To correct for multiple correlation tests, we estimated the false discovery rate (FDR) for a given P value, i.e. a q value, and significance levels were considered at q < 0.05, i.e. a 5% FDR cutoff.
We did multiple correlation tests (morphology, paramere shape, body shape vs. mtDNA branches) [ 29, 30] or ANOVA, with the amount of mtDNA change as explanatory variable, comparing it to different sets of branches (body shape, paramere shape, and morphology) grouped according to their amount of character state changes.
The significance of the increment in the squared multiple correlation was tested when the physical performance was entered after the control variables.
We have 14 predictor variables, requiring a minimum sample size of 162 per survey to test the multiple correlation, or 118 to test individual predictors.
A minimum sample size of 50 + 8 m, where m is the number of predictor variables, is recommended for testing the multiple correlation, and 104 + m for testing individual predictors [ 67, 68].
Data for all demographic variables and study variables were analyzed first using descriptive statistics with SPSS version 11.5.[ 18] Multiple correlation and paired t-tests were used to analyze significant differences in ratings between caregivers and patients and associations between related symptoms and conditions.
External bias happens when test scores have multiple correlations with non-test variables for two or more groups of examinees.
A vexing limitation is the issue of multiple correlations and significance testing, given also that the health and healthcare indicators are not independent.
The GLLMM regression model was tested with multiple correlation matrices, and was confirmed with a generalized estimating equation.
