IBM Support

How do I get all pairwise comparisons among the cells of a factorial design?

Question & Answer


Question

I'm running a factorial ANOVA in SPSS GLM/UNIANOVA and want to follow up by doing all pariwise comparisons among the cells. How do I do this in SPSS?

Answer

The SPSS linear modeling procedures currently do not offer an option to automatically compute pairwise comparisons among the cells of a multi-way design. Simple effects comparisons, including all pairwise for each simple effect, are available via EMMEANS, but for all pairwise comparisons among cells of a multi-way design, you would need to either use LMATRIX to specify the appropriate contrasts, or lay out the design as a one way by creating a new factor variable with a different value for each cell of the original design. Note that the overall model fitted when running a one way ANOVA with this new factor is the same as the full factorial model on the original design.

Beginning with Release 15, you can use the new GENLIN procedure for generalized linear models to obtain all pairwise comparisons in ANOVA models. In the EM Means tab, specify the interaction term and choose the Pairwise option under Contrast. GENLIN uses maximum likelihood to estimate the scale parameter in a normal distribution model, which produces different error estimates than you get in GLM and UNIANOVA. To match the error estimates from those procedures, go to the Estimation tab and specify either Deviance or Pearson chi-square for the Scale Parameter Method

[{"Product":{"code":"SSLVMB","label":"IBM SPSS Statistics"},"Business Unit":{"code":"BU059","label":"IBM Software w\/o TPS"},"Component":"Not Applicable","Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"Not Applicable","Edition":"","Line of Business":{"code":"LOB10","label":"Data and AI"}}]

Historical Number

50236

Document Information

Modified date:
16 April 2020

UID

swg21474877