Assumptions associated with GLM in SPSS

Home Forums Default Forum Assumptions associated with GLM in SPSS

Viewing 1 post (of 1 total)
  • Author
    Posts
  • #1297
    Alexander Quent
    Participant

    Hello,

    I am new to this group and I have questions both general and specific about linear models.

    I’ve read the book “Discovering statistics using IBM SPSS statistics” 4th Edition, mostly.

    General context of my statistical probelms: I want to run a GLM within subject design on data (classical stroop task) with 3 factor (baseline, congruent and incongruent) n= 20 for two outcome variables (RT & error rate).

    I want to work with this data conscientiously, but while working through the assumptions I realized that I don’t know how to check all of them.

    1. Question: How do I plot zpred vs. zresid and what do I have to infer from it?

    So far I have saved ZRESID and I plotted it against the z-standardized values ( (x – mean) / SD). And nearly all points lay exactly on the line. This result was also obtained when I plotted ZRESID against the values divided by SD. This is at odds with my preconceptions about the assumption for GLMs given that the data is not very close to a normal distribution, which may stem from the fact that n = 20. Is this the right way to plot zpred vs. zresid? The relevant chapters in the book do not provide a detailed explanation of this topic.

    It is reasonable to plot this except in a regression-like context?  (I’ve come across the conception that this whole GLM-testing is regression-like)

    2. Question: How do I check the assumption of addivity and linearity?

    Sadly, the remarks on this topic fall short, too. Do I have to check these?

Viewing 1 post (of 1 total)
  • The forum ‘Default Forum’ is closed to new topics and replies.