In Spss,
click Analysis/Descriptive/Crosstab
- then click at least one variable from the left column into the "Row" box
- and at least one from the left column into the "Column" box
- (sometimes you may want to click a control variable into the "Layers" box at the bottom
- (if you click several for each, all those in the row box will be run by all those in
the column box).
- Before you click "Ok" to get the tables always do the following:
click the [STATISTICS] box and once open, click the two statistics below
(open these pages to read about them)
- You can click on
Chi-Squared (but dont turn in these
results to me, I just want a table with Cramer's V and tau-b) as a test of to what degree the sum of discrepancies of observed cell values (Obs for each cell)
departs from what the expected value E of each cell (row total times col total divided by the grand
total. Chi-squared is the sum ((Obs - E)^2)/(Grand Total) for each cell (^2 means squared).
The df or degree of freedom for the table with NR = number of rows and NC = number of columns
is (NR-1)(NC-1).
- Click on "Phi and Cramer's V" if you have nominal or on "Kendall's tau" if you have ordinal data. Kendall's Tau-b
is for square tables; Kendal's tau-c is for rectangular tables.
- Click on "Continue," taking you back to the "Crosstabs" window.
Click the [CELLS] box at the bottom of the window.
- Click on "Row," or "Column" (but not both) in the "Percentages" box: this should correspond to your INDEPENDENT variable whose effects you will
see in the opposite row or column.
- Click on "Continue," again taking you back to the "Crosstabs" window.
Click on "OK." Your crosstabs table and related statistics will be added to the
output file. Note that if you double click on an element you can edit it or delete it.
Always delete the program script, the "Case Summary" and the label "Cross-Tab." I do not want
to see these in your papers, just the cross-tab table and the statistics.
The "Insert" pull-down menu includes a page break.