Prism organizes data for three-way ANOVA differently than do most other programs. Here is an example of data entry:
•Note that Prism does not use grouping variables. Instead, you only enter the outcome variable and designate treatments or groups by choice of rows and columns.
•Rows 1 and 2 represent two levels of one factor (male vs. female) in the example. You can use more than two rows if this factor has more than three levels.
•Data set columns A and B vs. C and D represent the second factor, Control vs. Treated in the example.
•Data set columns A and C vs. B and D represent the third factor, low vs. high dose in the example.
•Only data in the first four data set columns (A-D) are considered in three-way ANOVA.
•The example shows two subcolumns, for duplicate data. But you can choose any number of subcolumns you want (up to 256).
From the Welcome (or New Data Table and Graph) dialog, choose the Grouped tab.
If you are not ready to enter your own data, choose to use sample data and choose the three-way ANOVA sample data sets.
If you plan to enter your own data, create a table with the right number subcolumns to hold your replicated measurements.
You must enter two or more values for most combinations of conditions (alternatively, enter the mean, sample size and SD or SEM). Prism cannot compute three-way ANOVA with only a single value for each set of conditions.
If you have already averaged your replicates in another program, you can choose to enter and plot the mean and SD (or SEM) and n.
1.From the data table, click on the toolbar.
2.Choose Three-way ANOVA from the list of grouped analyses.
3.On the first tab (RM Design) define your experimental design -- which factors are repeated measures.
4.If you have repeated measures, use the second (RM Analysis) tab to specify whether you want to run repeated measures anova or fit a mixed model.
5. On the third (Factor Names) tab, optionally name the grouping variables that define the rows and columns. For the example shown above, A&B vs C&D could be named "treatment"; A&C vs. B&D could be named "dose", the rows could be named "sex".
6.On the fourth (Multiple Comparisons) tab, select your goal (if any) for multiple comparisons.
7.On the fifth (Options) tab, choose the details of which multiple comparisons test you want.
8.On the sixth (Residuals) tab, choose if and how you want to plot residuals, and if you want to test them for normality and equal variability (homoscedasticity).
Prism cannot run three-way ANOVA with huge data sets and presents an message telling you so. How huge is huge? Details here.