1

I am using the linear mixed function in SPSS and want to determine whether a continuous variable (attendance) has an effect on the repeated measurement of my DV (GLS). My hypothesis is that attendance does have an effect on the DV. It was suggested that I add attendance as a covariate but also as a fixed effect. From what I've read it seems like fixed effects are always categorical so does this make sense?

MIXED GLS BY TIME WITH ATTENDANCE

/FIXED=TIME ATTENDANCE |SSTYPE(3)

/METHOD=REML

/REPEATED=TIME | SUBJECT(PARTICIPANT_ID) COVTYPE(UN).

OR, should attendance only be a covariate? 

MIXED GLS BY TIME WITH ATTENDANCE

/FIXED=TIME ATTENDANCE |SSTYPE(3)

/METHOD=REML

/REPEATED=TIME | SUBJECT(PARTICIPANT_ID) COVTYPE(UN).
Sven Hohenstein
  • 6,833
amyk
  • 11
  • There is an underlying statistical question here but nevertheless this looks like a question on SPSS syntax to do something. If so, it is off-topic here. Please see advice on software-specific questions in the Help Center. – Nick Cox Nov 18 '15 at 12:15
  • 1
    @NickCox I agree, but I think the question could be edited to emphasize the statistical point. The main question seems to be statistics, not syntax. – Peter Flom Nov 18 '15 at 12:22

1 Answers1

1

Standard regression model includes slope and intercept, regression is related to ANOVA with dummy-coded group memberships. Mixed effects model can include, besides the fixed effects as in regression, also random effects: random slopes and random intercepts (cf. Gelman & Hill, 2006). Grouping variable cannot be continuous, so such case is impossible, however you can have fixed effects for dummy-coded groups - the same as in linear regression.

Gelman, A. & Hill, J. (2006). Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge.

Community
  • 1
Tim
  • 138,066