1 Open the Football.jmp sample data table.
 2 Select Analyze > Fit Model.
 3 Select Speed and click Y.
 4 Select Weight and Position2 and click Add.
Notice that Position2 is a nominal variable with values representing football positions.
 5 For Personality, select Stepwise.
 6 Click Run.
Position Hierarchy
 • The method first splits Position into two groups with the term Position2{wr&db&o&lb&te–ki&l}. One group (the slower group) consists of the wide receivers (wr), defensive backs (db), offensive backs (o), linebackers (lb), and tight ends (te). The other group (the faster group) is the kickers (ki), and linemen (l), which ultimately split as Position2{ki-l}.
 • The next split subdivides the slower group into wide receivers (wr), defensive backs (db), and offensive backs(o) versus linebackers (lb) and tight ends (te), shown as Position2{wr&db&o–lb&te}. The linebackers (lb) and tight ends split (te) to form Position2{lb–te}.
 • The slower group divides again giving Position2{wr&db-o}, wide receivers (wr) and defensive backs (db) versus offensive backs (o).
 • The last subdivision is the wide receivers (wr) and defensive backs (db), Position2{wr–db}.
Tree Structure of Terms and Corresponding Current Estimates Table
Using the default Combine rule for terms to enter a model or the Restrict rule, a term cannot enter the model unless all the terms above it in the hierarchy have been entered. Therefore, it is simple to bring in the term Position2{wr&db&o&lb&te-ki&l} because there is nothing above it in the hierarchy. But to enter Position2{lb-te} requires that the two other terms above it in the hierarchy are entered: Position2{wr&db&o-lb&te}, and Position2{wr&db&o&lb&te-ki&l}.
 • The hierarchy leads to natural stepping rules.
 • The groupings that make the greatest initial separation enter the model early.