Example of the Categorical Platform

This example uses the Consumer Preferences.jmp sample data table, which contains survey data on people’s attitudes and opinions, and some questions concerning oral hygiene (source: Rob Reul, Isometric Solutions).

1.	Select Help > Sample Data Library and open Consumer Preferences.jmp.

2.	Select Analyze > Consumer Research > Categorical.

3.	Select I am working on my career and click Responses on the Simple tab.

4.	Select Age Group and click X, Grouping Category.

Click OK.

6.	Select Crosstab Transposed from the Categorical red triangle menu.

7.	Select Test Response Homogeneity from the Categorical red triangle menu.

Survey Results by Age Group details the responses indicating that a respondent is currently working on his or her career and the age group. Of those responding positively, the highest majority working on their career were in the age group 25-29 at 84.1%. The highest majority of those responding oppositely were in the age group > 54 at 53.5%.

Survey Results by Age Group

Launch the Categorical Platform

Launch the Categorical Platform by selecting Analyze > Consumer Research > Categorical.

Categorical Platform Launch Window

The launch window includes tabs for a variety of response roles (Simple, Related, and Multiple) and a Structured tab where you can create your own structured responses. The following sections describe the different response types and effects.

Response Roles

Use the response roles buttons within the tabs to choose selected columns as responses with specified roles. You can also drag column names to the response list. The response roles are summarized in Response Roles.

Simple Tab

The default tab, Simple, contains a single button, Responses. This is appropriate for all basic analyses that do not have a special structure. You can drag column names from the Select Columns list to the Response list, or you can select columns and then click Responses. If a column has a Multiple Response column property, JMP automatically changes the handling of the column to recognize this property.

Aligned Responses

Performs the analysis like the default analysis, but shows the results more compactly by aligning the analyses side-by-side into one larger table.

Repeated Measures

Indicates that the columns reflect responses made by the same individual at different times, and you are interested in the changes between the times.

Rater Agreement

Is useful when each column is a rating for the same question, but by different individuals (raters) and you want to study how much the raters agree on their responses.

Multiple Tab

The Multiple tab is for multiple responses; when a question can involve checking off more than one choice. There are a variety of ways of storing multiple response data, so there are several buttons in the tab to accommodate the various means:

Multiple Response

Means that you have several columns acting like fill-in-the-blank columns to specify the multiple responses.

Multiple Response by ID

Indicates that you have several rows in a table corresponding to the multiple responses in one column, and the individuals are identified by an ID column.

Multiple Delimited

Signifies that you have one column that has several responses in it separated by a comma.

Indicator Group

Denotes that there is a column for each possible response, and each column is an indicator (for example, it has only two values, like 0 or 1).

Response Frequencies

Also has a column for each possible response, but has frequency counts instead of an indicator.

Free Text

Is used for comment fields where the analysis counts the frequency of each word used. Free Text gives word counts in both word order and frequency order, and the rate of non-empty text. For more information about Free Text, refer to Free Text Report Options.

Structured Tab

The Structured tab enables you to construct complex tables of descriptive statistics by dragging column names into green icon drop zones to create side-by-side and nested results. You can nest a variable within or beside another variable according to the structure that you want for the top and side of the table. Continue to drag columns, either beside or nested within another column, to specify the structure. For more information about structured reports, refer to Structured Report Options.

Structured Tab

To create a structured table:

1.	Drag a column name to the green drop zone at the Top or Side of the table. The drop zone is highlighted in pink.

2.	Add more variables by dragging a column name the appropriate green drop zone. You can also drag a column name between two columns.

To remove a selection, click the selection and then select Undo.

To add a selection back that you just removed, select Redo.

To clear all of the selections, select Clear.

3.	When you are finished creating the table, click Add=> to add the variables to the response list.

To make a revision once you have added your selection to the response list, click the selection and then select <=Edit.

4.	Complete the remainder of the launch window as necessary and click OK.

The Categorical report window appears.

5.	Should you want to make a change to the table, select Relaunch Dialog from the Categorical red triangle menu. The launch window reappears where you can make edits to your selections.

A few guidelines with Structured effects:

•	Structured always assumes that the innermost terms on the Side are responses, and that all other terms are sample level grouping factors.

•	You can analyze multiple response terms in the form of delimited multiple response columns, but you must indicate that it is a multiple response by having a Multiple Response column property. Use the Col Info window to add this property, as needed.

Response Roles
Response Role	Description	Example Data
Simple Tab
Responses	Separate responses are in each column, resulting in a separate analysis for each column.
Related Tab
Aligned Responses	Responses share common categories across columns, resulting in better-organized reports.
Repeated Measures	Aligned responses from an individual across different times or situations.
Rater Agreement	Aligned responses from different raters evaluating the same unit, to study agreement across raters.
Multiple Tab
Multiple Response	Aligned responses, where multiple responses are entered across several columns, but treated as one grouped response.
Multiple Response by ID	Multiple responses across rows that have the same ID values.
Multiple Delimited	Several responses in a single cell, separated by commas.
Indicator Group	Binary responses across columns, like selected or deselected, yes or no, but all in a related group.
Response Frequencies	Columns containing frequency counts for each response level, all in a related group.
Free Text	Counts the frequency of each word used in a comment field.
Structured Tab
	Drag variables to green drop zones to create your own structured table.

Cast Selected Columns into Roles

The lower right panel of the Launch window has the following options:

X, Grouping Category

Defines sample level groups to break the counts into. By default, it tabulates each combination of X values, but uses the Grouping Option below for other combinations.

Sample Size

Defines the number of individual units in the group for which that frequency is applicable to, for multiple response roles with summarized data. For example, a Freq column might indicate 50 defects, where the sample size variable would reflect the defects for a batch of 100 units.

Freq

Specifies the column containing frequency counts for each row for presummarized data.

Only required and used when Multiple Response by ID is selected.

Identifies a variable to produce a separate analysis for each value that appears in the column.

Other Launch Window Options

Several launch options are presented in the lower left panel of the window that can be specified before the analysis. The options can also be selected later from the Categorical red triangle menu, and have the effect of rerunning the platform with the new setting. The default settings for some of the launch options can be changed in the Categorical red triangle menu. For more information, refer to Set Preferences.

Grouping Option

Specifies whether you want to use the X columns individually or in a combination. Use this option only to specify more than one X (Grouping) column, and denote whether you want to treat the Xs one at a time, or in a fully nested grouping, or both. For example, if the X Columns are Region and Age Group, you can get separate tables for Response by Region and Response by Age Group (each individually) or get a nested table with each age group within each region (combinations), or both.

Combinations gives frequency results for combinations of the X variables.

Each Individually gives frequency results for each X variable individually.

Both gives frequency results for combinations of the X variables, and individually.

Unique Occurrences within ID

Allows duplicate response levels within a subject to be counted only once. An ID variable must be specified.

The following options can be specified on the launch window as well as from the Categorical report red triangle menu. They are also available as Preference settings. For more information, refer to Statistical Options and Set Preferences.

Count Missing Responses

Changes the behavior to tabulate missing values as categories, while still excluding them from statistical comparisons. When you have missing values, this specifies whether you want to see them tabulated beside the nonmissing data, or just excluded. Missing values can be either standard (numeric NAN or character empty) or a code declared as missing with the column property Missing Value Codes. Note that if a column contains only missing values, the missing values are counted regardless of the state of this option.

Order Response Levels High to Low

Changes the response order but keeps the X order from low to high. The default ordering is low to high. You can control the ordering with a column property (Value Ordering), but if you always want to see the high values first, then select this option. Often, ordered categories are ratings, and you want to see the positive ratings first. In the red triangle menu, this option is under Category Options.

Shorten Labels

Shortens labels by removing common prefixes and suffixes. Sometimes surveys code a lengthy label that contain a common prefix or suffix. For example, “Occurred 5 to 10 times in the last year” might be a level, but the phrase “in the last year” is repeated for each value label, and you do not need to see it repeated in the report. This option trims the common prefixes and suffixes. It also changes multiple blanks into single blanks. The option only applies to value labels, not column names.

Include Responses Not in Data

Includes a count for values that were not in the data. Sometimes when you conduct a survey and give choices, one of the choices is not selected. If you still want to see the choices that are not in the data, use this option. It determines the missing categories from the value labels in the column.

Supercategories

When ratings are involved in a data set (for example, a five point scale), you might want to know the percent of the responses in the top two or other subset of ratings. Such a group of ratings can be defined in the data through the column property, Supercategories.

The term Supercategories refers to the extra slots in a table to aggregate over groups of categories. The Supercategories property supports four keywords: Group, Mean, Std Dev, and All. Mean and Std Dev calculate statistics for value scores, and All aggregates across all levels. For example, a “Top Two” Group supercategory could aggregate the two top categories in a response, as specified. Although we support Mean, Std Dev, and All, we do not recommend using them because they are available as built-in statistics as well as supercategories. Supercategories appear in Crosstab tables and Frequency Charts. Share Charts, however, do not show supercategories and ignore the Hide option described below.

Create Supercategories by selecting a column and then selecting Column Info > Column Properties > Supercatagories. On the column properties window, select a column, enter a Supercategory Name, and click Add.

The Supercategories Options red triangle menu item provides the following commands for a selected category:

Hide

Omits columns for each category from the report. Only a column for the supercategory is included.

Net

Prevents individuals from being counted twice when they appear in more than one supercategory. Net is available only for a multiple response column.

The following red triangle options are also provided:

Add Mean

Includes mean statistics in the report.

Add Std Dev

Includes standard deviation statistics in the report.

Add All

Includes total responses in the report. By default, the Total Responses column is always included.

Note: Supercategories are supported for all response effects except Repeated Measures and Rater Agreement. Some response effects do not support Mean and Std Dev slots, because the effects do not have a natural score.