@jkosie and I simultaneously found the following issue:
When the gender variable is edited in 02_variable_validation.Rmd the saved values (e.g., replacing gender = 0 with NA or gender = MALE with M) are saved in a variable called gender2. When gender is used in later analyses, the original, unedited gender variable is used.
As consequence, one of the conclusions, but the estimates, change in the analyses including gender, the estimates get even smaller than reported in the paper. (0.007 vs 0.01 for the main effect and -0.00826 vs -0.01 for the interaction effect, which round to the same values).
@jkosie and I simultaneously found the following issue: When the
gender
variable is edited in02_variable_validation.Rmd
the saved values (e.g., replacing gender = 0 with NA or gender = MALE with M) are saved in a variable calledgender2
. When gender is used in later analyses, the original, uneditedgender
variable is used.As consequence, one of the conclusions, but the estimates, change in the analyses including gender, the estimates get even smaller than reported in the paper. (0.007 vs 0.01 for the main effect and -0.00826 vs -0.01 for the interaction effect, which round to the same values).