[Feature Request]: Change comparison group for mixed-models to fit other analyses types

[chrisharrisUU]

Description

For all other analyses (that I checked) the top level of a factor represents the baseline. For the mixed models module it is the opposite.

Purpose

Uniform interpretation of factors across analyses/modules

Use-case

Analyses with factorial independent variables

Is your feature request related to a problem?

No response

Is your feature request related to a JASP module?

Mixed Models

Describe the solution you would like

Unify which level of a factor is the baseline (e.g., always top level or always bottom level)

Describe alternatives that you have considered

No response

Additional context

Factor with two groups:

The audiovisual group has a larger mean than the audio-only group:

For the linear regression (ignoring model assumptions), the audiovisual group seems to be the baseline (negative coefficient for the variable modality):

However, for the mixed-model, the reverse seems to be the case:

[tomtomme]

@chrisharrisUU
Confirmed. Still an issue with JASP 0.19 beta.

As a workaround you can switch the levels of the factor in JASPs label editor, see end of https://jasp-stats.org/wp-content/uploads/2018/01/LabelEditing.gif

[FBartos]

Hi

For all other analyses (that I checked) the top level of a factor represents the baseline. For the mixed models module it is the opposite.

Please note that this is not a correct interpretation of the fixed effect estimates in the current version of Mixed Effect models. The note highlights that the analysis uses Sum contrast coding, which is more appropriate for mixed-effect models and interactions. As such, the "(1)" corresponds to the first contrast rather than the factor level.

Since this is a repeated request, I just added an option to change the factor contrast coding in the Options section (possibly merged in for 0.19).