Linear Mixed Model - How to set up

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

Meitis
Posts: 1
Joined: Thu Feb 09, 2017 9:46 pm

Linear Mixed Model - How to set up

Postby Meitis » Thu Feb 09, 2017 9:51 pm

Hello,

I am quite new to SPSS, but I have been asked to perform an analysis of a data set and I am trying very hard to achieve this. My final question will be if this model is correct or not and where it's needed to be edited.

**Experiment**
First a quick note of the experiment: We tested the knees of 6 subjects. Both the left and the right side were tested to compare. In total this gives an amount of 12. All knees were tested in their pre-operative condition, condition A (n = 12). After this phase a separation was made between the left-sided and the right-sided knees. The left-sided ones were put in a condition B (n = 6). The right-sided knees were put in a condition C (n = 6). Finally, both the left-sided and right-sided knees were put together again in condition D (n = 12).
Another variable that has to be accounted for is the bending of the knee, which goes from 0° to 120° (interpolated for every 10°).
The dependent variable is the position of the knee in a plane (continuous variable).

I am interested in the effect of the 4 conditions (A, B, C and D) on the position of the knee for the bending of the knee in general (not a specific angle). This means repeated measures are used here. Therefor, I was advised by a statistician to make a Mixed Model.

**Data Set**
My data is organized in the following format (long format, so every line represents 1 case):
Variable1: Subject number (1 - 6)
Variable2: Left or Right (1 - 2)
Variable3: Condition (1 - 4)
Variable4: Bending of the knee (1 - 13) (0° - 10° - 20° - 30° - 40° - 50° - 60° - .. - 120°)

This means I have 6 x 2 x 4 x 13 (= 624) cases.

Variable5: Dependent variable (continuous variable)

**Linear Mixed Model**
When I fill in the dialogue boxes of SPSS for the Mixed Models I receive this, but I am not sure whether this is correct or not. Could someone help me with this? And how do I perform further post-hoc tests to see which condition differs from which condition and by how much?

IXED DependentVariable BY Conditions WITH BendingKnee
/CRITERIA=CIN(95) MXITER(100) MXSTEP(10) SCORING(1) SINGULAR(0.000000000001) HCONVERGE(0,
ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE)
/FIXED=Conditions BendingKnee | SSTYPE(3)
/METHOD=REML
/PRINT=CPS DESCRIPTIVES SOLUTION
/REPEATED=Conditions*KneeBending | SUBJECT(SubjectNr*LeftRight) COVTYPE(DIAG)
/EMMEANS=TABLES(Conditions) .

Who is online

Users browsing this forum: No registered users and 2 guests

cron