Stratified analysis

Learn me by follow me exercises

This module tests for associations between risk factor and outcome stratified by each of potential confounders or effect-modifiers.

Given a set of categorical covariates, this module tests the associations of risk factor and outcome within each level of each covariate. The module can also report associations within levels of joint distributions of more than one covariate.

The module automatically detects whether the outcome variable is binary or continuous and selects logistic or linear regression as appropriate.  You also can specify the distribution and link function for each outcome manually, right click the variable and then select “Change functions”. In the popup window, select the distribution and link function.

What it’s used for

Stratifying on potential confounders and/or effect modifiers can provide insight into the structure of relationships within a dataset and help to guide decisions on whether covariates should be controlled or used to define stratifications of the data.  

If the associations between risk factor and outcome differ a lot in different levels of a covariate, it might be more appropriate to present a stratified analysis or specify an interaction term in the model.

Screen shot of sample designs:

t3_input.gif    t3_input2.gif

Sample output tables:

t3_sreg.gif    t3_sreg_2.gif