Feature Index
A guide to JMP^{®} software features
With JMP installed on your PC or Mac, you can run an example of an analysis by clicking on the Example Script (JSL) link. This link will download a script, which can be run in JMP software to demonstrate the feature. Navigate this index by the letter links below or use the search box.
Suggestions or feedback for this page? Contact jmpweb@jmp.com.
Want to try out these example scripts for yourself and don't own a JMP license?
Download a free 30day trial today.
 = Capability only available in JMP Pro.
 = Denotes new capability in JMP 12.
Term  Definition  Example of how to access in JMP 

3D biplot: Gabriel 
A multivariate plot in principal components space, which shows both points and rays showing variables directions. 
Analyze > Multivariate Methods > Multivariate > Biplot 
3D scatterplot 
A threedimensional spinnable view of your data. 
Graph > Scatterplot 3D 
A
Term  Definition  Example of how to access in JMP 

ABCD design 
Screening design for mixtures. 
DOE > Mixture Design > Choose Mixture Design Type > ABCD Design 
accelerated failuretime models 
Fits a regression model to the parameters of a life distribution, such as Weibull. 
Analyze > Quality and Process > Fit Life by X > Fit All Distributions 
accelerated life test (ALT) design 
Used to design high stress tests; used when the time required to test the product until it fails is prohibitive. 
DOE > Accelerated Life Test Design 
adaptive elastic net 
A generalized regression estimation method that applies both an L1 (absolute value) and an L2 (squared) penalty to the likelihood when estimating parameters. 
Analyze > Fit Model > Personality > Generalized Regression > Estimation Method 
adaptive lasso 
A generalized regression estimation method that applies an L1 (absolute value) penalty to the likelihood when estimating parameters. 
Analyze > Fit Model > Personality > Generalized Regression > Estimation Method > Lasso 
addedvariable plot (leverage plot) 
A plot such that the distance from a point to the sloped line is the residual, and the distance to the horizontal line is what the residual would be under the hypothesis. 
Analyze > Fit Model > Personality:Standard Least Square > Leverage Plot 
adjusted inertia 
Benzecri (1979) stated that the relative inertia was a poor estimate of the quality of fit of the Multiple Correspondence Analysis solution and proposed an adjusted inertia. Greenacre (1984) argued the Benzecri adjustment overestimates the quality of fit and provided an alternate adjustment. 
Analyze > Consumer Research > Multiple Correspondence Analysis 
adjusted means 
The predicted value at each level of the indicated term, with other terms being set to neutral values. The Fit Model platform produces these automatically for nominal terms. 
General 
adjusted R^{2} 
A measure of degree of fit that has been adjusted to reflect the number of parameters in the model. Unlike the unadjusted R^{2}, the adjusted R^{2} does not always increase as more terms are added to the model. 
Analyze > Fit Model > Personality:Standard Least Square > Summary of Fit > RSquare Adj 
AIC, AICc, Akaike's 'A' Information Criterion 
A measure of the goodness of fit of an estimated statistical model that can be used to compare two or more models. The model with the lowest AIC value is the best. The AICc is a modification of the AIC adjusted for small samples. 
General 
alias matrix 
Shows the aliasing between the model terms and the terms that you specify in the Alias Terms panel. It enables you to see the confounding patterns. 
DOE > Custom Design > Make Design > Design Evaluation > Alias Matrix 
alias optimal design 
This design minimizes the sum of squares of the entries in the alias matrix subject to constraining the Defficiency of the design to be above some lower bound. 
DOE > Custom Design > Optimality Criterion > Make Alias Optimal Design 
all possible models 
Runs all possible models using combinations of the regression parameters specified. 
Analyze > Fit Model > Personality:Stepwise > All Possible Models 
ALT design 
Used to design high stress tests; used when the time required to test the product until it fails is prohibitive. 
DOE > Accelerated Life Test Design 
analysis of covariance  ANCOVA  same slopes 
When main effects model needs adjusting for a covariate. Use Fit Model, specifying the main effect, the covariate. 
Analyze > FIt Model > Add Effects 
analysis of covariance  ANCOVA  separate slopes 
The slope on a covariate is different in different groups. Use Fit Model, specifying the main effect, the covariate, and a crossed effect for main effect by covariate. 
Analyze > Fit Model > Personality:Standard Least Square 
analysis of mean ranges 
A chart which shows how ranges vary across groups. 
Analyze > Quality and Process > Measurement Systems Analysis > Range Chart 
analysis of means 
Compares group means to the overall mean. This method assumes that your data are approximately normally distributed. 
Analyze > Fit Y by X > Oneway > Analysis of Means > ANOM 
analysis of means  ANOM 
Compares group means to the overall mean. This method assumes that your data are approximately normally distributed. 
Analyze > Fit Y by X > Oneway > Analysis of Means Method > ANOM 
analysis of means  ANOM for proportions 
Compares response proportions for the X levels to the overall response proportion. Only appears if the response has exactly two levels. 
Analyze > Fit Y by X > Contingency > Analysis of Means for Proportions 
analysis of means  ANOM for transformed ranks (ANOMTR) 
This is the nonparametric version of the ANOM analysis. Use this method if your data are clearly nonnormal and cannot be transformed to normality. Compares each group mean transformed rank to the overall mean transformed rank. 
Analyze > Fit Y by X > Oneway > Analysis of Means Method > ANOM with Transformed Rank 
analysis of means  ANOM for variances (ANOMV) 
Tests for variance heterogeneity by comparing group standard deviations to the root mean square error. 
Analyze > Fit Y by X > Oneway > Analysis of Means Method > ANOM for Variances 
analysis of means  ANOM for variances with Levene (ADM) 
This is the nonparametric version of the ANOM for Variances analysis. Use this method if you suspect your data are nonnormal and cannot be transformed to normality. Compares the group means of the absolute deviation from the median (ADM) to the overall mean ADM. 
Analyze > Fit Y by X > Oneway > Analysis of Means Method > ANOM for Variances with Levene 
analysis of variance  ANOVA general (two or more factors) 
For almost any linear model, use Fit Model and fill in the dialog. 
Analyze > Fit Model > Personality:Standard Least Square > Analysis of Variance 
analysis of variance  ANOVA oneway 
Fitting means across a grouping variable, and testing if they are significantly different. 
Analyze > Fit Y by X > Oneway > t Test 
anticipated response 
Response values at the specified design settings calculated using the Anticipated Coefficients specified in the Power Analysis report. Found in the DOE Design report. 
DOE > Custom Design > Make Design 
AR (auto regressive) model 
Group of linear prediction formulas that attempt to predict an output of a system based on the previous outputs. 
Analyze > Modeling > Time Series > ARIMA Model Group > Auto Regressive 
area plot 
Shows a response summarized by categories. 
Graph > Graph Builder > Area 
area under the curve (AUC) 
AUC is equal to the probability that a classifier will rank a randomly chosen positive instance higher than a randomly chosen negative one. 
Analyze > Fit Y by X > Logistic > ROC Curve 
ARIMA 
Fitting ARIMA (autoregressive integrated moving average) models for times series analysis. 
Analyze > Modeling > Time Series > ARIMA 
ARL 
The average run length at a given quality level is the average number of samples (subgroups) taken before an action signal is given. 
Analyze > Quality and Process > Control Chart > CUSUM > Show ARL (Also available through Control Chart Builder) 
Arrhenius transformation 
A transformation of temperature used in accelerated life testing. This transformation is supported for singlecolumn continuous effects. 
Analyze > Reliability and Survival > Fit Life by X > Relationship:Arrhenius Transformation 
association 
A relationship between two variables. Association can be tested for or visualized in many platforms in JMP. 
General 
Attribute Gauge R&R 
Analyzing the agreement between raters of an attribute, such as accept/reject. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart > Chart Type:Attribute > Gauge Studies > Gauge RR 
AUC 
AUC is equal to the probability that a classifier will rank a randomly chosen positive instance higher than a randomly chosen negative one. 
Analyze > Fit Y by X > Logistic > ROC Curve 
augment design 
To modify an existing design data table. 
DOE > Augment Design 
autocorrelation 
The correlation between a value and the value next to it in a time series. 
Analyze > Time Series > Autocorrelation 
autoregression 
See ARIMA. Note, autoregression models with autocorrelated and heteroscedastic errors are available in the SAS/ETS^{®} addin using SAS PROC Autoreg. 
Analyze > Modeling > Time Series 
average chart 
XBar Control Chart. Shows the process mean and its variability. 
Analyze > Quality and Process > Measurement Systems Analysis > Average Chart 
average run length 
The average run length at a given quality level is the average number of samples (subgroups) taken before an action signal is given. 
Analyze > Quality and Process > Control Chart > CUSUM > Show ARL (Also available through Control Chart Builder) 
B
Term  Definition  Example of how to access in JMP 

backwards selection 
Describes how an effect leaves a model; backward removes the regressor that affects the fit the least, given that term is not significant at the specified level. 
Analyze > Fit Model > Personality:Stepwise > Stepwise Regressional Control > Direction:Backwards 
bar graph (bar chart) 
Bar heights across groups represent statistics. 
Graph > Chart > Options: Bar Chart 
bar plot 
Shows a response summarized by categories. 
Graph > Graph Builder > Bar 
Bartlett's test 
Testing that the variances are equal in a oneway layout and providing a weighted (Welch) Anova in case they aren't. 
Analyze > Fit Y by X > Oneway > Unequal Variances 
Bayes Plot (BoxMeyer) 
For screening designs, this helps pick out the active effects, using a Bayesian approach that views the distribution of inactive effects as being contaminated by a distribution of active effects with ktimes larger variance. 
Analyze > Fit Model > Effect Screening > Bayes Plot 
Bayesian aliasoptimal 
This is a modification of the Alias Optimal criterion that seeks to minimize the aliasing between model effects and alias effects. At the same time, this design has the ability to detect and estimate some higherorder terms. 
DOE > Custom Design > Optimality Criterion > Make Alias Optimal Design 
Bayesian Doptimal 
This type of design allows the precise estimation of all of the Necessary terms while providing omnibus detectability and some estimability of the If Possible terms. 
DOE > Custom Design > Optimality Criterion > Make DOptimal Design 
Bayesian designs 
Modifies the given optimality criterion so that the design has the ability to detect and estimate some higher order terms. 
DOE > Custom Design > Make Design 
Bayesian Ioptimal 
This type of design minimizes the average prediction variance over the design region and at the same time has the ability to detect and estimate some higherorder terms. 
DOE > Custom Design > Optimality Criterion > Make IOptimal Design 
Bayesian Information Criterion (BIC) 
This is a measure (that is in part, based on the likelihood function) of model fit that is helpful when comparing different models. 
Analyze > Modeling > Time Series > Stopping Rule:Minimum BIC 
Bayesian split plot 
This type of design contains hardtochange factors which only change between one whole plot and the next. In addition, this design has the ability to detect and estimate some higher order terms. 
DOE > Custom Design > Design Generation > Number of Whole Plots 
BCI 
A measure of the impact of a component to system reliability over time. A large BCI indicates that the system is sensitive to the component. 
Analyze > Reliability and Survival > Reliability Block Diagran > Show BCI 
Bernoulli (trials) 
Random numbers from a binomial distribution with parameters that you enter as function arguments. 
Cols > Formula > Random Functions 
best subsets regression 
Runs all possible models using combinations of the regression parameters specified. 
Analyze > Fit Model > Personality:Stepwise > All Possible Models 
bias comparison 
Analysis of Means test to determine whether the group differences are real differences or whether they are due to measurement error. 
Analyze > Quality and Process > Measurement Systems Analysis > Bias Comparison 
bias report 
Shows a graph and summary table for each X variable. The average bias, or differences between the observed values and the standard values, is given for each level of the X variable. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart > Gauge Studies > Bias Report 
BIC 
This is a measure (that is in part, based on the likelihood function) of model fit that is helpful when comparing different models. 
Analyze > Fit Model > Personality:Stepwise > Stepwise Regression Control > Stopping Rule: Minimum BIC 
biplot (principal components) 
Biplots are a type of exploratory graph used in statistics, a generalization of the simple twovariable scatterplot. A biplot allows information on both samples and variables of a data matrix to be displayed graphically. Samples are displayed as points while variables are displayed either as vectors, linear axes or nonlinear trajectories. 
Analyze > Multivariate Methods > principal Components > Biplot 
Birnbaum's Component Importance (BCI) 
A measure of the impact of a component to system reliability over time. A large BCI indicates that the system is sensitive to the component. 
Analyze > Reliability and Survival > Reliability Block Diagran > Show BCI 
BlandAltman plot 
A plot for matched pairs analysis which shows the relationship between the differences versus the means of the paired observations. 
Analyze > Matched Pairs > Plot Dif by Mean 
BLUP (best linear unbiased prediction) 
Output for random effects in mixed model analysis. 
Analyze > Fit Model Personality: Standard Least Squares  Method: REML (Recommended) > Random Effect Predictions: BLUP 
Bonferroni adjustment for multiple comparisons 
Access in Fit Y by X (set alpha level). Apply the Bonferroni adjustment for multiple comparisons by dividing alpha by the number of comparisons being made. 
General 
boosted neural network 
The process of building a large additive neural network by fitting a sequence of smaller models. 
Analyze > Modeling > Neural > Boosting 
boosting (boosted tree) 
The process of building a large, additive decision tree by fitting a sequence of smaller trees. 
Analyze > Modeling > Partition > Method:Boosted Tree 
boosting (neural network) 
The process of building a large additive neural network by fitting a sequence of smaller models. 
Analyze > Modeling > Neural > Boosting 
bootstrap forest (random forest technique) 
Creates many trees and computes the final predicted value by averaging the predicted values. 
Analyze > Modeling > Partition > Method:Bootstrap Forest 
bootstrapping 
A resampling method that allows for estimation of the standard error of a statistic. Available in most analytical platforms in JMP Pro. 
Right click any statistic in JMP Pro report and click bootstrap. 
Bowker's test of symmetry 
A test of the symmetry of kbyk tables that assumes corresponding nondiagonal elements are equal; for 2x2 tables, Bowker's Test is equivalent to McNemar's Test. 
Analyze > Fit Y by X > Contingency > Agreement Statistics 
box plot (Graph Builder) 
Shows a compact view of a variable's distribution, with quartiles and outliers. 
Graph > Graph Builder > Box Plot 
box plot (One level) 
A graph to look at a distribution, which draws a box around the middle half of the data, and lines extending until the outlying data. The Distribution platform does this across one or more variables. 
Analyze > Distribution > Outlier Box Plot 
box plot  groups 
Sidebyside box plots comparing the distribution across groups. The Oneway platform does this across groups. 
Analyze > Fit Y by X > Oneway > Display Option > Box Plots 
box plot  multiple groupings 
Sidebyside box plots across several grouping variables. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart > Multiple Groupings > Show Box Plots 
boxandwhisker plot 
A box plot enclosing the inner quartiles of points with lines to the farthest point within 1.5 interquartile ranges from the quartiles. 
Analyze > Distribution > Outlier Box Plot 
BoxBehnken design 
A response surface experimental design with points midway between vertices. 
DOE > Response Surface Design > Choose a Design > BoxBehnken 
BoxCox power transformation 
A power transformation, usually on the response, proportional to y^{lambda1}. 
Analyze > Fit Model > Personality:Standard Least Square > Factor Profiling > Bo Co Y Transformation 
BoxJenkins Methods 
Fitting ARIMA (autoregessive integrated moving average) models for time series analysis. 
Analyze > Modeling > Time Series > ARIMA 
BoxMeyer Bayes Plot 
For screening designs, this helps pick out the active effects, using a Bayesian approach that views the distribution of inactive effects as being contaminated by a distribution of active effects with ktimes larger variance. 
Analyze > Fit Model > Effect Screening > Bayes Plot 
BoxWilson designs 
An experimental design for response surface analysis involving three sets of points: Vertex points, Center points, and Axial points. 
DOE > Custom Design 
BrownForsythe test/test variances equal across groups/Oneway(Test Equal Variances) 
Testing that the variances are equal in a oneway layout and providing a weighted (Welch) Anova in case they are not. 
Analyze > Fit Y by X > Oneway > Unequal Variances 
bubble chart (bubble plot) 
A scatterplot that draws its points as circles (bubbles). The bubbles can be sized or colored depending on other columns. A bubble plot can also be animated through time, resulting in a fifth displayed variable. 
Graph > Bubble Plot 
Burt table 
A Burt table is a partitioned symmetric table of all pairs of categorical variables. 
Analyze > Consumer Research > Multiple Correspondence Analysis > (Red Triangle) > Cross Table 
C
Term  Definition  Example of how to access in JMP 

C chart 
A plot showing the numbers of nonconformities in subgroup samples. 
Analyze > Quality and Process > Control Chart > C 
C(p) 
Mallow’s C(p) is a measure to use when comparing models. Usually C(p) is plotted against p, the number of regressors. 
Analyze > Fit Model > Personality: Stepwise 
calibration 
To obtain a prediction and (fiducial) confidence interval for an X, given the Y value and other X values. 
Analyze > Fit Model > Personality:Nominal Logistic > Inverse Partition 
canonical correlation (Discriminant Analysis) 
Canonical Correlation in discriminant analysis is the correlation between the classification categories and the canonical scores calculated on the discriminant factors. 
Analyze > Multivariate > Discriminant, hotspot > canonical options > show canonical details 
canonical correlation (MANOVA) 
Canonical correlations between the multiple response canonical scores and the model factors can be calculated. 
Analyze > Fit Model > Personality:MANOVA For a fitted model (based on the MANOVA correlation structure you want to use), choose the hotspot for the whole model or model effect of interest and choose test details. 
canonical plot (discriminant analysis) 
Shows the points and multivariate means in the two dimensions that best separate the groups in linear discriminant analysis. Each row in the data is a point. The multivariate means are labeled circles. The directions of the variables in the canonical space are shown by labeled rays emanating from the grand mean. 
Analyze > Multivariate Models > Discriminant > Canonical Plot 
capability analysis 
Computes capability analyses for each column and creates a goal plot with one point for each column. Capability analysis uses the standardized mean and standard deviation as the X and Y coordinates. 
Analyze > Quality and Process > Capability 
capability box plots 
Shows or hides together in one graph, one box plot for each column, where the box plots are each standardized to their respective specification limits. 
Analyze > Quality and Process > Capability > Capability Box Plots 
caption box (Graph Builder) 
Shows a summary statistic value for the data. 
Graph > Graph Builder > Caption Box 
categorical analysis 
Analysis for data with discrete (categorical) levels. The strength of the Categorical platform is that it can handle responses in a wide variety of formats without needing to reshape the data. 
Analyze > Modeling > Categorical 
Cauchy regression 
A robust regression technique. 
Analyze > Fit Y By X > (Red Triangle) > Robust > Fit Cauchy 
cause and effect diagrams 
A hierarchical diagram to lay out root causes. Also called Ishikawa or fishbone diagrams. 
Analyze > Quality and Process > Diagram 
CDF plot 
Plot of the empirical cumulative distribution function. Available in the Distribution platform as well as in Fit Y by X (Oneway), when comparing distributions across groups. 
Analyze > Distribution > CDF Plot 
center points 
Additional runs to be placed at the center of each continuous factor’s range. 
DOE > Custom Design > Design Generation > Number of Center Points 
central composite design 
An experimental design for response surface analysis involving three sets of points: Vertex points, Center points, and Axial points. 
DOE > Response Surface Design > Choose a Design > Central Composite Design 
chisquare (chi squared) – test for association 
Used to determine whether there is a significant association between the two categorical variables from a single population. 
Analyze > Fit Y by X: Contingency 
chisquare (chi squared) – test for independence 
Used to determine whether there is a significant association between the two categorical variables from a single population. 
Analyze > Fit Y by X: Contingency 
ChiSquare  for a general categorical response model 
The likelihood ratio ChiSquare compares the fit of a model with how good the fit would be without certain effects in the model. 
Analyze > Fit Model > Personality:Nominal Logistic > Whole Model Test > ChiSquare 
ChiSquare  for twoway table of frequencies 
To test for independence (or marginal homogeneity) for two categorical variables using a table of counts. There are two versions: the likelihood ratio G^{2}, and the Pearson X^{2}. 
Analyze > Fit Y by X > Contingency > Tests > ChiSquare 
choice analysis 
Fits choice models for market research. Conjoint Experiments. 
Analyze > Modeling > Choice 
choice design 
Creates experiments with factors that are product attributes. The purpose of a choice experiment is to define a product that people want to buy. 
DOE > Choice Design 
classification trees 
A decision tree for a categorical response variable. 
Analyze > Modeling > Partition > Method 
cluster analysis 
Clustering clumps together points that are close to each other (points that have similar values). JMP provides two types of clustering, hierarchical and kmeans. 
Analyze > Multivariate Methods > Cluter > Hierarchical 
Cochran Armitage trend test 
Tests for trends in binomial proportions across levels of a single variable. 
Analyze > Fit Y by X > Contingency > Cochran Armitage Trend Test 
CochranMantelHaenszel test 
Computes ChiSquare statistics for stratifications of a twoway contingency table by a third variable. 
Analyze > Fit Y by X > Contingency > Cochran Mantel Haenszel 
coefficient of variation 
The standard deviation estimate divided by the mean, expressed as a percentage. 
Analyze > Distribution > Customize Summary Statistics > CV 
coefficient of variation (C_{v}) 
Measure of dispersion, which is the standard deviation divided by the mean multiplied by 100. 
General 
collinearity 
Collinearity occurs when one or more model effects in a regression model have strong correlations. This results in inflated estimates of the model parameter estimates, making it more difficult to show statistical significance for these parameter estimates. Collinearity can be seen in many ways, including: a) Collinearity shows as a shrunken X scale on the factors’ leverage plots. b) Variance Inflation Factor: See statistics index entry for Variance Inflation Factors. 
Analyze > Fit Model&;Personality:Standard Least Square > Leverage Plots 
color by variable levels (Graph Builder) 
Drop variables here to color the graph. If you are using a map, the map shapes are colored. If you are using a contour plot, colored contours appear. If your graph contains points, they are colored. 
Graph > Graph Builder > Color 
compare data tables 
Compare two data tables. Compare data, tables’ metadata, as well as columns’ metadata. 
Tables > Compare Data Tables 
comparison circles 
A graphical method of comparing means, lining up circles vertically with center at means and diameter as the confidence interval, the angle of intersection indicates significance. Used in conjunction with a means comparison option. 
Analyze > Fit Y by X > Oneway > Display Options > Comparison Circles 
concatenate data tables 
Concatenate to append tables end to end. 
Tables > Concatenate 
conditional logistic regression 
Regression models for matched or grouped data with a binary response variable. A casecontrol study could be analyzed using conditional logistic regression using the Choice platform. 
Analyze > Modeling > Choice 
confidence intervals 
JMP can produce confidence intervals around estimates in most reports. 
General 
confounding 
Confounding is when terms in a model are linearly related. The DOE and Fit Model facilities both deal with this as needed. 
General 
confusion matrix 
A matrix that tabulates the predictive ability of a model for a categorical response. Available for logistic regression in Fit Model, Partition, Neural and in JMP Pro, Model Comparison. 
General 
constellation plot 
Plot which shows cluster joins as points and observations as endpoints. 
Analyze > Multivariate Methods > Cluster > Options = Hierarchical > OK > Constellation Plot 
constraints for DOE factors 
Defines the allowable region for factors in a designed experiment. 
DOE > Custom Design > Continue > Define Factor Constraints 
contour plot 
A twodimensional plot that displays level curves of a function or a bivariate density. 
Graph > Contour Plot 
contour plot (Graph Builder) 
A twodimensional plot that displays level curves of a function or a bivariate density. 
Graph > Graph Builder > Contour 
contour profiler 
A twodimensional plot that shows level curves of one or more functions, that may also depend on other factors. 
Graph > Contour Profiler 
contrasts 
Contrasts are tests that means or least squares means are different. The linear function across the means should sum to zero, and the positive and negative elements should sum to +1 and 1 respectively. In JMP, the Fit Model platform provides features for estimating and testing contrasts. 
Analyze > Fit Model > Personality:Standard Least Squares > LSMeans Contrast 
control chart with stages (phase) 
Phases generate for each level of the specified phase variable a new σ, set of limits, zones, and resulting tests. 
Analyze > Quality and Process > Control Chart > Choose option > Phase (Also available through Control Chart Builder) 
control charts (C) 
A plot showing the numbers of nonconformities in subgroup samples. 
Analyze > Quality and Process > Control Chart > C 
control charts (CUSUM) 
A plot showing cumulative sums of the deviations of the subgroup means from a target value. 
Analyze > Quality and Process > Control Chart > CUSUM 
control charts (EWMA) 
A plot showing an exponentially weighted moving average chart. 
Analyze > Quality and Process > Control Chart > EWMA 
control charts (G) 
A control chart for rare events that plots a count of units or occurrences between rare events. 
Analyze > Quality and Process > Control Chart Builder > Rare Event Charts > gchart 
control charts (IMR) 
An individual measurement chart and moving range chart. 
Analyze > Quality and Process > Control Chart > IR > Individual Measurement/Moving Range (Also available through Control Chart Builder) 
control charts (IMRR) 
Uses both betweensubgroup and withinsubgroup variations to generate an individuals, moving range, and R chart. 
Analyze > Quality and Process > Control Chart > XBar > R (Also available through Control Chart Builder) 
control charts (IMRS) 
Uses both betweensubgroup and withinsubgroup variations to generate an individuals, moving range, and S chart. 
Analyze > Quality and Process > Control Chart > XBar > S (Also available through Control Chart Builder) 
control charts (individual measurement) 
An individual measurement chart and moving range chart. 
Analyze > Quality and Process > Control Chart > IR > Individual Measurement (Also available through Control Chart Builder) 
control charts (IR) 
An individual measurement chart and moving range chart. 
Analyze > Quality and Process > Control Chart > IR (Also available through Control Chart Builder) 
control charts (Levey Jennings) 
Chart that shows a process mean with control limits based on a longterm σ. The control limits are placed at 3*σ distance from the center line. 
Analyze > Quality and Process > Control Chart > Levey Jennings (Also available through Control Chart Builder) 
control charts (MR) 
Control chart that displays the moving ranges of two or more successive measurements. 
Analyze > Quality and Process > Control Chart > Right Click:Limits > Moving Range 
control charts (multivariate) 
Creates a chart to view summaries for monitoring problems where several related variables are of interest. 
Analyze > Quality and Process > Multivariate Control Chart 
control charts (NP) 
A plot showing the numbers of nonconforming items in subgroup samples. 
Analyze > Quality and Process > Control Chart > NP 
control charts (P) 
A plot showing the proportions of nonconforming items in subgroup samples. 
Analyze > Quality and Process > Control Chart > P 
control charts (presummarized) 
A plot showing presummarized sample means. 
Analyze > Quality and Process > Control Chart > Presummarize (Also available through Control Chart Builder) 
control charts (rare events) 
Control charts used to determine whether rare events are occurring more frequently. Gcharts plot the number of events between rare events, and Tcharts plot the time between rare events. 
Analyze > Quality and Process > Control Chart Builder > Rare Event Charts 
control charts (run) 
A plot showing a run of the samples. 
Analyze > Quality and Process > Control Chart > Run Chart (Also available through Control Chart Builder) 
control charts (S) 
Control chart that displays the subgroup standard deviations. 
Analyze > Quality and Process > Control Chart > XBar > S (Also available through Control Chart Builder) 
control charts (T) 
A control chart for rare events to determine whether rare events are occurring more frequently than expected by graphing time between events. 
Analyze > Quality and Process > Control Chart Builder > Rare Event Charts > tchart 
control charts (U) 
A plot showing the numbers of nonconformities per inspection unit in subgroup samples. 
Analyze > Quality and Process > Control Chart > U 
control charts (UWMA) 
A plot showing a uniformly weighted moving average chart. 
Analyze > Quality and Process > Control Chart > UWMA 
control charts (XBar/R) 
A plot showing a sequence of samples to detect an outofcontrol situation in statistical process control. 
Analyze > Quality and Process > Control Chart > XBar > XBar/R (Also available through Control Chart Builder) 
control charts (XBar/S) 
A plot showing a sequence of samples to detect an outofcontrol situation in statistical process control. 
Analyze > Quality and Process > Control Chart > XBar > XBar/S (Also available through Control Chart Builder) 
Cook’s D 
A regression diagnostic used to identify outliers and influential observations. 
Analyze > Fit Model > Personality:Standard Least Squares > Save Columns > Cook’s D Influence 
corrected sum of squares 
Determines the total amount of variation that occurs with the individual observations of Y about the mean estimate of Y. 
General 
correlation  for many variables 
Correlation measures how linearly related two variables are, ranging from 1 to 1. For many variables, the matrix of correlations is the gateway to many multivariate techniques. 
Analyze > Multivariate Merthods > Multivariate 
correlation  two variables 
Correlation measures how linearly related two variables are, ranging from 1 to 1. 
Analyze > Fit Y by X > Bivariate 
correspondence analysis 
A plot using 2way frequency counts, showing relationship between levels of nominal/ordinal variables. 
Analyze > Fit Y by X > Contingency > Correspondence Analysis 
Cotter designs/DOE(Screening(Cotter)) 
A varyonefactoratatime design in which sums of even and odd effects are estimable. The DOE screening facility can construct these if they have not been disabled by preference setting. 
DOE > Screening Design > Deselect Suppress Cotter Design 
counts; see frequency counts 
Distribution or Contingency. 
Analyze > Distribution 
covariance 
Measure of how two random variables change together. 
Analyze > Multivariate Method > Multivariate > Covariance Matrix 
covering array design 
A design in which for any t columns all possible combinations of factor levels occur at least once. 
DOE > Covering Array 
Cox Proportional Hazards Model 
A semiparametric model to fit survival times. 
Analyze > Reliability and Survival > Fit Proportional Hazard 
CP 
Estimates what the process is capable of producing if the process mean were to be centered between the specification limits. 
Analyze > Distribution > Capability Analysis > Long Term Sigma > CP 
CPK 
Estimates what the process is capable of producing, considering that the process mean might not be centered between the specification limits. 
Analyze > Distribution > Capability Analysis > Long Term Sigma > CPK 
CPL 
Estimates process capability for specifications that consist of a lower limit only. 
Analyze > Distribution > Capability Analysis > Long Term Sigma > CPL 
CPM 
Estimates process capability around a target. 
Analyze > Distribution > Capability Analysis > CPM 
CPU 
Estimates process capability for specifications that consist of an upper limit only. 
Analyze > Distribution > Capability Analysis > CPU 
Cramervon Mises 
See Reliability Growth documentation. 
Reliability and Survival > Reliability Growth 
Cronbach’s a 
A measure of item reliability on how consistently a set of variables measures overall response. 
Analyze > Mulivariate Methods > Multivariate > Item Reliability > Cronbach’s a 
cross correlation 
The linear relationship between two time series variables. 
Analyze > Modeling > Time Series > Cross Correlation 
cross validation 
The process of using parts of a data set to estimate model parameters, as well as to assess the predictive ability of the fitted model. Many modeling platforms in JMP Pro support a validation column role. 
General 
cross validation 
Many modeling platforms in JMP support kfold or leaveoneout cross validation. See kfold cross validation. 
General 
crosstabulation 
Frequency counts arranged in a table. JMP does this for 2 variables only. 
Tables > Tabulate > Add Columns by Categories 
Crow AMSAA model 
Model that allows for tracking reliability of a process during development testing to determine whether changes to the process are improving process reliability over time. 
Analyze > Reliability and Survival > Reliability Growth > Fit Model > Crow AMSAA 
cube plot 
A cube with predicted values shown on the vertices. 
Analyze > Fit Model > Personality:Standard Least Square > Factor Profiling > Cube Plots 
cumulative density function plot 
Plot of the empirical cumulative distribution function. 
Analyze > Distribution > CDF Plot 
cumulative gains chart 
Shows overlaid cumulative gains curves, which measure the effectiveness of a predictive model, for each level of the response. 
Analyze > Modeling > Model Comparison > Cum Gains Curve 
curvature 
JMP can deal with curvature in models in many platforms. See documentation for Custom Designer, Fit Model, and Fit Y by X. 
General 
custom tests 
Tests for customized model hypotheses. 
Analyze > Fit Model > Personality:Standard Least Square > Estimates > Custom Test 
CUSUM chart 
A plot showing cumulative sums of the deviations of the subgroup means from a target value. 
Analyze > Quality and Process > Control Chart > CUSUM 
CV 
The standard deviation estimate divided by the mean, expressed as a percentage. 
Analyze > Distribution > Summary Statistics > Customize Summary Statistics > CV 
D
Term  Definition  Example of how to access in JMP 

DOptimal design 
An experimental design that minimizes the determinant of the variance matrix of the regression parameters. The DOE platform does this under Custom Design. 
DOE > Custom Design > Optimality Criterion > Make DOptimal Design 
Daniel plot (halfnormal plot) 
Plots the absolute values of the estimates against the normal quantiles for the absolute value normal distribution. 
Analyze > Fit Model > Personality:Standard Least Square > Effect Screening > Normal Plot > Half Normal Plot 
data mining 
General term to refer to extracting patterns from large sets of data. 
General 
databases (connection to) 
You can import data from a database if you have an ODBC (Open Database Connectivity) driver for the database. 
File > Database > Open Table 
decision Trees 
Recursively partition the data to predict a response. Classification and regression trees. 
Analyze > Modeling > Partition > Method:Decision Tree 
definitive screening design 
Threelevel designs where: main effects are independent of twofactor interactions; quadratic effects are estimable; secondorder effects are only partially aliased. 
DOE > Definitive Screening Design 
degradation analysis 
Analyzes product degradation (or deterioration) over time and anticipates product quality in the future. 
Analyze > Reliability and Survival > Degradation 
demonstration test plans 
Let you design a test to compare the reliability of a new product to a standard. 
DOE > Sample Size and Power > Reliability Demonstration 
dendrogram 
A tree diagram used to illustrate the arrangement of clusters produced by hierarchical clustering. 
Analyze > Multivariate Methods > Cluster > Options:Hierarchical 
density ellipse (Graph Builder) 
Shows a bivariate normal density ellipse. 
Graph > Graph Builder > Ellipse 
density ellipses 
Draws an ellipse that contains a specified mass of points; the number of points is determined by a specified probability. 
Analyze > Fit Y by X > Bivariate > Density Ellipses 
design of experiments 
The issues for planning experimental runs efficiently so that the responses can be fit to answer the questions of interest. 
DOE 
desirability profiling (Optimization) 
A technique of setting up desirability functions, and searching for factor values that optimize a composite desirability of a number of responses. 
Many modeling platforms such as Neural, Fit Model and Partition can add desirability functions to their profilers for optimization. 
diagram 
Cause and effect diagrams. Also called Ishikawa or fishbone diagrams. A hierarchical diagram to lay out root causes. 
Analyze > Quality and Process > Diagram 
DickeyFuller tests 
Diagnostic tests for stationarity performed in the Time Series platform. 
Analyze > Modeling > Time Series > ADF 
disallowed combinations 
Disallows any combination of levels of categorical factors in Design of Experiments. 
DOE > Custom Design > Disallowed Combinations 
discrete choice analysis 
Fits choice models for market research. Conjoint Experiments. 
Analyse > Modeling > Choice 
discrete choice design 
Creates experiments with factors that are product attributes. The purpose of a choice experiment is to define a product that people want to buy. 
DOE > Choice Design 
discriminant analysis 
Classifying points to groups according to which group means that the column values are closest to. 
Analyze > Multivariate Methods > Discriminant 
discrimination ratio 
Compares the total variance of the measurement with the variance of the measurement error. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart > Gauge Studies > Discrimination Ratio 
distance matrix 
Matrix containing the distances between the observations. 
Analyze > Multivariate Methods > Cluster > Options:Hierarchical > Save Distance Matrix 
distribution 
Use the Distribution platform to describe the shape, centering and spread of variables with graphical displays and summary statistics. Explore and fit the underlying distribution using Distribution or Life Distribution. 
Analyze > Distribution or Analyze > Reliability and Survival > Life Distribution 
Duane plot 
Plot in reliability growth analysis that fits a line to the points (log10(X), log10(Y)), where Y is the estimated cumulative mean time between failures (MTBF) and X is the time to event variable. 
Analyze > Reliability and Survival > Reliability Growth > OK 
Dunn All Pairs for Joint Ranks 
Performs a comparison of each pair, similar to the SteelDwass All Pairs option. The Dunn method is different in that it computes ranks on all of the data, not just the pair being compared. The Dunn method calculates pvalues using the Bonferroni method. 
Analyze > Fit Y by X > Oneway > Nonparametric > Nonparametric Multiple Comparisons > Dunn All Pairs for Joint Ranks 
Dunn with Control for Joint Ranks 
Compares each level to a control level, similar to the Steel With Control option. The Dunn method is different in that it computes ranks on all of the data, not just the pair being compared. The Dunn method calculates pvalues using the Bonferroni method. 
Analyze > Fit Y by X > Oneway > Nonparametric > Nonparametric Multiple Comparisons > Dunn with Control for Joint Ranks 
Dunnett’s test for treatments vs. control 
Test that group means are different from mean of a control in a oneway ANOVA. The test controls the significance level for multiple comparisons. See ‘multiple comparisons’. 
Analyze > Fit Y by X > Oneway > Compare Means > With Control, Dunnett’s 
DurbinWatson test 
In regression, a test that the residuals are autocorrelated. In JMP, an exact significant level is available. 
Analyze > Fit Model > Personality:Standard Least Squares > Row Diagnostics > Durbin Watson Test 
E
Term  Definition  Example of how to access in JMP 

effective resolution 
Determines the proper number of digits to record for a measurement. 
Analyze > Quality and Process > Measurement Systems Analysis > Effective Resolution 
effectiveness report 
The ratio of the number of correct decisions to the total number of opportunities for a decision. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart > Chart Type:Attribute > Effectiveness Report 
elastic net 
A generalized regression estimation method that applies both an L1 (absolute value) and an L2 (squared) penalty to the likelihood when estimating parameters. 
Analyze > Fit Model > Personality > Generalized Regression 
ellipses  bivariate density 
The contours of the bivariate normal distribution are ellipses. 
Analyze > Fit Y by X > Bivariate > Density Ellipse 
EMP Gauge R&R 
A Gauge R&R analysis based on the EMP (Evaluating the Measurement Process) method developed by Donald Wheeler. 
Analyze > Quality and Process > Measurement Systems Analysis > EMP Gauge RR Results 
EMP study 
An approach to evaluating a measurement process based on Donald Wheeler’s book “Evaluating the Measurement Process.” Example Script (JSL) 
Analyze > Quality and Process > Measurement Systems Analysis > EMP Results 
empirical CDF plot 
Plot of the empirical cumulative distribution function. 
Analyze > Distribution > CDF Plot 
empirical cumulative distribution function plot 
Plot of the empirical cumulative distribution function. 
Analyze > Distribution > CDF Plot 
entropy R^{2} 
Measure of fit that compares the loglikelihoods from the fitted model and the constant probability model. 
Analyze > Fit Model > Personality:Nominal Logistics > Whole Model Test > Entropy R^{2} 
equal variance test 
Testing that the variances are equal in a oneway layout and providing a weighted (Welch) Anova in case they are not. 
Analyze > Fit Y by X > Oneway > Unequal Variances 
equivalence test 
Equivalence tests assess whether there is a practical difference in means. 
Analyze > Fit Model > (Red Triangle) > Estimates > Multiple Comparisons > Turkey HSD All Pairwise Comparisons (Red Triangle) > Equivalence Tests 
error 
Pure error, within error, lack of fit error, residual error, mean square error. 
General 
error variance heterogeneity 
Tests for variance heterogeneity by comparing group standard deviations to the root mean square error. 
Analyze > Fit Y by X > Oneway > Unequal Variances 
escape rate 
The probability that a nonconforming part will be produced and not detected. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart > Chart Type:Attribute > Conformance Report > Calculate Escape Rate 
estimation efficiency 
A report that gives the fractional increase in confidence interval length and relative standard error of parameters for each model effect in a designed experiment. 
DOE General Design Evaluation 
evaluate design 
Evaluates existing designs (from JMP and other software products). 
DOE > Evaluate Design 
exact agreement statistic 
Performs an exact test for testing agreement between variables. This is an exact test for the Kappa statistic. This is available only when the two variables have the same levels. 
Analyze > Fit Y by X > Contingency > Exact Test > Exact Agreement Statistic 
exact Cochran Armitage trend test 
Performs the exact version of the Cochran Armitage Trend Test. This test is available only when one of the variables has two levels. 
Analyze > Fit Y by X > Contingency > Exact Test > Exact Cochran Armitage Trend Test 
exact KolmogorovSmirnov test 
Performs the exact version of the KolmogorovSmirnov test. This option is available only when the X factor has two levels. 
Analyze > Fit Y by X > Oneway > Nonparametric > Exact Test > Kolmogorov Smirnov Exact Test 
exact van der Waerden test 
Performs the exact version of the van der Waerden test. This option is available only when the X factor has two levels. 
Analyze > Fit Y by X > Oneway > Nonparametric > Exact Test > Van Der Waerden Exact Test 
exact Wilcoxon rank sum test 
Performs exact versions of the Wilcoxon test. This option is available only when the X factor has two levels. 
Analyze > Fit Y by X > Nonparametric&Exact Test > Wilcoxon Exact Test 
exact Wilcoxon test 
Performs exact versions of the Wilcoxon test. This option is available only when the X factor has two levels. 
Analyze > Fit Y by X > Oneway > Nonparametric > Exact Test > Wilcoxon Exact Test 
Excel (integration) 
Transfer data from an Excel worksheet into a JMP table or launch basic JMP analysis platforms. Visualize and explore Excel models in JMP using the Prediction Profiler (Windows Only). 
General 
experimental design 
The issues for planning experimental runs efficiently so that the responses can be fit to answer the questions of interest. 
DOE 
exponential distribution 
A family of continuous probability distributions describing the time between Poisson events. 
General 
exponential plot  Survival 
A plot of log(Survival) by time for an estimated survival curve. If the survival distribution is exponential, the curve tends to be a straight line. 
Analyze > Reliability and Survival > Survival > Exponential Plot 
exponential smoothing 
Fitting a moving average process for forecasting a time series. 
Analyze > Modeling > Time Series > Smoothing Model > Exponential Smoothing 
Exponentially weighted moving average (EWMA) chart 
A plot showing an exponentially weighted moving average chart. 
Analyze > Quality and Process > Control Chart > EWMA 
extreme vertices design 
For mixture experiments, this design is based on taking corners of the constrained factor space. 
DOE > Mixture Design > Choose Mixture Design Type > Extreme Vertices 
F
Term  Definition  Example of how to access in JMP 

F test 
A test, most commonly used in ANOVA and regression, in which the test statistic follows the F distribution under the null hypothesis. The test statistic, the F ratio, is the ratio of two scaled sums of squares. 
General 
factor analysis 
Finding directions among a large set of variables that seem to simplify the structure of the variables into a small number of ‘factors’. Usually the factor solution is rotated to be more interpretable. JMP implements three types of factor analysis: principal component analysis, noniterated principal factor analysis with SMC, and maximum likelihood factor analysis. 
Analyze > Multivariate Methods > Multivariate > Factor Analysis 
factor loading plot 
A scatterplot matrix where each cell is a plot of the loadings for each variable on a pair of factors. 
Analyze > Consumer Research > Factor Analysis > OK > Rotation Matrix 
factorial designs 
An experimental design in which all possible combinations of factor levels occur. 
DOE > Full Factorial Design 
factorial models 
A factorial model is one in which all possible interactions are specified. In the JMP Model dialog, there is a menu item to make factorial effect sets. 
Analyze > Fit Model > Personality:Nominal Logistic 
failure time 
Failure times can be modeled in the Survival and Fit Model platforms. 
Analyze > Reliability and Survival > Survival > Plot Failure instead of Survival 
false discovery rate (FDR) 
The expected proportion of Type I errors, or false discoveries, when conducting multiple hypothesis tests. Used to control error rate in multiple testing. 
Analyze > Modeling > Response Screening > Select X and Y > OK 
fast flexible filling design 
A design whose points are quasiuniformly distributed throughout the design space. Useful when the design region is not rectangular. 
DOE > Space Filling Design > Set factor level values to 0 and 1 > Continue > Specificy a sample size > Linear Constraint > Fast Flexible Filling > Make Table 
Fast Ward 
Modification of Ward’s method that is more efficient for large numbers of rows. 
Analyze > Multivariate Methods > Cluster > Options:Hierarchical, Fast Ward 
FDR 
The expected proportion of Type I errors, or false discoveries, when conducting multiple hypothesis tests. Used to control error rate in multiple testing. 
Analyze > Modeling > Response Screening > Select X and Y > OK 
final communality estimate 
In factor analysis, the sum of the squared loadings for a variable. Estimates the proportion of the variance of that variable that is explained by the common factors. 
Analyze > Consumer Research > Factor Analysis > OK > Final Community Estimates 
fishbone diagrams 
A hierarchical diagram to lay out root causes. Also called Ishikawa or cause and effect diagrams. 
Analyze > Quality and Process > Diagram 
Fisher’s Exact Test (2x2) 
For 2by2 tables only, the Contingency platform performs a Fisher’s Exact Test as well as a traditional ChiSquare. 
Analyze > Fit Y by X > Contingency 
Fisher’s Exact Test (mxn) 
JMP Pro can perform a Fisher’s exact test on an m x n contingency table. 
Analyze > Fit Y by X > Contingency > Exact Test > Fisher’s Exact Test 
fitting lines or polynomial 
Using the Fit Line command, you can add straight line fits to your scatterplot using least squares regression. Using the Fit Polynomial command, you can fit polynomial curves of a certain degree using least squares regression. Available in Fit Y by X as well as Graph Builder. 
General 
forecasting 
Predicting future values based on fitting a time series model to data. 
Analyze > Modeling > Time Series > Forecast Periods 
formula (Graph Builder) 
Shows a function defined by a column formula. 
Graph > Graph Builder > Formula 
forward selection 
Describes how an effect enters a model; forward brings in the regressor that most improves the fit, given that term is significant at the level specified. 
Analyze > Fit Model > Personality:Stepwise > Direction:Forward 
forward selection (forward stepwise regression) 
In stepwise regression, forward brings in the regressor that most improves the fit, given that term is significant at the level specified by Prob to Enter. 
Analyze > Fit Model: Stepwise > Direction: Forward 
fraction of design space (FDS) plot 
A way to see how much of the model prediction variance lies above (or below) a given value. 
DOE > Custom Design > Design Evaluation > Fraction of Design Space Plot 
fractional factorial design (2 and 3 level) 
An experimental design that is a factorial in nk of the n factors, with the remaining factors set to interactions of the first set. Used for screening designs. 
DOE > Custom Design > Design Evaluation > Fraction of Design Space Plot 
freq (Graph Builder) 
Drop a variable here to use it as a frequency or weight for graph elements that use statistics, such as mean or counts. 
Graph > Graph Builder > Freq 
frequency counts  2way crosstabulation 
Contingency Platform. 
Analyze > Fit Y by X > Contingency > Contingency Table 
frequency counts  general 
Summary Command. 
Analyze > Distribution > Freq 
frequency counts  oneway classification 
Distribution Platform. 
Analyze > Distribution > Freq 
frequency table, frequency distribution 
The Distribution platform shows the univariate frequency distribution for one or more categorical variables, showing the number of times each value occurs. Use Fit Y by X for the joint frequency distribution for two categorical variables. 
Analyze > Distribution, Analyze > Fit Y by X: Contingency 
Friedman’s test for nonparametric repeated measures ANOVA 
Not supported directly in JMP but can be determined by calculating the ranks within each block and then doing a two way ANOVA on ranks using Fit Y by X (treating one of the effects as a blocking variable). 
Analyze > Fit Y by X 
full factorial design 
Design that contains all combinations of the levels of the factors. 
DOE > Full Factorial Design 
G
Term  Definition  Example of how to access in JMP 

G chart 
A control chart for rare events that plots a count of units or occurrences between rare events. 
Analyze > Quality and Process > Control Chart Builder > Control Chart Types 
G^{2} 
For a node in a partition analysis of a categorical response, G² is 2 times the entropy. Candidate G² is the change in entropy if the next split occurred on that candidate. 
Analyze > Modeling > Partition 
Gabriel biplot 
A multivariate plot in principal components space, showing variable directions for both points and rays. 
Graph > Scatterplot 3D > Biplot Rays 
Gamma (measure of association) 
A nonparametric measure of ordinal association that uses the counts of concordant and discordant pairs. A pair is concordant if an observation with a larger value of X also has a larger value of Y. A pair is discordant if an observation with a larger value of X has a smaller value of Y. Only appropriate when both variables are ordinal. 
Analyze > Fit Y by X > Contingency > Measures of Association > Gamma 
GASP models 
Models a surface by interpolating across a set of data points with respect to a distancecovariance specification, with each coordinate distance component parameterized by a different value. 
Analyze > Modeling > Gaussian Process 
gauge chart 
Chart that is used to study how a measurement process varies across gauges. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart 
Gauge R&R 
For analyzing measurement systems, the variation is characterized as due to measurements (repeatability) or operators (reproducibility). Use the Variability Chart platform. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart > Gauge Studies > Gauge RR 
Gaussian process 
Models a surface by interpolating across a set of data points with respect to a distancecovariance specification, with each coordinate distance component parameterized by a different value. 
Analyze > Modeling > Gaussian Process 
Gaussian process Ioptimal design 
Design that minimizes the integrated mean squared error of the Gaussian process model over the experimental region. 
DOE > Space Filling Design > Space Filling Design Methods > Gaussian Process IMSE Optimal 
Gaussian process IMSE design 
Design that minimizes the integrated mean squared error of the Gaussian process model over the experimental region. 
DOE > Space Filling Design > Space Filling Design Methods > Gaussian Process IMSE Optimal 
Generalized Linear Models or GLIM 
An advanced technique for modeling nonnormally distributed responses. JMP’s Fit Model platform supports Normal, Poisson, and binomial distributions in its Generalized Linear Model Personality. 
Analyzed > Fit Model > Personality:Generalized Linear Model 
generalized R^{2} 
Generalization of the R^{2} measure of fit that simplifies to the regular R^{2} measure for continuous normal responses. It is similar to the entropy R^{2} measure, but instead of using the loglikelihood, it uses the 2/n root of the likelihood. 
Analyze > Fit Model > Personality:Nominal Logistic > Whole Model Test > Generalized RSquare 
generalized regression 
A collection of techniques that fit models using shrinkage techniques. The method accommodates a variety of response distributions. Two methods, the lasso and elastic net, perform variable reduction as part of the fitting procedure. Useful in fitting correlated and highdimensional data. 
Analyze > Fit Model > Generalized Regression Models 
geometric mean 
Indicates the central tendency or typical value of a set of numbers. The numbers are multiplied and then the n^{th} root of the resulting product is taken. 
Analyze > Distribution > Summary Statistics > Customize Summary Statistics > Geometric Mean 
goal plot 
A goal plot shows a point corresponding to the mean and standard deviation of each variable, standardized to its specification limits, on X and Y axes. 
Analyze > Quality and Process > Capability > Goal Plot 
goodness of fit 
Tests the hypothesis that the data comes from a particular distribution. 
Analyze > Distribution > Summary Statistics > Show All Modes > Fitted Generalized Logarithm > Goodness of Fit 
GreenhouseGeisser 
An adjustment to the degrees of freedom to adjust for compound symmetry violations in fitting a multivariate model with univariate calculations. 
Analyze > Fit Model > Personality:MANOVA > Univariate Tests Also > Choose Response:Repeated Measures > Within Subjects > Univar GG 
group X (Graph Builder) 
Subsets or partitions the data based on the variable or variables that you select. Displays the variable horizontally. Once a variable is placed here, no variable can be placed in Wrap. 
Graph > Graph Builder > Group X 
group Y (Graph Builder) 
Subsets or partitions the data based on the variable or variables that you select. Displays the variable vertically. 
Graph > Graph Builder > Group Y 
groups (grouping variables) 
Subset or partition the data based on a grouping variable. 
General 
H
Term  Definition  Example of how to access in JMP 

halfnormal plot 
Plots the absolute values of the estimates against the normal quantiles for the absolute value normal distribution. 
Analyze > Fit Model > Personality:Standard Least Square > Effect Screening > Normal Plot > Half Normal Plot 
hazard profiler 
Plots the hazard rate (or instantaneous failure rate) over time. 
Analyze > Reliability and Survival Methods > Reliability Block Diagrams > Show Hazard Profiler 
heatmap matrix (Graph Builder) 
Shows counts using color for X and Y categories. 
Graph > Graph Builder > Heatmap 
heteroschedasticity by groups 
Testing that the variances are different in different groups. 
Analyze > Fit Y by X > Oneway > Fit Line > Plot Residuals? 
hierarchical clustering 
A clustering technique that starts out with each point being its own cluster, then at each step combining the clusters that are closest to each other. 
Analyze > Multivariate Methods > Cluster > Options:Hierarchical 
histogram 
A bar graph in which the height of the bars represent frequency counts. Histograms help visualize the density of a distribution. 
Analyze > Distribution > Histogram Options > Histogram 
histogram (Graph Builder) 
Shows a variable’s distribution using binning. If you specify the same variable for X and Y, then the Y role is ignored and a single histogram appears. 
Graph > Graph Builder > Histogram 
historical mean 
Mean gathered from a previous process that can be used in computing Gauge R&R summaries. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart > Chart Type:Variability > Gauge Studies > Gauge RR > Historical Mean 
Hoeffding’s D 
A nonparametric measure of association. 
Analyze > Multivariate Methods > Multivariate > Nonparametric Correlations > Hoeffding’s D 
homogeneity of variances, across groups 
Testing that the variances are equal in a oneway layout and providing a weighted (Welch) Anova in case they are not. 
Analyze > Fit Y by X > Oneway > Unequal Variances 
Hotelling’s T^{2} 
A measure of multivariate distance that takes into account the variances and covariances. Equal to the squared Mahalanobis distance. 
Analyze > Multivariate Methods > Multivariate > Outlier Analysis > T^{2} 
HotellingLawley Trace 
Four multivariate tests are supported in the MANOVA personality of the Fit Model platform. Wilks’ Lambda, Pillai’s Trace, HotellingLawley Trace, and Roy’s Maximum Root Criterion. 
Analyze > Fit Model > Personality:MANOVA > Choose Response:Identity > Whole Model > HotellingLawley 
Hsu’s MCB test 
Test that a level has the highest or lowest mean in a oneway ANOVA. See ‘multiple comparisons’. 
Analyze > Fit Y by X > Oneway > Compare Means > With Best, Hsu MCB 
HuynhFeldt 
An adjustment to the degrees of freedom to adjust for compound symmetry violations in fitting a multivariate model with univariate calculations. 
Analyze > Fit Model > Personality:MANOVA > Univariate Tests Also > Choose Response:Repeated Measures > Within Subjects > Univar HF 
hypergeometric 
The hypergeometric distribution models the total number of successes in a fixed sample drawn without replacement from a finite population. 
Cols > Formula > Hypergeometric 
hypothesis testing 
Hypothesis tests are used to answer the question: Assuming the null hypothesis is true, what is the probability of observing a test statistic that is at least as extreme as the value that was observed? 
General 
hypothesis testing for one sample means 
A test that the mean is some hypothesized value. 
Analyze > Distribution > Test Mean 
hypothesis testing for one sample two level proportions 
Test if proportions are different than hypothesized values. In the twosided case, this test is a chisquare test; in either of the onesided cases, this test is an exact onesided binomial test. 
Analyze > Distribution > Test Mean 
I
Term  Definition  Example of how to access in JMP 

IMR chart 
An individual measurement chart and moving range chart. 
Analyze > Quality and Process > Control Chart > IR > Individual Measurement/Moving Range (Also available through Control Chart Builder) 
IMRR chart 
Uses both betweensubgroup and withinsubgroup variations to generate an individuals, moving range, and R chart. 
Analyze > Quality and Process > Control Chart > XBar > R (Also available through Control Chart Builder) 
IMRS chart 
Uses both betweensubgroup and withinsubgroup variations to generate an individuals, moving range, and S chart. 
Analyze > Quality and Process > Control Chart > XBar > S 
Ioptimal design 
Design that minimizes the average variance of prediction over the region of the data. 
DOE > Custom Design > Optimality Criterion > Make IOptimal Design 
ICC (intraclass correlation) 
Indicates the proportion of the total variation that you can attribute to the part. 
Analyze > Quality and Process > Measurement Systems Analysis > EMP Results > Intraclass Correlation 
impute missing data 
Imputes data in a partial least squares analysis using either an iterative EMalgorithm or the average value for that variable. The EM method produces a Missing Value Imputation report. 
Analyze > Multivariate Methods > Partial Least Squares > Impute Missing Data 
independence 
Independence of two events means that the occurrence of one event does not make the other event more or less likely to occur. 
General 
indicator variables 
The indicator variable uses the value of 1 for the chosen category or level profile row and 0 elsewhere. 
General 
individual measurement chart 
An individual measurement chart and moving range chart. 
Analyze > Quality and Process > Control Chart > IR > Individual Measurement (Also available through Control Chart Builder) 
inertia 
The total Pearson Chisquare for a twoway frequency table divided by the sum of all observations in the table. 
Analyze > Consumer Research > Multiple Correspondence Analysis 
informative missing 
Method for handling missing data that adds an indicator variable for missing continuous values into the model and uses the mean for the missing values. For categorical columns, the missing value is treated as a valid (nonmissing) level of the variable. 
Analyze > Fit Model > Model Specification 
interaction plots  profile plots 
How the response varies differently over one factor depending on levels of another factor. 
Analyze > Fit Model > Personality:Standard Least Square > Profilers > Profiler > Interaction Profiler 
interactions 
Crossproduct or interaction terms. See Fit Model documentation. 
General 
intercept 
Point where a line crosses an axis. 
General 
interquartile range 
The difference between the first and third quartiles. Displayed in a box plot. Box plots are also available in Distribution and Graph Builder 
Analyze > Fit Y by X: Oneway > Quantiles 
inverse prediction 
Predicting which x value led to a particular y value, given other values. 
Analyze > Fit Y by X > Inverse Prediction 
IQR (interquartile range) 
Difference between the first and 3rd quartiles. 
Analyze > Distribution > Summary Statistics > Customize Summary Statistics > Interquartile Range 
IR chart 
An individual measurement chart and moving range chart. 
Analyze > Quality and Process > Control Chart > IR (Also available through Control Chart Builder) 
Ishikawa diagrams 
A hierarchical diagram to lay out root causes. Also called fishbone or cause and effect diagrams. 
Analyze > Quality and Process > Diagram 
item analysis 
Using Item Response Theory (IRT), jointly models ability and probability of answering each question correctly from a data set of correct and incorrect question responses. 
Analyze > Multivariate Methods > Item Analysis 
item analysis  Cronbach’s a 
A measure of item reliability on how consistently a set of variables measures overall response. 
Analyze > Multivariate Methods > Multivariate > Item Reliability > Cronbach’s a 
item response theory 
Item Response Theory (IRT) jointly models ability and probability of answering each question correctly from a data set of correct and incorrect question responses. 
Analyze > Multivariate Methods > Item Analysis 
iteratively reweighted least squares 
A technique where the weights depend on the estimates, and are thus done iteratively. Only the Nonlinear platform supports recalculated weights. 
Analyze > Modeling > Nonlinear 
J
Term  Definition  Example of how to access in JMP 

jackknife values (outlier analysis) 
Values obtained using the jackknife method, which is a resampling method that estimates sampling variance by sequentially running an analysis with m observations deleted. The delete1 jackknife is used in JMP. 
Analyze > Multivariate Methods > Multivariate > Outlier Analysis > Jackknife Distances 
join data tables 
Join tables side by side, or match values from one table in the other. 
Tables > Join 
K
Term  Definition  Example of how to access in JMP 

K means cluster 
The kmeans approach to clustering performs an iterative alternating fitting process to form the number of specified clusters. 
Analyze > Multivariate Methods > Cluster > Options:KMeans 
K out of N node 
A KoutofN node in a reliability block diagram requires that K of the N paths leading into it are functional in order for the system to function. 
Analyze > Realiability and Survival Methods > Reliability Block Diagram > Weibull 
kfold cross validation 
Divides the original data into K subsets. In turn, each of the K sets is used to validate the model fit on the rest of the data, fitting a total of K models. 
General 
KaplanMeier Survival Estimates 
A stepfunction estimate for the univariate survival distribution function. Also called Product limit estimates. 
Analyze > Reliability and Survival > Survival > Plot Options > Show Kaplan Meier 
kappa statistic 
measures the degree of agreement of two similarly valued categorical variables on a scale up to 1. JMP does this automatically in the Contingency platform when the two variables have the same set of values. 
Analyze > Fit Y by X > Contingency > Exact Test > Exact Agreement Statistic > Kappa Coefficient 
Kendall’s t 
A nonparametric measure of association. It is based on counting concordances and discordances on comparisons of pairs of rows on pairs of columns. 
Analyze > Multivariate Methods > Multivariate > Nonparametric Correlations > Kendall’s t 
Kendall’s TauB 
A nonparametric measure of ordinal association that uses the counts of concordant and discordant pairs. A pair is concordant if an observation with a larger value of X also has a larger value of Y. A pair is discordant if an observation with a larger value of X has a smaller value of Y. Only appropriate when both variables are ordinal. 
Analyze > Fit Y by X > Contingency > Measures of Association > Kendall’s TauB 
knot node 
A knot node in a reliability block diagram allows you to configure a KoutofN block shape for shapes having different distribution property settings. 
Analyze > Realiability and Survival Methods > Reliability Block Diagram > Weibull 
KolmogorovSmirnov test 
Uses the empirical distribution function to test whether the distribution of the response is the same across groups. 
Analyze > Fit Y by X > Oneway > Nonparametric > Kolmogorov Smirnov Test 
KolmogorovSmirnovLilliefors test 
A test that a distribution is normally distributed. In the Distribution platform, when you fit a Normal, this test is used if n>2000. Otherwise a more powerful test is used, the ShapiroWilk test. 
Analyze > Distribution > Continuous Fit > Normal > Goodness of Fit > KolmogorovSmirnovLilliefors Test 
kriging 
Models a surface by interpolating across a set of data points with respect to a distancecovariance specification, with each coordinate distance component parameterized by a different value. 
Analyze > Modeling > Gaussian Process 
KruskalWallis test 
A test that compares several distributions by ranking the data and comparing the ranks from each group. 
Analyze > Fit Y by X > Oneway > Nonparametric > Wilcoxon Test 
kurtosis 
The statistic that measures the 4^{th} moment. 
Analyze > Distribution > Summary Statistics > Customize Summary Statistics > Kurtosis 
L
Term  Definition  Example of how to access in JMP 

L18 L36 designs 
Orthogonal Designs. 
DOE > Custom Design 
lack of fit 
The difference between the total error from the fitted model and pure error is called LackofFit error. It represents all the terms that might have been added to the model, but were not. 
Analyze > Fit Model > Personality:Nominal Logistic > Lack of Fit 
lack of fit 
Test to compare error against pure error due to exact replicates. 
Analyze > Fit Model > Personality:Standard Least Square > Regression Reports > Lack of Fit 
lagged variables 
A lagged variable is a variable that takes on its past values. 
Analyze > Modeling > Time Series 
lasso 
A generalized regression estimation method that applies an L1 (absolute value) penalty to the likelihood when estimating parameters. 
Analyze > Fit Model > Personality > Generalized Regression > Estimation Method 
Latin Hypercube 
A design that tries to fill space such that it is distributed evenly along each factor. 
DOE > Space Filling Design > Space Filling Design Methods:Latin Hypercube 
Latin hypercube design 
A design that tries to fill space such that it is distributed evenly along each factor. 
DOE > Space Filling Design > Space Filling Design Methods:Latin Hypercube 
least squares means 
The predicted value at each level of the indicated term, with other terms being set to neutral values. The Fit Model platform produces these automatically for nominal terms. 
Analyze > Fit Model > Personality:Standard Least Squares 
least squares regression 
Regression method that minimizes the sum of squared differences from each point to the fitted line (or curve). 
Analyze > Fit Model > Personality:Standard Least Squares 
leaveoneout 
A cross validation technique where every observation is used as a validation set. For each observation, the model is estimated on the rest of the data and validated on that observation. The validation measures are combined over all observations. 
Analyze > Multivariate Methods > Partial Least Squares > Validation Method 
legend (Graph Builder) 
Shows descriptions of graph elements. If you attempt to drop a variable here, the variable defaults to Overlay. 
Graph > Graph Builder > Legend 
Lenth’s Pseudostandard error 
For saturated models when the residual standard error cannot be estimated well, if you assume that most of your effects are ‘inactive’, This is a way to use the inactive effects to estimate the error. 
Analyze > Screening 
Levene’s test/Testing Variances across groups 
Testing that the variances are equal in a oneway layout and providing a weighted (Welch) Anova in case they are not. 
Analyze > Fit Y by X > Oneway > Analysis of Means Method > ANOM for Variances with Levene (ADM) 
leverage plot 
A plot such that the distance from a point to the sloped line is the residual, and the distance to the horizontal line is what the residual would be under the hypothesis. 
Analyze > Fit Model > Personality:Standard Least Squares > Leverage Plot 
Levey Jennings chart 
Charts that show a process mean with control limits based on a longterm σ. The control limits are placed at 3*σ distance from the center line. 
Analyze > Quality and Process > Control Chart > Levey Jennings 
lift curve 
Similar to an ROC curve, but constructed to show the initial ordering. 
Analyze > Modeling > Partition > Lift Curve 
likelihoodratio ChiSquare 
Formed by twice the difference in the loglikelihoods due to the hypothesis. The likelihoodratio ChiSquare statistic is found in many different platforms. 
Analyze > Fit Y by X > Contingency > Tests > Likelihood Ratio (Also available through Fit Model) 
line (fitting, equation for) 
See Fitting lines or polynomial. 
General 
line of fit (Graph Builder) 
Shows a linear regression with confidence intervals. 
Graph > Graph Builder > Line of Fit 
line plot (Graph Builder) 
Shows a response summarized by categories. 
Graph > Graph Builder > Line 
linear discriminant analysis 
Classifying points using discriminant analysis with the same withincovariance matrix for all groups. See ‘discriminant analysis’. 
Analyze > Multivariate Methods > Discriminant > Discriminant Method:Linear, Common Covariance 
linear regression 
An approach to modeling the relationship between a scalar variable y and one or more explanatory variables denoted X. 
Analyze > Fit Y by X > Fit Line 
linearity study 
Performs a regression analysis using the standard variable as the X variable, and the bias as the Y. This analysis examines the relationship between bias and the size of the part. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart > Gauge Studies > Linearity Study 
loglinear models 
Enables you to model both the expected value and the variance of a response using regression models. The log of the variance is fit to one linear model simultaneously with the expected response fit to a different linear model. 
Analyze > Fit Model > Personality:Loglinear Variance 
loglogistic survival model; see Nonlinear documentation 
A parametric survival distribution. Not supported directly by JMP, but you could use Nonlinear. 
Nonlinear platform. 
lognormal survival model  regression 
A distribution for modeling survival times. 
Analyze > Reliability and Survival > Survival > LogNormal Fit 
lognormal survival model  univariate 
A distribution for modeling survival times. 
Analyze > Reliability and Survival > Survival > LogNormal Fit 
logrank test (Survival) 
Test that the survival distribution is the same across groups. 
Analyze > Reliability and Survival > Survival > Tests Between Groups > LogRank 
logarithms 
JMP can perform log transformations using a formula, or in Fit Y by X (bivariate fit special). Graphs can also be loglog or semilog x or y. 
General 
logistic regression 
Fitting the probability of a categorical response to a linear model. 
Analyze > Fit Model > Personality:Nominal Logistic 
logit transformation 
Calculates the inverse of the logistic function for the selected column. The column values must be between 0 and 1. 
Analyze > Fit Model > Construct Model Effects > Transform 
LSD  least significant difference 
Tests of pairwise comparisons using Student’s t test. 
Analyze > Fit Y by X > Oneway > Compare Means > Mean Comparisons > LSD Threshold Matrix 
LSN  least significant N 
The smallest sample size that would still yield a significant test statistic given the α level, effect size, and error variance. 
Analyze > Fit Model > Personality:Standard Least Square > Estimates > Parameter Power > LSN 
LSV  least significant Value 
The smallest value of the estimate that would still yield a significant test statistic given the α, sample size, and error variance. Also the radius of the confid. interval for an estimate. 
Analyze > Fit Model > Personality:Standard Least Square > Estimates > Parameter Power > LSV 
M
Term  Definition  Example of how to access in JMP 

Mahalanobis Distance 
Measures how far a multivariate point is from a multivariate mean with respect to the covariance structure. 
Analyze > Multivariate Methods > Multivariate > Outlier Analysis > Mahalanobis Distance 
Mallow’s C(p) 
Mallow’s C(p) is a measure to use when comparing models. Usually C(p) is plotted against p, the number of regressors. 
Analyze > Fit Model > Personality:Stepwise > Cp 
MannWhitney U Test 
A test that is exactly equivalent to the Wilcoxon 2sample (or Kruskal Wallis ksample) test. See ‘Wilcoxon two group test’. 
Analyze > Fit Y by X > Oneway > Nonparametric > Wilcoxon Test 
MANOVA  multivariate analysis of variance 
When several responses are fit to the same linear model, and tests are needed that go across the responses. 
Analyze > Fit Model > Personality:MANOVA 
MantelHaenszel test 
Computes ChiSquare statistics for stratifications of a twoway contingency table by a third variable. 
Analyze > Fit Y by X > Contingency > Cochran Mantel Haenszel 
map shape (Graph Builder) 
Drop variables here to create map shapes. See Create Map Shapes. If you have a variable in the Map Shape zone, the X and Y zones disappear. 
Graph > Graph Builder > Map Shape 
map shapes (Graph Builder) 
Creates a map on the graph. 
Graph > Graph Builder > Map Shapes 
margin of error 
A measure of sampling error that includes the standard error along with a percentile from a test statistic. 
General 
marginal distributions (marginal probability) 
The probability distribution of one variable derived from a joint probability distribution with a second variable. 
General 
marginal means 
The predicted value at each level of the indicated term, with other terms being set to neutral values. The Fit Model platform produces these automatically for nominal terms. 
Analyze > Fit Model > Personality:Standard Least Square > LSMeans 
matched pairs sign test 
This is a nonparametric version of the paired ttest that uses only the sign (positive or negative) of the difference for the test. 
Analyze > Matched Pairs > Sign Test 
Mauchly Criterion 
Multivariate problems can be made univariate if the covariances have ‘sphericity’. The Mauchly Criterion is for testing this assumption of a spherical covariances. 
General 
maximum entropy design 
Design that optimizes a measure of the amount of information contained in an experiment. 
DOE > Space Filling Design > Space Filling Design Methods:Maximum Entropy 
maximum likelihood 
A general method in statistics to get estimates that maximize the likelihood, i.e. adjust the estimates to maximize the probability attributed to the data that you actually have. Many platforms in JMP already use maximum likelihood. If your problem does not fit, you might try using the Nonlinear platform, which enables you to specify the negative loglikelihood as a loss function to minimize. 
General 
maximum likelihood factor analysis 
Finding directions among a large set of variables that seem to simplify the structure of the variables into a small number of ‘factors’. Usually the factor solution is rotated to be more interpretable. JMP implements three types of factor analysis: principal component analysis, noniterated pincipal factor analysis with SMC, and maximum likelihood factor analysis. 
Analyze > Multivariate Methods > Multivariate > Factor Analysis 
maximum R^{2} 
The maximum attainable value of R^{2} for the given data if you had to fit a parameter for each unique combination of the regressors. Maximum R^{2} is included in the lackoffit report in many regression reports. 
Analyze > Fit Model > Personality:Standard Least Square > Regression Reports > Lack of Fit > Max R Sq 
McNemar’s test 
A nonparametric method used on nominal data; this test is applied to a 2x2 contingency table, with matched pairs of subjects, to determine whether the row and column marginal frequencies are equal. Bowker’s Test is a generalization of McNemar’s Test. 
Analyze > Fit Y by X > Contingency > Exact Test > Exact Agreement Statistic > Bowker’s Test 
mean of a single population 
The expected value of the underlying distribution for the response variable. 
Analyze > Distribution > Summary Statistics > Custom Summary Statistics > Means 
mean time to failure (MTTF) 
Mean or average life in reliability analysis. 
Analyze > Realiability and Survival Methods > Reliability Block Diagram > Mean Time to Failure 
meandifference plot 
A plot for matched pairs analysis which shows the relationship between the differences versus the means of the paired observations. 
Analyze > Matched Pairs > Plot Dif by Mean 
means across groups grouping facility 
Use the Summary Command in Tables menu. 
Tables > Summary 
means across groups oneway layout 
Use the Oneway Platform, or Fit Y by X. 
Analyze > Fit Y by X > Oneway 
measurement systems analysis (MSA) 
A platform that evaluates a measurement system for numeric responses using either traditional Gauge R&R or the EMP (Evaluating the Measurement Process) method. 
Analyze > Quality and Process > Measurement Systems Analysis 
median  across groups 
The middle value, for which half the data is above, and half below. 
Analyze > Fit Y by X > Oneway > Nonparametric > Median Test 
median  one sample 
The middle value, for which half the data is above, and half below. 
Analyze > Distribution > Summary Statistics > Customize Summary Statistics > Median 
median absolute deviation (MAD) 
Robust measure of the variability of a univariate sample of quantitative data. It is the median of the absolute deviations from the median of the data. 
Analyze > Distribution > Summary Statistics > Customize Summary Statistics > Median Absolute Deviation 
median test 
A test that compares several distributions by finding the median and counting how many in each group are greater than the median. 
Analyze > Fit Y by X > Oneway > Nonparametric > Median Test 
minimum potential design 
Space filling design that spreads points out inside a sphere. 
DOE > Space Filling Design > Space Filling Design Methods:Minimum Potential 
missing data pattern 
Shows pattern of missing values in data table. 
Tables > Missing Data Pattern 
mixed model 
Random effects are effects, like subjects, where the levels are randomly selected from a larger population, and their effect on the response can be assumed to vary normally with some variance (the variance component). In Fit Model there are two methods of estimating mixed models. 
Analyze > Fit Model > Personality:Standard Least Square 
mixed model 
A continuousresponse linear model that can include both fixed and random effects as well as a specified covariance structure. Such models include random coefficient, repeated measures, splitplot, spatial, and hierarchical models. 
Analyze > Fit Model > Mixed Models 
mixedlevel designs 
Experimental designs when the factors do not all have the same number of levels. 
DOE > Custom Design 
mixture designs 
Experimental designs where the factors sum to 1, as ingredients to a mixture. 
DOE > Mixture Design 
mode 
The value that occurs most frequently in a set of univariate data. 
Analyze > Distribution > Summary Statistics > Customize Summary Statistics > Mode 
model averaging 
Average fits for a number of models, resulting in a model with improved prediction capability. 
Analyze > Fit Model > Personality:Stepwise > Model Averaging 
model comparison 
Compares the fit of different models. Provides measures of fit, ROC, diagnostic plots, model averaging and profilers. 
Analyze > Modeling > Model Comparison 
moments 
Quantitative measures of the shape of a set of points. 
Analyze > Distribution > Summary Statistics > Customize Summary Statistics 
monitor classification 
Don Wheeler’s classification method that helps interpret the intraclass correlation. 
Analyze > Quality and Process > Measurement Systems Analysis > EMP Results > Intraclass Correlation 
mosaic plot 
A mosaic plot is like a set of sidebyside divided bar charts, with the sides of rectangles showing the marginal or conditional rates and the areas showing frequency counts. 
Analyze > Fit Y by X > Contingency > Mosaic Plot 
mosaic plot (Graph Builder) 
Shows counts using size for X and Y categories. 
Graph > Graph Builder > Mosaic 
moving average 
Simple moving averages are calculated by taking arithmetic means based on a given set of most recent values. 
General 
moving range chart 
Control chart that displays the moving ranges of two or more successive measurements. 
Analyze > Quality and Process > Control Chart > Control Chart Builder > Right Click:Limits > Sigma > Moving Range (Also available through Control Chart Builder) 
MTTF 
Mean or average life in reliability analysis. 
Analyze > Realiability and Survival Methods > Reliability Block Diagram > Mean Time to Failure 
multivari chart 
Shows variation from group to group. Like a Shewhart Control Chart, but used for variation across measurement groups, rather than across time. Use the Variability Chart platform. 
Analyze > Quality and Process > Control Chart > Multivariate Control Chart 
multicollinearity 
Collinearity is when the factors in a model are almost linear combinations of other factors. Collinearity shows as a shrunken X scale on the factors’ leverage plots. 
General 
multinomial logit 
Fits choice models for market research. Conjoint Experiments. 
See Choice Analysis 
multiple comparisons 
Tests that compare group means in a oneway ANOVA, especially those that recognize that special care is needed to protect the significant level for multiple tests. JMP offers Tukey HSD for a general test, Hsu MCB for testing extreme means, and Dunnett’s test for testing with respect to a control group. 
Analyze > Fit Y by X > Oneway > Nonparametric > Median Test 
multiple comparisons 
Comparisons involving several userdefined groups. Includes comparisons with overall average, with a control, Tukey and Student’s t pairwise comparisons, and equivalence tests. 
General 
multiple correspondence analysis 
Extends correspondence analysis to more than two nominal or ordinal variables. 
Analyze > Consumer Research > Multiple Correspondence Analysis 
multiple regression 
When a response is predicted by a linear combination of several factors. 
Analyze > Fit Model > Personality:Standard Least Square 
multivariate analysis 
A wide collection of methods that analyze a group of responses. 
Analyze > Multivariate Methods > Multivariate 
multivariate analysis of variance 
When several responses are fit to the same linear model, and tests are needed that go across the responses. 
Analyze > Fit Model > Personality:MANOVA 
multivariate control chart 
Creates a chart to view summaries for monitoring problems where several related variables are of interest. 
Analyze > Quality and Process > Multivariate Control Chart 
multivariate tests 
Four multivariate tests are supported in the MANOVA personality of the Fit Model platform. Wilks’ Lambda, Pillai’s Trace, HotellingLawley Trace, and Roy’s Maximum Root Criterion. 
Analyze > Fit Model > Personality:MANOVA > Choose Response:Identity > Identity > Whole Model 
N
Term  Definition  Example of how to access in JMP 

near orthogonal design 
A screening design with orthogonal main effects. Useful when interactions are considered negligible. 
DOE > Screening Design > Continue > Near Orthogonal 
needle plot 
A plot where lines are drawn from zero to the point. 
Graph > Chart > Options:Needle Plot 
nested designs, effects 
In an experiment, when the label of a term B necessarily involves another term A, it is ‘nested’ and the term is written ‘B[A]’. The Fit Model dialog supports this with the ‘Nest’ button. 
Analyze > Fit Model > Nest > Personality:Standard Least Square 
neural net 
Neural Network. Flexible fitting of Y’s to X’s within a specific framework of layering and sshaped functions. 
Analyze > Modeling > Neural 
NIPALS 
Method for partial least squares analysis that works by extracting one factor at a time. By working on one factor at a time, NIPALS does not require calculation of the overall covariance matrix. 
Analyze > Multivariate Methods > Partial Least Squares 
nominal factors 
Nominal factors are categorical factors where there is no special treatment for the ordering of the values. In JMP fitting, they are parameterized with respect to the difference of each level from the average over levels. 
General 
nonlinear design 
A nonlinear design is an optimal design that is nonlinear in its parameters. Such a design is generated using data tables that contain factors, a single response, and a formula column for the model. 
DOE > Nonlinear design 
nonlinear iterative partial least squares (NIPALS) 
Method for partial least squares analysis that works by extracting one factor at a time. By working on one factor at a time, NIPALS does not require calculation of the overall covariance matrix. 
Analyze > Multivariate Methods > Partial Least Squares 
nonlinear regression 
Fitting equations that are nonlinear in the parameters. Nonlinear regression requires that you specify a formula for the column with parameters to estimate. Nonlinear regression is an iterative method that has no guarantees that it will converge to the best estimate unless you start out near enough to it. 
Analyze > Modeling > Nonlinear 
nonparametric density 
Shows patterns in the point density of a scatterplot, which is useful when the scatterplot is so darkened by points that it is difficult to distinguish patterns. 
Analyze > Fit Y by X > Nonpar Density 
nonparametric: exact KolmogorovSmirnov test 
Performs the exact version of the KolmogorovSmirnov test. This option is available only when the X factors has two levels. 
Analyze > Fit Y by X > Oneway > Nonparametric > Exact Test > Kolmogorov Smirnov Exact Test 
nonparametric: exact van der Waerden test 
Performs the exact version of the van der Waerden test. This option is available only when the X factor has two levels. 
Analyze > Fit Y by X > Oneway > Nonparametric > Exact Test > Van Der Waerden Exact Test 
nonparametric: exact Wilcoxon test 
Performs exact versions of the Wilcoxon test. This option is available only when the X factor has two levels. 
Analyze > Fit Y by X > Oneway > Nonparametric > Exact Test > Wilcoxon Exact Test 
nonparametric: Hoeffding’s D 
A nonparametric measure of association. 
Analyze > Multivariate Methods > Multivariate > Nonparametric Correlations > Hoeffding’s D 
nonparametric: Kendall’s t 
A nonparametric measure of association. It is based on counting concordances and discordances on comparisons of pairs of rows on pairs of columns. 
Analyze > Multivariate Methods > Multivariate > Nonparametric Correlations > Kendall’s t 
nonparametric: KolmogorovSmirnov test 
Uses the empirical distribution function to test whether the distribution of the response is the same across groups. 
Analyze > Fit Y by X > Oneway > Nonparametric > Kolmogorov Smirnov Test 
nonparametric: KruskalWallis 
A test that compares several distributions by raking the data and comparing the ranks from each group. 
Analyze > Fit Y by X > Oneway > Nonparametric > Wilcoxon Test 
nonparametric: Median test 
A test that compares several distributions by finding the median and counting how many in each group are greater than the median. 
Analyze > Fit Y by X > Oneway > Nonparametric > Median Test 
nonparametric: Spearman’s Rho 
Spearman’s Rho is a correlation coefficient computed on the ranks of the data values instead of on the values themselves. 
Analyze > Multivariate Methods > Multivariate > Nonparametric Correlations > Spearman’s Rho 
nonparametric: Steel with control test 
Performs the SteelDwass test on each pair. 
Analyze > Fit Y by X > Oneway > Nonparametric > Nonparametric Multiple Comparisons > Steel with Control 
nonparametric: SteelDwass all pairs test 
Performs the SteelDwass test on each pair. This is the nonparametric version of the All Pairs, Tukey HSD option found on the Compare Means menu. 
Analyze > Fit Y by X > Oneway > Nonparametric > Nonparametric Multiple Comparisons > SteelDwass All Pairs 
nonparametric: Van der Waerden test 
A test that compares several distributions by ranking the data, using the ranks to form normal scores and comparing the mean scores across groups. 
Analyze > Fit Y by X > Oneway > Van Der Waerden Test 
nonparametric: Wilcoxon signedranks test 
Nonparametric test that a mean is equal to a given value. 
Analyze > Distribution > Test Mean > Wilcoxon Signed Rank 
nonparametric: Wilcoxon test 
Performs the Wilcoxon test (rank test for errors with logistic distributions) on each pair, and does not control for the overall α level. 
Analyze > Fit Y by X > Oneway > Nonparametric > Wilcoxon Test 
nonparametric: Wilcoxon two group test 
A test that compares several distributions by ranking the data and comparing the ranks from each group. 
Analyze > Fit Y by X > Oneway > Nonparametric > Wilcoxon Test 
normal curve 
The Normal or Gaussian distribution is the familiar bellshaped curve, the distribution that arises of summing many small independent random values. 
Analyze > Distribution > Continuos Fit > Normal 
normal mixtures clustering 
Clustering method appropriate when the data are assumed to come from a mixture of multivariate normal distributions. 
Analyze > Multivariate Methods > Cluster > Options:KMeans > Method:Normal Mixtures 
normal mixtures fit 
Fits a mixture of k normal distributions, where k is the number of groups or clusters. 
Analyze > Distribution > Continuous Fit > Normal Mixtures 
normal plot 
Helps you identify effects that deviate from the normal lines. 
Analyze > Fit Model > Personality:Standard Least Square > Effect Screening > Normal Plot 
normal probability plot 
A plot of a batch of values versus normal quantiles calculated on ranks. If the data is normal, the points tend to be in a straight line. 
Analyze > Distribution > Normal Quantile Plot 
normal quantile plot 
A plot of a batch of values versus normal quantiles calculated on ranks. If the data is normal, the points tend to be in a straight line. 
Analyze > Distribution > Normal Quantile Plot 
normal quantile plot by group 
A plot of a batch of values versus normal quantiles calculated on ranks. If the data is normal, the points tend to be in a straight line. 
Analyze > Fit Y by X > Oneway > Normal Quantile Plot 
normal quantile plot of effects 
In Fitting models with large numbers of effects assuming that only a few effects are nonzero, plotting the effect sizes in a normal quantile plot helps identify which effects are active, rather than due to random variation. The feature here ensures that the effects are uncorrelated and have the same variance. 
Analyze > Personality:Standard Least Square > Effect Screening > Normal Plot 
normality: normal curve on histogram 
Use the Distribution Platform, Fit Distribution command, Normal. 
Analyze > Distribution > Histogram Options > Histogram > Continuous Fit > Normal 
normality: quantile plot 
A plot of a batch of values versus normal quantiles calculated on ranks. If the data is normal, the points tend to be in a straight line. 
Analyze > Distribution > Normal Quantile Plot 
normality: ShapiroWilk and KSL tests 
To test if a distribution is nonnormal. The ShapiroWilk test is used up to sample sizes of 2000, and the KolmogorovSmirnovLilliefors test is used above that. 
Analyze > Distribution > Continuous Fit > Normal > Goodness of Fit 
normalized box plots 
Shows one box plot for each column, where the box plots are standardized to a mean of zero and a standard deviation of one. 
Analyze > Quality and Process > Capability > Normalized Box Plots 
NP chart 
A plot showing the numbers of nonconforming items in subgroup samples. 
Analyze > Quality and Process > Control Chart > NP 
O
Term  Definition  Example of how to access in JMP 

O’Brien’s test 
Testing that the variances are equal in a oneway layout and providing a weighted (Welch) Anova in case they are not. 
Analyze > Fit Y by X > Oneway > Unequal Variances 
OC (operating characteristic) curve 
The OC curve shows how the probability of accepting a lot changes with the quality of the sample. 
Analyze > Quality and Process > Control Chart > XBar > Chart Type:XBar > OC Curve 
odds ratio confidence interval 
The odds ratios are exponentiated linear functions of the parameters. Therefore, the odds ratio confidence interval is calculated by computing a confidence interval for the linear function of the parameters and exponentiating the resulting interval. 
Analyze > Fit Model > Personality:Nominal Logistic > Odds Ratio 
odds ratios 
The ratio of the probability that the event of interest occurs versus the probability that it does not occur. 
Analyze > Fit Model > Personality:Nominal Logistic > Odds Ratio 
one sample ttest 
A test that the mean is some hypothesized value. 
Analyze > Fit Y by X > Oneway > Compare Means > Each Pair, Student’s T 
one sample two level proportion test 
Test if proportions are different than hypothesized values. In the twosided case, this test is a chisquare test; in either of the onesided cases, this test is an exact onesided binomial test. 
Analyze > Fit Y by X > Contingency > Tests 
one way ANOVA 
Fitting means across a grouping variable, and testing if they are significantly different. 
Analyze > Fit Y by X > Oneway > ANOVA 
operating characteristic curve 
The OC curve shows how the probability of accepting a lot changes with the quality of the sample. 
Analyze > Quality and Process > Control Chart > XBar > Chart Type:XBar > OC Curve 
optimal Bayesian Doptimal designs 
This type of design allows the precise estimation of all of the Necessary terms while providing omnibus detectability and some estimability of the If Possible terms. 
DOE > Custom Design 
optimal Bayesian Ioptimal designs 
This type of design minimizes the average prediction variance over the design region and at the same time has the ability to detect and estimate some higherorder terms. 
DOE > Custom Design 
optimal design 
An experimental design that minimizes the determinant of the variance matrix of the regression parameters. The DOE platform does this under Custom Design. 
Go to custom designer. 
optimal randomized block designs 
Optimal designs with a blocking factor where runs within a block are more homogeneous than runs in different blocks. 
Go to custom designer. 
optimal splitsplit plot and stripplot designs 
Designs in which the VeryHardtochange factors stay fixed within each whole plot. In the middle stratum, the Hardtochange factors stay fixed within each subplot. 
General 
ordinal factors 
Ordinal factors are categorical factors where the ordering of levels matters. In JMP fitting, they are parameterized with respect to the difference of each level from the previous level. For example, the design matrix columns are coded [level>=2], [level>=3] and so on, and the parameters measure the response mean differences [level=2][level=1], [level=3][level=2], and so on. 
General 
ordinal logistic regression 
Fitting an ordinal (ordered categorical) response probabilities to a linear model. The model fits the probability that a response is less than or equal to specific response levels. Ordinal logistic regression is distinguished from nominal logistic regression in that each response level needs only one new intercept parameter rather than a full new set of parameters for each effect in the model. 
Analyze > Fit Model > Personality:Ordinal Logistic 
ordinal responses 
When a response in a model has modeling type "ordinal", then the fitting system uses ordinal logistic regression to fit it. 
Analyze > Fit Model > Personality:Ordinal Logistic 
orthogonal array design 
Any design in which the columns are orthogonal to each other. 
General 
orthogonal regression 
Regression method that minimizes the orthogonal (perpendicular) distances from the data points to the fitted line. 
Analyze > Fit Y by X > Fit Orthogonal 
outlier box plot 
A box plot enclosing the inner quartiles of points with lines to the farthest point within 1.5 interquartile ranges from the quartiles. 
Analyze > Distribution > Outlier Box Plot 
outlier: Mahalanobis distance 
Measures how far a multivariate point is from a multivariate means with respect to the covariance structure. 
Analyze > Multivariate Methods > Multivariate > Outlier Analysis > Mahalanobis Distance 
overlay plot (Graph Builder) 
Groups the Y variables by the selected variables, overlays the response, and marks the levels with different colors. 
Graph > Graph Builder > Overlay 
overlay plots 
Use the Overlay Plot platform. 
Graph > Overlay Plot 
P
Term  Definition  Example of how to access in JMP 

P chart 
A plot showing the proportions of nonconforming items in subgroup samples. 
Analyze > Quality and Process > Control Chart > P 
pvalue 
The probability of observing a value of the sample statistic at least as extreme as the value observed in the data if the null hypothesis is true. Compared with the level of significance to make a decision about the null hypothesis. 
General 
paired t test 
For a pair of matched responses, tests that they have the same mean, using Student’s t distribution. 
Analyze > Matched Pairs > Reference Frame 
parallelism plot 
An overlay plot that reflects the average measurement values for each part. 
Analyze > Quality and Process > Measurement Systems Analysis > Parallelism Plots 
parametric survival models 
Fits a regression model to the parameters of a life distribution, such as Weibull. 
Analyze > Reliabilty and Survival > Fit Parametric Survival 
Pareto chart  general 
A bar chart from highest to lowest that also shows the cumulative total. 
Analyze > Quality and Process > Pareto Plot 
Pareto Plot (effects in a model) 
In fitting models to screening designs, shows how the absolute effect sizes add up to the total, in decreasing order from the largest. Fit Model is the platform. Distribution of Effects is the option. Bayes plot is a suboption. 
Analyze > Quality and Process > Pareto Plot 
partial correlation  adjusted 
The correlations between two variables after they have become residuals from being fit to all the other variables in the set. 
Analyze > Multivariate Methods > Multivariate > Partial Correlation 
partial correlation  group 
The correlations of a group of variables after they have become residuals from being to fit to another set of variables. 
Anayze > Fit Model > Personality:MANOVA > Partial Correlation 
partial least squares 
Partial Least Squares. Predicting Y’s with many X’s, especially when there are more X’s than rows. 
Analyze > Multivariate Methods > Partial Least Square 
partial plot (leverage plot) 
A plot such that the distance from a point to the sloped line is the residual, and the distance to the horizontal line is what the residual would be under the hypothesis. 
Analyze > Fit Model > Personality:Standard Least Square > Leverage Plot 
partition 
Recursively partition the data to predict a response. Classification and regression trees. 
Analyze > Modeling > Partition 
path diagrams 
Available via the structural equation modeling (SEM) SAS addin. 
General 
PCA 
Finds the rotation of the variables that orders the variation from largest to smallest. 
Analyze > Multivariate Methods > Multivariate > Principal Components 
PDF plot 
A plot of the probability density function. 
Analyze > Distribution > Fit Continuous > Normal 
Pearson ChiSquare 
The ChiSquare statistic in frequency tables for categorical variables which is formed by a weighted sum of squares between observed and expected counts. 
Analyze > Fit Y by X > Contingency > Tests > Pearson 
penalized regression 
Penalized regression methods introduce bias into the estimation of b to reduce variability in the estimates. 
Analyze > Fit Model > Generalized Regression 
percentiles 
The value of a variable below which a certain percent of observations fall. 
Analyze > Distribution > Quantiles > Percentiles 
phase chart 
For each phase level, a new σ, set of limits, zones, and resulting tests are done for the control chart. 
Analyze > Quality and Process > Control Chart Builder > Add Phase 
pie chart 
Use the Chart Platform. 
Graph > Chart > Options:Pie Chart 
pie chart (Graph Builder) 
Shows portions of a whole. 
Graph > Graph Builder > Pie 
piecewise Weibull change point detection 
A method to detect a time point where the reliability model changes. 
Analyze > Reliability and Survival > Reliability Growth > Fit Model > Piecewise Weibull NHPP Change Point Detection 
Pillai’s Trace 
Four multivariate tests are supported in the MANOVA personality of the Fit Model platform. Wilks’ Lambda, Pillai’s Trace, HotellingLawley Trace, and Roy’s Maximum Root Criterion. 
Analyze > Fit Model > Personality:MANOVA > Choose Response:Identity > Identity > Whole Model > Pillai’s Trace 
pivot table (Tabulate) 
Drag and drop variables to create a table of counts and statistics. 
Tables > Tabulate 
PlackettBurman designs 
A type of experimental design for screening. 
DOE > Screening Design > Design List > PlackettBurman 
plot  scatterplot 
In JMP, there are two primary scatterplot platforms: Bivariate [which supports fitted lines], and OverlayPlot [which supports multiple Y variables]. 
Analyse > Fit Y by X > Bivariate > Show Points 
PLS 
Partial Least Squares. Predicting Y’s with many X’s, especially when there are more X’s than rows. 
Analyze > Multivariate Methods > Partial Least Square 
points (Graph Builder) 
Shows data values. 
Graph > Graph Builder > Points 
Poisson 
Distribution, probability, quantile. See Scripting Index in JMP for examples. 
Cols > Formula or for a discrete column Distribution > Discrete Fit > Poisson 
Poisson regression model 
Poisson regressions are useful for data sets with counts as the response. They are fit using JMP’s Fit Model platform with the Generalized Linear Model personality. 
Analyzed > Fit Model > Personality:Generalized Linear Model 
polynomial regression 
Fitting a response to a polynomial of a specified order in one term. 
Analyze > Fit Y by X > Fit Polynomial 
pooled variance, sum of squares 
Average of several group variances, where each group variance is weighted for the size of the group. 
General 
post hoc tests 
Tests that compare group means in a oneway ANOVA. See least significant difference (LSD), Hsu’s MCB test, TukeyKramer HSD test, and Dunnett’s test for treatments vs. control. 
General 
power 
The probability of getting a significant result. The probability of rejecting the NULL hypothesis. For prospective power, use the Sample Size platform. For retrospective power, fit a model and look for Power Details. 
DOE > Sample Size and Power 
power for K sample means 
The probability of concluding that there are differences among k means when there truly are differences among the k means. 
DOE > Sample Size and Power > k Sample Means 
power for one sample mean 
The probability of concluding that the mean is different from a hypothesized (null) value when the mean truly is different from the hypothesized value. 
DOE > Sample Size and Power > One Sample Mean 
power for one sample proportion 
The probability of concluding that the proportion is different from a hypothesized (null) value when the proportion truly is different from the hypothesized value. 
DOE > Sample Size and Power > One Sample Proportion 
power for one sample standard deviation 
The probability of concluding that the standard deviation is different from a hypothesized (null) value when the standard deviation truly is different from the hypothesized value. 
DOE > Sample Size and Power > One Sample Standard Deviation 
power for two sample means 
The probability of concluding that two means are different when the two means truly are different from each other. 
DOE > Sample Size and Power > Two Sample Means 
power for two sample proportion 
The probability of concluding that two proportions are different when the two proportions truly are different from each other. 
DOE > Sample Size and Power > Two Sample Proportions 
PP 
An indicator of Process Performance. 
Analyze > Distribution > Capabilty Analysis > PP 
PPK 
Adjustment of PP for the effect of noncentered distribution. 
Analyze > Distribution > Capabilty Analysis > CPK 
PPL 
Adjustment of PP for a lower limit only. 
Analyze > Distribution > Capabilty Analysis > CPL 
PPM 
Adjustment of PP around a target. 
Analyze > Distribution > Capabilty Analysis > CPM 
PPU 
Adjustment of PP for an upper limit only. 
Analyze > Distribution > Capabilty Analysis > CPU 
prediction interval 
Interval that, with a specified degree of confidence, contains either a single observation, or the mean and standard deviation of the next randomly selected sample. 
Analyze > Distribution > Prediction Interval 
PRESS RMSE 
Residual sum of squares where the residual for each row is computed after dropping that row from the computations. 
Analyze > FIt Model > Personality:Standard Least Square > Row Diagnostics > Press > Press RMSE 
presummarized chart 
A plot showing presummarized sample means. 
Analyze > Quality and Process > Control Chart > Presummarize 
prewhitening 
Finds an adequate model for the input series, applies the model to the output, and gets residuals from both series. 
Analyze > Modeling > Time Series > Input Time Series Panel > Input Series > Prewhitening 
principal components 
Finds the rotation of the variables that orders the variation from largest to smallest. 
Analyze > Multivariate Methods > Multivariate > Principal Component 
prior communality estimates: SMC 
In factor analysis, for each variable, a prior estimate of the proportion of the variance of that variable that is explained by the common factors. Obtained using the squared multiple correlation (SMC) for the regression of that variable on the others. 
Analyze > Consumer Research > Factor Analysis > Common Factor Analysis 
probability density function plot 
A plot of the probability density function. 
Analyze > Distribution > Fit Continuous > Normal 
probability Ioptimal design 
A design that minimizes the prediction variance when predicting the failure probability for the times given in Diagnostic Choices. 
DOE > Custom Design 
probability models or distributions 
Available via the Formula Editor. 
Cols > Formula 
probable error 
Median error for a single measurement. 
Analyze > Quality and Process > Measurement Systems Analysis > EMP Results > Probablility Warning 
probit model 
Models the probability that a categorical response is a certain level as a function of regressors. Either use Logistic regression, which is similar, or you can look up how to use Nonlinear. 
Analyze > Nonlinear 
Process Capability Index 
Estimates what the process is capable of producing, considering that the process mean might not be centered between the specification limits. 
Analyze&; Quality and Process > Capability&; Capability Indices Report 
productlimit (KaplanMeier) Survival Estimates 
An stepfunction estimate for the univariate survival distribution function. Also called Product limit estimates. 
Analyze > Reliability and Survival > Survival > Plot Options > Show Kaplan Meier 
profilelikelihood confidence intervals 
Confidence limits for parameters corresponding to changes in the likelihood function. These are produced by various nonlinear models, like logistic regression. 
Analyze > Fit Model > Personality:Nominal Logistic > Confidence Intervals 
profit matrix 
A specification of weights to categorical outcomes. This specification can then be used to weight the outcomes of a prediction model to obtain a decision model. Can be specified as a column property. 
Analyze > Modeling > Partition > OK > Specify Profit Matrix 
proportion 
Estimating, confidence intervals, testing. 
General 
proportional hazards (Cox) Model 
A semiparametric model to fit survival times. 
Analyze > Reliability and Survival > Fit Proportional Hazard 
pure error 
The error variance estimate for exact replicates is called pure error because it is independent of whether the model is right or wrong. It represents the best that can be done in fitting these terms to the model for this data. 
Analyze > Fit Model > Personality:Standard Least Square > Lack of Fit > Pure Error 
Q
Term  Definition  Example of how to access in JMP 

QQ plot see normality: normal quantile plot 
A plot of a batch of values versus normal quantiles calculated on ranks. If the data is normal, the points tend to be in a straight line. 
Analyze > Distribution > Normal Quantile Plot 
quadratic discriminant analysis 
Classifying points using discriminant analysis with a different covariance matrix for each group. See ‘discriminant analysis’. 
Analyze > Multivariate Methods > Discriminant > Discriminant Method:Quadratic, Different Covariances 
quadratic discriminant analysis 
Uses a separate covariance matrix for each group. 
Analyze > Multivariate Methods > Discriminant > Discriminant Method:Quadratic, Different Covariances 
quality 
A measure of how well the Multiple Correspondence Analysis solution represents the level of a variable. 
Analyze > Consumer Research > Multiple Correspondence Analysis 
quality control 
See "Quality and Reliability Methods" documentation book. 
General 
quantile box plot 
A box plot that shows the quantiles that would be equally spaced if the data were normally distributed. 
Analyze > Distribution > Quantile Box Plot 
quantile: normal quantile plot, across groups 
A plot of a batch of values versus normal quantiles calculated on ranks. If the data is normal, the points tend to be in a straight line. 
Analyze > Fit Y by X > Oneway > Normal Quantile Plot 
quantile: normal quantile plot, single population 
A plot of a batch of values versus normal quantiles calculated on ranks. If the data is normal, the points tend to be in a straight line. 
Analyze > Distribution > Normal Quantile Plot 
quantiles 
The value of a variable below which a certain percent of observations fall. 
Analyze > Distribution > Display Options > Quantiles 
quartiles 
Three values that cut a distribution of values into four equal groups. Also in Distribution under Quantiles 
Analyze > Fit Y by X: Oneway > Quantiles 
R
Term  Definition  Example of how to access in JMP 

R chart 
A plot showing a sequence of samples to detect an outofcontrol situation in statistical process control. 
Analyze > Quality and Process > Control Chart > XBar > R 
R^{2} 
A measure of degree of fit, ranging from 0 (no fit) to 1 (exact fit). R^{2} is reported in most regression fit reports. 
Analyze > Fit Model > Personality:Standard Least Square > Summary of Fit > RSquare 
R^{2}, adjusted 
A measure of degree of fit that has been adjusted to reflect the number of parameters in the model. Unlike the unadjusted R^{2}, the adjusted R^{2} does not always increase as more terms are added to the model. 
Analyze > Fit Model > Personality:Standard Least Square > Summary of Fit > RSquare Adj 
R^{2}, entropy 
Measure of fit that compares the loglikelihoods from the fitted model and the constant probability model. 
Analyze > Fit Model > Personality:Nominal Logistics > Whole Model Test > Entropy R^{2} 
R^{2}, generalized 
Generalization of the R^{2} measure of fit that simplifies to the regular R^{2} measure for continuous normal responses. It is similar to the entropy R^{2} measure, but instead of using the loglikelihood, it uses the 2/n root of the likelihood. 
Analyze > Fit Model > Personality:Nominal Logistic > Whole Model Test > Generalized RSquare 
R^{2}, maximum 
The maximum attainable value of R^{2} for the given data if you had to fit a parameter for each unique combination of the regressors. Maximum R^{2} is included in the lackoffit report in many regression reports. 
Analyze > Fit Model > Personality:Nominal Logistic > Lack of Fit > Rsquare Max 
random effect 
A discrete variable is considered random if the levels are randomly selected from a larger population. 
Analyze > Fit Model > Construct Model Effects > Attributes > Random Effect 
random effects 
Random effects are effects, like subjects, where the levels are randomly selected from a larger population, and their effect on the response can be assumed to vary normally with some variance (the variance component). In Fit Model there are two methods of estimating mixed models. 
General 
Random Forest^{tm} 
Creates many trees and computes the final predicted value by averaging the predicted values. 
Analyze > Modeling > Partition > Method:Bootstrap Forest 
range 
Difference between the maximum and minimum observation. 
Analyze > Distribution > Summary Statistics > Customize Summary Statistics > Range 
range chart 
A plot of the variability statistic for each combination of the part and X variables. 
Analyze > Quality and Process > Measurement Systems Analysis > Range Chart 
range odds ratios 
Calculates the change in the ratio of probabilities across the entire range of the continuous independent variable. 
Analyze > Fit Model > Personality:Nominal Logistic > Odds Ratio 
range risk ratio 
Shows the risk change over the whole range of the regressor in a proportional hazards model. 
Analyze > Reliability and Survivall > Fit Proportional Hazard > Risk Ratios 
rare events chart 
Control charts used to determine whether rare events are occurring more frequently. Gcharts plot the number of events between rare events, and Tcharts plot the time between rare events. 
Analyze > Quality and Process > Control Chart Builder > Control Chart Types 
recoding 
Discrete values in a column can be recoded to recategorize or make corrections to the existing values. 
Cols > Recode 
recurrence analysis 
Analyzes how a recurring event is distributed over time, per system, or until the system goes out of service. 
Analyze > Reliability and Survival > Recurrence Analysis 
recursive Partitioning 
Recursively partition the data to predict a response. Classification and regression trees. 
Analyze > Modeling > Partition 
regression diagnostics 
Analysis of residuals, influence of outliers, leverage and multicollinearity. 
General 
regression tree 
A decision tree for a continuous response variable. 
Analyze > Modeling > Partition > Display Options > Show Tree 
regression: logistic 
Fitting the probability of a categorical response to a linear model. 
Analyze > Fit Model > Personality:Nominal/Ordinal Logistic 
regression: multiple regression 
When a response is predicted by a linear combination of several factors. 
Analyze > Fit Model 
regression: one regressor 
Fitting a line through a set of points using the least squares criterion. 
Analyze > Fit Y by X > Fit Line 
regression: polynomial 
Fitting a response to a polynomial of a specified order in one term. 
Analyze > Fit Model > Macros > Polynomial to Degree 
regularized discriminant analysis 
Classifying points using discriminant analysis with a compromise between linear and quadratic approaches. See ‘discriminant analysis’. 
Analyze > Multivariate Methods > Discriminant > Discriminant Method:Regularized, Compromise Method 
regularized discriminant analysis 
Analysis that is a compromise between the linear and quadratic methods. Regularized discriminant analysis allows for specification of two parameters, lambda and gamma. Lambda specifies how to mix the individual and group covariance matrices. Gamma specifies whether to deflate the nondiagonal elements. 
Analyze > Multivariate Methods > Discriminant > Discriminant Method:Regularized, Compromise Method 
relative risk 
A measure of the relative likelihood of an event occurring between two distinct groups (in the context of a 2x2 contingency table). 
Analyze > Fit Y by X > Contingency > Relative Risk 
relative risk ratio 
A measure of the relative likelihood of an event occurring between two distinct groups (in the context of a 2x2 contingency tables). 
Analyze > Fit Y by X > Contingency > Relative Risk 
reliability  failure time 
Reliability models the failure times, in the same way that survival analysis does. For simple univariate models, this is done with the Survival platform. For models with more regressors, Fit Model is used with one of the Survival personalities. 
Analyze > Reliability and Survival > Reliability Growth 
reliability block diagram 
A diagram of a complex system, constructed from simple components. Distributions assigned to the components allow you to analyze the reliability of the entire system. 
Analyze > Realiability and Survival Methods > Reliability Block Diagram 
reliability demonstration 
A reliability demonstration consists of testing a specified number of units for a specified period of time. If fewer than k units fail, you pass the demonstration, and conclude that the product reliability meets or exceeds a reliability standard. 
DOE > Sample Size and Power > Reliability Demonstration 
reliability forecast 
Analysis that predicts future product failures based on previous production and failure counts. 
Analyze > Reliability and Survival > Reliability Forecast 
reliability growth model 
Model that allows for tracking reliability of a process during development testing to determine whether changes to the process are improving process reliability over time. 
Analyze > Reliability and Survival > Reliability Growth > Fit Model > Crow AMSAA 
reliability test plan 
Determines the sample size or the length of study needed to obtain a certain precision about a fitted quantile or probability. 
DOE > Sample Size and Power > Reliability Test Plan 
remaining life BCI 
A measure of the impact of a component to system reliability over time, given that the system has survived a specified amount of time. A large BCI indicates that the system is sensitive to the component. 
Analyze > Realiability and Survival Methods > Reliability Block Diagram > Show BCI 
repeated measures  multivariate 
Repeated measures can be modeled as a multivariate model. Each measurement on a subject is a separate column, and a response for the model. 
Analyze > Fit Model > Personality:MANOVA > Choose Response:Repeated Measures 
repeated measures  univariate sphericityadjusted 
Multivariate problems can be made univariate if the covariances have ‘sphericity’. The MANOVA of personality of ‘Fit Model’ with the ‘Univariate’ option tests sphericity (Mauchly Criterion) and calculates using two adjustments to degrees of freedom: GreenhouseGeisser and HuynhFeldt. 
Analyze > Fit Model > Personality:MANOVA > Univariate Tests Also > Choose Response:Repeated Measures > Within Subjects > Sphericity Test 
repeated measuresunivariate (mixedmodels) 
Repeated measures can be modeled in a mixed model with a random effect for subject. Each measurement is a row in the table. 
Analyze > Fit Model > Personality:MANOVA > Univariate Tests Also > Choose Response:Repeated Measures > Within Subjects > Sphericity Test 
replicates 
Fully repeated set of test conditions. 
DOE > Custom Design > Design Generation > Number of Replicate Runs 
residual analysis 
The use of residuals (including residual plots) to evaluate linear regression models. Different types of plots of the residuals from a fitted model provide information about the adequacy of different aspects of the model. 
Analyze > Fit Model > Personality:Standard Least Square > Row Diagnostics > Plot Residual 
response screening 
Automates the process of conducting tests across a large number of responses. Provides plots of false discovery rate pvalues and tests based on robust estimates. 
Analyze > Specialized Models > Response Screening 
response surface designs 
An experimental designs that allow curvature terms with the idea of finding the optimum value for a response. Central composite and BoxBehnken. 
DOE > Response Surface Design 
response surface methodology 
The fitting and analysis of response surfaces to help determine optimums, or directions to look for the optimum. 
Analyze > Fit Model > Attributes:Response Surface Effect > Personality:Standard Least Square 
ridge regression 
A generalized regression estimation method that applies an L2 penalty in estimating parameters. 
Analyze > Fit Model > Generalized Regression Models > Ridge Regression 
risk ratios (proportional hazard) 
A measure of the relative likelihood of an event occurring between two distinct groups (in the context of a proportional hazards model). 
Analyze > Reliability and Survivall > Fit Proportional Hazard > Risk Ratios 
robust fit (bivariate) 
Fits a line using estimates for the parameters that are less sensitive to outliers than the usual least squares estimates. Uses Huber Mestimation. 
Analyze > Oneway Analysis > Robust Fit 
robust fit (oneway) 
Provides estimates that are resistant to outliers. Conducts the ANOVA test using these statistics. Uses Huber Mestimation. 
Analyze > Oneway Analysis > Robust Fit 
robust mean 
Provides an estimate of the mean that is resistant to outliers. Uses Huber Mestimation. 
Analyze > Distribution > Customize Summary Statistics > Robust Mean 
robust regression 
Fitting models such that the fit is not sensitive to outliers or to a nonnormal distribution. JMP does not offer specific techniques for this, but the documentation for Nonlinear describes techniques for it. 
General 
robust standard deviation 
Provides an estimate of the standard deviation that is resistant to outliers. Uses Huber Mestimation. 
Analyze > Distribution > Customize Summary Statistics > Robust Std Dev 
ROC curve 
The Receiver Operating Characteristic curve is a graphical plot of the sensitivity, or true positive rate, versus false positive rate. 
Analyze > Fit Model > Personality > Ordinal Logistic > ROC Curve 
root means square error (RMSE) 
Standard error of regression. 
General 
rotated factor loading 
In factor analysis, provides loadings for each variable on the rotated factors. A loading reflects the correlation between a variable and the factor. 
Analyze > Consumer Research > Factor Analysis > OK > Rotated Factor Loading 
Roy’s Maximum Root Criterion 
Four multivariate tests are supported in the MANOVA personality of the Fit Model platform. Wilks’ Lambda, Pillai’s Trace, HotellingLawley Trace, and Roy’s Maximum Root Criterion. 
Analyze > Fit Model > Personality:MANOVA > Choose Response:Identity > Identity > Whole Model > Roy’s Max Root 
run chart 
A plot showing a run of the samples. 
Analyze > Quality and Process > Control Chart > Run Chart 
S
Term  Definition  Example of how to access in JMP 

S chart 
Control chart that displays the subgroup standard deviations. 
Analyze > Quality and Process > Control Chart > XBar > S 
sample size 
The Sample Size dialog calculates the sample size needed to achieve a given power for a test. 
DOE > Sample Size and Power 
sampling 
In statistics, sampling refers to the selection of a subset of individuals from a population in order to estimate certain characteristics of the population. 
General 
scatterplot 
In JMP, there are two primary scatterplot platforms: Bivariate [which supports fitted lines], and OverlayPlot [which supports multiple Y variables]. 
Analyse > Fit Y by X > Bivariate > Show Points 
scatterplot 3D 
A threedimensional spinnable view of your data. 
Graph > Scatterplot 3D 
scatterplot matrix 
A grid of scatterplots of all pairings of a set of variables. 
Graph > Scatterplot Matrix 
schematic plot 
Another name for an outlier box plot. 
Analyze > Distribution > Outlier Box Plot 
score plot 
A scatterplot matrix where each cell is a plot of the scores for each variable on a pair of factors. 
Analyze > Consumer Research > Factor Analysis > OK > Score Plot 
score summaries 
Provides a table showing misclassification results for classification based on the scores. 
Analyze > Multivariate Methods > Discriminant Analysis > Score Summaries 
scree plot 
A plot of the eigenvalues versus the number of components in principal components analysis. This plot is useful for visualizing the dimensionality of the data space. 
Analyze > Multivariate Methods > Multivariate > Scree Plot 
screening analysis 
Model data where most effects are assumed to be inactive. The smaller estimates can help estimate the error in the model and determine whether the larger effects are real. 
Analyze > Modeling > Screening 
screening design 
A screening design is used to discover active factors from a large number of potential factors. 
DOE > Screening Design 
screening designs 
Use the DOE platform, either Screening or Custom designs. 
DOE > Screening/Custom Design 
seasonal models 
Seasonal Models capture and estimate the regular pattern of changes that repeats over time (i.e., seasonality) in a time series. 
General 
selforganizing maps 
(SOM) is a type of neural network; the goal is to form clusters on a cluster grid, such that points in clusters that are near each other in the SOM grid are also near each other in multivariate space. 
Analyze > Multivariate Methods > Cluster > Options:KMeans > Method:Self Organizing Maps 
sequential sum of squares 
Sums of squares that depend on the order of the effects in the model. Sequential tests show the reduction in the residual sum of squares as each effect is entered into the model. Also called Type I sums of squares (Type I SS). 
Analyze > Fit Model > Personality:Standard Least Squares > Estimates > Sequential Tests 
ShapiroWilk test for normality 
To test if data is normally distributed. 
Analyze > Distribution > Continuous Fit > Normal > Goodness of Fit > ShapiroWilk W Test 
Shewhart Charts 
The standard type of control chart for statistical quality control. 
Analyze > Quality and Process > Control Chart > XBar > S (Also available through Control Chart Builder) 
signedrank test 
Nonparametric test that a mean is equal to a given value. 
Analyze > Distribution > Test Mean > Wilcoxon Signed Rank 
simple linear regression 
Fitting a line through a set of points using the least squares criterion. 
Analyze > Fit Y by X > Fit Line 
simple moving average 
Time series model based on the unweighted mean of the previous n data points. 
Analyze > Modeling > Time Series > Smoothing Model > Simple Moving Average 
simplex centroid design 
An experimental design for mixtures that takes the vertices and various degrees of means (centroids) of those vertices. 
DOE > Mixture Design > Choose Mixture Design Type > Simplex Centroid 
simplex lattice design 
An experimental design for mixtures that creates a triangular grid of values. 
DOE > Mixture Design > Choose Mixture Design Type > Simplex Lattice 
SIMPLS 
Method for partial least squares analysis that seeks to optimize a statistical criterion. 
Analyze > Multivariate Methods > Partial Least Squares 
Six Sigma 
A business management strategy that seeks to improve the quality of process outputs by identifying and removing the causes of defects (errors) and minimizing variability in manufacturing and business processes. 
General 
size by variable (Graph Builder) 
Scales map shapes according to the size variable, minimizing distortion. 
Graph > Graph Builder > Size 
skewness 
The third moment of a distribution, which helps measure the asymmetry of a distribution. 
Analyze > Distribution > Summary Statistics > Custom Summary Statistics > Skewness 
slope 
The gradient or steepness of a line. See fitting lines. 
General 
smoothed empirical likelihood quantiles 
Quantiles with smoothed empirical likelihood confidence intervals. 
Analyze > Distribution > Display Options > Custom Quantiles 
smoother, smoothing spline (Graph Builder) 
Shows a smooth curve through the data. The smoother is a cubic spline with a lambda of 0.05 and standardized X values. 
Graph > Graph Builder > Smoother 
smoothers  splines 
A fitted line that is gently curved as needed to better fit points. It is optimizing a compromise between a better fit and a smoother line. It is formed by 3^{rd} degree polynomial segments that are spliced together to be continuous and smooth. 
Analyze > Fit Y by X > Fit Spline 
solution path

The solution path is the path that parameter estimates take to solve the estimation problem using elastic net, Lasso, or ridge regression (and their adaptive versions). 
Analyze > Fit Model > Generalized Regression 
Somer’s D 
A nonparametric measure of ordinal association that uses the counts of concordant and discordant pairs. A pair is concordant if an observation with a larger value of X also has a larger value of Y. A pair is discordant if an observation with a larger value of X has a smaller value of Y. Only appropriate when both variables are ordinal. 
Analyze > Fit Y by X > Contingency > Measures of Association > Somer’s D 
sort data tables 
Sort by any number of variables in either ascending or descending order. 
Tables > Sort 
Space filling 
A design that tries to fill space, so that all points in the factor space are not far from a design point. 
DOE > Space Filling Design 
space filling design 
A design constructed so as to minimize bias in estimating models for systems that are deterministic or neardeterministic. The Space Filling Design option provides a number of methods that spread the points over the design space to achieve this goal. 
DOE > Mixture Design > Continue > Space Filling 
Spearman’s rho 
A nonparametric measure of association. Calculated as the correlations of the ranks of the two variables. 
Analyze > Multivariate Methods > Multivariate > Nonparametric Correlations > Spearman’s Rho 
Sphere packing 
A design that tries to fill space such that the largest sphere can be drawn around each point without spheres intersecting. 
DOE > Space Filling Design > Space Filling Design Methods:Sphere Packing 
sphericity test and adjustments 
Multivariate problems can be made univariate if the covariances have ‘sphericity’. The MANOVA personality of ‘Fit Model’ with the ‘Univariate’ option tests sphericity (mauchly Criterion) and calculates using two adjustments to degrees of freedom: GreenhouseGeisser and HuynhFeldt. 
Analyze > Fit Model > Personality:MANOVA > Univariate Tests Also > Choose Response:Repeated Measures > Within Subjects > Sphericity Test 
spinning plot 
A scatterplot that becomes 3dimensional by rotating it in real time. 
Graph > Scatterplot 3D 
splines  smoothing 
A fitted line that is gently curved as needed to better fit points. It is optimizing a compromise between a better fit and a smoother line. It is formed by 3^{rd} degree polynomial segments that are spliced together to be continuous and smooth. 
Analyze > Fit Y by X > Fit Spline 
split data tables 
Split columns to create a shorter, wider table. 
Tables > Split 
Split Plot Design 
An experimental design that has been randomized in layers, with whole plot factors defining the upper layer, usually with hard to change factors, and subplot factors nested to form the lower layer. 
DOE > Custom Design > Design Generation > Hard to change...to change factors 
Split Plot Fitting 
It is important to declare an effect defining the Whole plots with the Random attribute for a correct analysis of split plot designs. 
Analyze > Fit Model 
splitsplit plot design 
Designs in which the VeryHardtochange factors stay fixed within each whole plot. In the middle stratum, the Hardtochange factors stay fixed within each subplot. 
DOE > Custom Design > Design Generation > Hard to change...to change factors 
squared cosines 
Squared cosines indicate the quality of each point for the dimension listed in Multiple Correspondence Analysis. 
General 
stack data tables 
Stack columns to create a long, narrow table. 
Tables > Stack 
standard deviation 
A measure of true spread in a distribution around the mean, or sometimes its estimate. Standard deviation is a pervasive concept, but if you are looking for a simple estimate in a batch of values, the Distribution platform or the Summary command are used. 
Analyze > Distribution > Summary Statistics > Custom Summary Statistics > Std Dev 
standard error 
The standard deviation of a statistic, such as the mean or the difference between two means. 
General 
standardize robustly 
In cluster analysis, standardizes the variables so as to minimize the influence of outliers in distance calculations. This is done by inflating the standard deviation estimate. 
Analyze > Multivariate Methods > Cluster Analysis > Hierarchical > Standardize Robust 
standardized regression coefficients 
Parameter estimates that would result if you standardized your variables before fitting a model. In the Fit Model platform, they are available by unhiding columns in the parameter estimates table, using a contextclick. 
Analyze > Fit Y by X > Bivariate > Parameter Estimates > Columns > Std Beta 
standardizing 
Create a new column and create for it a formula like: (xCol Mean(x))/Col Std Dev(x). Or use Distribution and select ‘Save Standardized’. For grouped data use Oneway and its ‘Save Standardized’ command. 
Analyze > Distribution > Save > Standardized 
statistically inspired modification of PLS (SIMPLS) 
Method for partial least squares analysis that seeks to optimize a statistical criterion. 
Analyze > Multivariate Methods > Partial Least Squares 
Steel with control test 
Performs the SteelDwass test on each pair. 
Analyze > Fit Y by X > Oneway > Nonparametric > Nonparametric Multiple Comparisons > Steel with Control 
SteelDwass all pairs test 
Performs the SteelDwass test on each pair. This is the nonparametric version of the All Pairs, Tukey HSD option found on the Compare Means menu. 
Analyze > Fit Y by X > Oneway > Nonparametric > Nonparametric Multiple Comparisons > SteelDwass All Pairs 
stem and leaf plot 
The stem and leaf plot is similar to a histogram, but the bars are consist of digits representing the later digits of the values. 
Analyze > Distribution > Stem Leaf Plot 
step plots 
A pointtopoint plot where the connecting lines are drawn as a horizontal and vertical step, rather than as a direct line. 
Graph > Overlay Plot > Y Options > Step 
stepwise regression 
A regression approach that involves incrementally adding or deleting terms to a regression model. 
Analyze > Fit Model > Personality:Stepwise 
strip plot design 
An experimental design involving Hard and Very Hard to change factors. Hard to change factors can vary independently of Very Hard to change factors. 
DOE > Custom Design > Design Generation > Hard to change...to change factors 
Stuart’s TauC 
A nonparametric measure of ordinal association that uses the counts of concordant and discordant pairs. A pair is concordant if an observation with a larger value of X also has a larger value of Y. A pair is discordant if an observation with a larger value of X has a smaller value of Y. Only appropriate when both variables are ordinal. 
Analyze > Fit Y by X > Contingency > Measures of Association > Stuart’s TauC 
Student’s t 
The distribution that results when a normally distributed estimator with [hypothesized] mean zero is divided by an independent estimator of its standard error. Test statistics are formed that measure significance by how improbably large a value the estimate is if the true mean were zero. 
Analyze > Fit Model > Personality:Standard Least Square > LSMeans Student’s t 
Student’s t: Each Pair 
Individual pairwise comparisons of group means using Student’s t test. See ‘multiple comparisons’. 
Analyze > Fit Y by X > Oneway > Compare Means > Each Pair, Student’s t 
Student’s t: Paired 
For a pair of matched responses, tests that they have the same means, using Student’s T distribution. 
Analyze > Matched Pairs > Reference Frame 
Student’s t: test parameter in linear model 
In fitting platforms, estimates are routinely reported with their standard error and the tratio, the ratio of their value to the standard error, which has a Student’s t distribution under the hypothesis that the true parameter is zero. 
Analyze > Distribution > Parameter Estimates 
Student’s t: test that mean=value 
A test that the mean is some hypothesized value. 
Analyze > Distribution > Test Mean 
Student’s t: two groups equal variance 
The standard test that the means between two groups are equal. It is assumed that the error variance is the same in the two groups. This is a special case of the F test in oneway anova. 
Analyze > Fit Y by X > Oneway > Compare Means > Each Pair, Student’s t 
Student’s t: two groups unequal variance 
A Student’s t test that has been weighted to adjust for different variances in the two groups. 
Analyze > Fit Y by X > Oneway > Compare Means > Each Pair, Student’s t 
subset data tables 
Subset selected rows and columns. 
Tables > Subset 
summarize data tables 
Summarize columns from the active tables. 
Tables > Summary 
summary statistics 
Mean, Std Dev, Std Err Mean, Upper Mean Confidence Limits, Lower Mean Confidence Limits, N, Sum Weight, Sum, Variance, Skewness, Kurtosis, CV, N Missing, N Zero, N Unique, Uncorrected SS, Corrected SS, Autocorrelation, Median, Mode, Trimmed Mean, Geometric Mean, Range, Interquartile Range, Median Absolute Deviation. 
Distribution >Summary Statistics > Customize Summary Statistics 
surveys 
See documentation for categorical analysis. 
Analyze > Modeling > Categorical 
survival  parametric survival models 
Fits a regression model to the parameters of a life distribution, such as Weibull. 
Analyze > Reliability and Survival > Fit Parametric Survival 
survival analysis 
Models the distribution of ‘failure’ event times. For simple univariate models, this is done with the Survival platform. For more models with regressors, Fit Model is used with one of the Survival personalities. 
Analyze > Reliability and Survival > Survival 
survival estimates  product limit (KaplanMeier) 
An stepfunction estimate for the univariate survival distribution function. Also called Product limit estimates. 
Analyze > Reliability and Survival > Survival > Plot Options > Show Kaplan Meier 
T
Term  Definition  Example of how to access in JMP 

T chart 
A control chart for rare events to determine whether rare events are occurring more frequently than expected by graphing time between events. 
Analyze > Quality and Process > Control Chart Builder > Sigma > Weibull 
t distribution (Student’s t distribution) 
A family of continuous probability distributions defined by the degrees of freedom, used to estimate population parameters when the sample size is small and/or when the population variance is unknown. 
General 
t test: See Student’s t 
The distribution that results when a normally distributed estimator with [hypothesized] mean zero is divided by an independent estimator of its standard error. Test statistics are formed that measure significance by how improbably large a value the estimate is if the true mean were zero. 
Analyze > Fit Y by X > Oneway > Compare Means > Each Pair, Student’s t 
T^{2} statistic 
A measure of multivariate distance that takes into account the variances and covariances. Equal to the squared Mahalanobis distance. 
Analyze > Multivariate Methods > Multivariate > Outlier Analysis > T^{2} 
Taguchi robust parameter design 
A type experimental design that has an inner design for control factors, and an outer design over noise factors, with a response calculated from a signaltonoise statistic over the noise factors. 
DOE > Taguchi Arrays 
ternary plot 
A triangular plot showing how three factors that add up. Usually they are ingredients in a mixture, and they add up to 1. Because each factor is a function of the other two, it can be represented in two dimensions. If there are more than three factors two are identified and the others are grouped into ‘other factors’. 
Graph > Ternary Plot 
test of proportions 
Test if proportions are different than hypothesized values. In the twosided case, this test is a chisquare test; in either of the onesided cases, this test is an exact onesided binomial test. 
Analyze > Distribution > Test of Proportions 
testretest error 
Used to check for detectable differences between the different levels of the potential nuisance component. 
Analyze > Quality and Process > Measurement Systems Analysis > TestRetest Error Comparison 
threelevel designs 
Use the DOE platform, either Screening or Custom designs. 
DOE > Screening/Custom Design 
time Ioptimal design 
Design that minimizes the prediction variance when predicting the time to failure for the probability given in Diagnostic Choices. 
DOE > Custom Design 
time series analysis 
Analysis of how values change across equallyspaced time. 
Analyze > Modeling > Time Series 
tobit model 
A model where the response is truncated with a lower bound of zero. This is handled similarly to censoring in survival models. JMP does not have a specific feature, but a Tobit example is shown in the documentation for Nonlinear. 
Analyze > Modeling > Nonlinear 
tolerance interval 
Interval defined as the upper specification limit minus the lower specification limit. 
Analyze > Distribution > Tolerance Interval 
transfer function model 
Model based on the past values of explanatory variables and residuals are modeled as an ARIMA model. 
Analyze > Modeling > Time Series > Transfer Function 
transformations 
In Fit Y by X, fit special for transformations, or use a column formula. 
General 
transpose data tables 
Transpose to interchange rows and columns. 
Tables > Transpose 
treatment 
In experiments, a treatment is a condition applied to the experimental units. 
General 
tree map 
Chart that tiles a rectangle with category rectangles. This is useful when there are a lot of categories and a bar chart of the data creates a long line of bars. 
Graph > Tree Map 
treemap (Graph Builder) 
Shows a response summarized by categories. 
Graph > Graph Builder > Treemap 
trees 
Recursively partition the data to predict a response. Classification and regression trees. 
Analyze > Modeling > Partition 
trimmed mean 
The mean calculated after removing the smallest p% and the largest p% of the data. 
Analyze > Distribution > Summary Statistics > Custom Summary Statistics > Trimmed Mean 
Tukey meandifference plot 
A plot for matched pairs analysis which shows the relationship between the differences versus the means of the paired observations. 
Analyze > Matched Pairs > Plot Dif by Mean 
TukeyKramer HSD test 
A general test for differences among a set of means that controls the significance level for multiple comparisons. See ‘multiple comparisons’. 
Analyze > Fit Y by X > Oneway > Compare Means > All Pairs, Tukey HSD 
two sample proportion test 
Test if proportions between two samples are different. You can test it by Pearson’s or likelihood ratio chisquared test. 
Analyze > Fit Y by X > Contingency >Tests 
two sample ttest 
Used to determine whether two population means are equal. 
Analyze > Fit Y by X > Oneway > Means/ANOVA/Pooled t 
twolevel designs 
Use the DOE platform, either Screening or Custom designs. 
DOE > Screening/Custom Design 
twoway ANOVA 
Examines the influence of two different categorical independent variables on one dependent variable. 
Analyze > Fit Model > Personality:Standard Least Square > Analysis of Variance 
type I and type II errors 
Type I error (false positive): The null hypothesis is rejected when it is true. Type II error (false negative): The null hypothesis is not rejected when it is false. 
General 
types of sums of squares 
JMP produces results similar to SAS GLM’s Types III and IV. JMP’s SS agree with GLM when Types III and IV agree themselves, which occurs in all designs without missing cells. 
General 
U
Term  Definition  Example of how to access in JMP 

U chart 
A plot showing the numbers of nonconformities per inspection unit in subgroup samples. 
Analyze > Quality and Process > Control Chart > U 
uncorrected sum of squares 
Total sum of squares, uncorrected for the mean. 
Analyze > Distribution > Summary Statistics > Custom Summary Statistics > Uncorrected SS 
uniform precision see response surface designs 
Designs whose variance is relatively uniform. See DOE Platform for Response Surface. 
DOE > Response Surface Design > Choose a Design > CCDUniform Precision 
Uniform Space filling Design 
A design that tries to fill space such that the multivariate empirical cumulative distribution function is closest to the uniform distribution. 
DOE > Space Filling Design > Space Filling Design Methods:Uniform 
Uniformly weighted moving average (UWMA) chart 
A plot showing a uniformly weighted moving average chart. 
Analyze > Quality and Process > Control Chart > UWMA 
unit odds ratios 
Calculates the change in the ratio of probabilities as the continuous independent variable changes by 1 unit. 
Analyze > Fit Model > Personality:Nominal Logistic > Odds Ratio 
unit risk ratio 
Shows the risk change over one unit of the regressor in a proportional hazards model. 
Analyze > Reliability and Survivall > Fit Proportional Hazard > Risk Ratios 
univariate distribution 
Use Distribution Platform. 
Analyze > Distribution 
univariate repeated measures 
Set up a new column and add a formula to it. Some fitting platforms provide features for doing this without needing a new column. 
Multiple 
univariate repeated measures 
See repeated measures topics. 
Analyze > Fit Model > Personality:MANOVA > Univariate Tests Also > Choose Response:Repeated Measures > Within Subjects > Sphericity Test 
update data tables 
Update one table with values from another table. 
Tables > Update 
uplift 
A partition model which helps identify characteristics of individuals who are likely to respond to an intervention or treatment. 
Analyze > Consumer Research > Uplift Model Report 
upper and lower control limit 
See control charts. 
General 
V
Term  Definition  Example of how to access in JMP 

validation column 
Validation column role in many modeling platforms. Used to provide honest assessment of model performance by splitting data into training, validation and test sets. 
Use validation column in many modeling platforms 
Van der Waerden Test 
A test that compares several distributions by ranking the data, using the ranks to form normal scores and comparing the mean scores across groups. 
Analyze > Fit Y by X > Oneway > Van Der Waerden Test 
variability analysis 
Variability Chart shows variation across groups, and can fit variance components. 
Analyze > Quality and Process > Variability/Attribute Gauge Chart 
variable importance 
A method for assessing the importance of variables that is independent of model type. Only available for continuous responses. 
Predicition Profiler > Assess Variable Importance 
variable importance for projection (VIP) 
A VIP score is a measure of a variable’s importance in modeling both X and Y. 
Analyze > Multivariate Methods > Partial Least Squares > (red triangle) > Variable Importance Plot 
variables clustering (principal component analysis) 
Variable clustering option for predictor variable reduction prior to modeling. 
Analyze > Multivariate Methods > Principal Components > Cluster Variables 
variance 
A measure of how far a set of numbers is spread out. 
Analyze > Distribution > Summary Statistics > Custom Summary Statistics > Variance 
variance components 
Random effects are effects, like subjects, where the levels are randomly selected from a larger population, and their effect on the response can be assumed to vary normally with some variance (the variance component). In Fit Model there are two methods of estimating mixed models. 
ANalyze >Quality and Process >Variability/Attribute Gauge Chart 
variance homogeneity 
Testing that the variances are equal in a oneway layout and providing a weighted (Welch) Anova in case they are not. 
Analyze > Fit Y by X > Oneway > Unequal Variances 
variance inflation factors (VIF) 
In regression where the regressors are highly correlated, a measure of interest is how much inflated the variance of the estimator compared with what its variance would be without the effect of the other regressors. In Fit Model, the VIF is available by contextclicking in the Parameter Estimates report table. 
Analyze > Fit Model > Personality:Standard Least Square > Parameter Estimates > Right Click:Columns > VIF 
violin plot (Graph Builder) 
Shows regions of density or value contours. If you specify only one continuous variable for X or Y, a violin plot appears instead of a contour plot. 
Graph > Graph Builder > Contour 
VIP 
A VIP score is a measure of a variable’s importance in modeling both X and Y. 
Analyze > Multivariate Methods > Partial Least Squares > (red triangle) > Variable Importance Plot 
W
Wald Test 
A ChiSquare test that is a linear approximation to a likelihood ratio ChiSquare. It is cheaper but considered less reliable than the likelihood ratio ChiSquare. The Fit Model platform for nominal/ordinal responses calculates Wald tests automatically, but waits for a request to do the more consuming likelihood ratio ChiSquares. 
Analyze > Fit Model > Personality:Ordinal Logistic > Wald Tests 
Weibayes analysis 
A reliability analysis that can be used when there are no failures in the observed data. 
Analyze > Reliability and Survival > Life Distribution > Weibull 
Weibull Plot  survival 
A Weibull plot for failure times puts the log(log(Survival probability)) on the y axis, and the log(time) on the x axis. If the events are Weibulldistributed, the points tend to follow a straight line. 
Analyze > Reliability and Survival > Survival > Weibull Plot 
Weibull survival model 
Weibull is the most common probability distribution used to model failure time responses, in survival and reliability studies. Weibull models are supported in the Survival platform for univariate studies, and in the parametric survival personality of Fit Model. 
Analyze > Reliability and Survival > Survival > Weibull Fit 
weighting 
Weighting is a statistical method that adjusts your sample data in the estimation of population parameters. 
General 
Welch Anova 
A oneway analysis of variance that is weighted according to the different variances estimated from the data. The 2group Student’s t test for unequal variances is a special case of this. 
Analyze > Fit Y by X > Oneway > Unequal Variances > Welch’s Test 
Western Electric rules 
Tests for special causes. 
Analyze > Quality and Process > Control Chart > XBar > XBar/R > Tests > All Tests 
Westgard rules 
Tests for special causes that follow the Westgard Rules. 
Analyze > Quality and Process > Control Chart > Levey Jennings > Westgard Rules 
Wilcoxon each pair test 
Performs the Wilcoxon test (rank test for errors with logistic distributions) on each pair, and does not control for the overall α level. 
Analyze > Fit Y by X > Oneway > Nonparametric > Nonparametric Multiple Comparisons > Wilcoxon Each Pair Test 
Wilcoxon rank sum test 
A nonparametric statistical hypothesis test for assessing whether one of two samples of independent observations tends to have larger values than the other. Also called the MannWhitney U test. When more than two groups called a KruskalWallis test. 
Analyze > Fit Y by X > Nonparametric > Wilcoxon Test 
Wilcoxon signedranks test 
Nonparametric test that a mean is equal to a given value. 
Analyze > Distribution > Test Mean > Wilcoxon Signed Rank 
Wilcoxon test (Survival) 
For univariate survival models across several groups, this is one of the tests that the survival distribution is the same across groups. 
Analyze > Reliability and Survival > Survival > Tests Between Groups > Wilcoxon 
Wilcoxon two group test 
A test that compares several distributions by ranking the data and comparing the ranks from each group. 
Analyze > Fit Y by X > Oneway > Nonparametric > Wilcoxon Test 
Wilks’ Lambda 
Four multivariate tests are supported in the MANOVA personality of the Fit Model platform. Wilks’ Lambda, Pillai’s Trace, HotellingLawly Trace, and Roy’s Maximum Root Criterion. 
Analyze > Fit Model > Personality:MANOVA > Choose Response:Identity > Identity > Whole Model > Wilks’ Lambda 
willingness to pay 
How much a price must change allowing for the new feature settings to produce the same predicted outcome. 
Analyze > Consumer Research > Choice Models > Willingness to Pay 
Winter’s Method 
Fitting a seasonal moving average process for forecasting a time series. 
Analyze > Modeling > Time Series > Smoothing Model > Winters Method 
wrap (Graph Builder) 
Subsets or partitions the data based on the variable or variables that you select. Wraps the data horizontally and vertically. Once a variable is placed here, no variable can be placed in Group X. 
Graph > Graph Builder > Wrap 
X
Term  Definition  Example of how to access in JMP 

X,Y (Graph Builder) 
Drop variables here to assign them the X or Y role 
Graph > Graph Builder > X/Y 
XBar chart 
A plot showing a sequence of samples to detect an outofcontrol situation in statistical process control. 
Analyze > Quality and Process > Conrol Chart > XBar 
Z
Term  Definition  Example of how to access in JMP 

z test 
A test on means that is appropriate if the standard deviation of the data is known. If estimates are substituted for the standard deviations, then this becomes the Student’s t test. 
Analyze > Distribution > Test Mean > Enter True Standard Deviation to do ztest 
zscore (zvalue, normal score, standard score) 
The distance between the raw score and the population mean in units of the standard deviation. 
Cols > Formula 