Technical details
The precise meaning the Overall significance level is determined by other Statistical Assumptions settings as follows:
- Where Multiple comparisons method is set to None or Fisher LSD in both Cell comparisons and Column comparisons then Overall significance level has the standard interpretation (i.e., the specified Overall significance level is the \(\alpha\) and the result is marked as being “significant” if the \(p \lt = \alpha\).
- Where Multiple comparisons method is set to False Discovery Rate (FDR), False Discovery Rate (pooled t-test) or Non-parametric False Discovery Rate, the Overall significance level is interpreted as the False Discovery Rate (often referred to as \(q\), and it is an estimate of the proportion of the cells shown as being significant which will in fact be “false positives” (i.e., not reflective of differences in the population). Additionally, the way that the false discovery rate is computed is also determined by the selection of the Within row and span options (when selected, the false discovery rate is recomputed within each set of tests conducted in within a span in a row; otherwise, the computation is done using the entire table).
- When the Multiple comparisons method is set any other option (i.e., which does not include None or False Discovery Rate), the Overall significance level is the familywise error rate. Where Bonferroni, Bonferroni (pooled t-test) or Non-parametric Bonferroni is selected, the Within row and span option also has an effect. This is described in more detail in Multiple Comparisons (Post Hoc Testing).