custom hypotheses in MIXED model
Locked
custom hypotheses in MIXED model
 
|
Hi All,
I am sure that this should be a very simple problem but I am having a lot of trouble performing this task: I am analysing a study where we had 12 Patients hooked up to EEGs and asked them a series of questions with yes/no answers and looked at the left and right side of the brain (seperately) before, during and after the question was asked. I was looking at whether there were significant interactions between the fixed effects using the MIXED command in SPSS as follows: 12 Patients (Random Factor) Side: (Fixed Factor) - Left, Right Time: (Fixed Factor) - T1, T2, T3 Question: (Fixed Factor) - Same, Different Specifically the Syntax is: DATASET ACTIVATE DataSet1. MIXED POWER_EEG BY PatientNum Epoch Location SemanticCondition /CRITERIA=CIN(95) MXITER(100) MXSTEP(5) SCORING(1) SINGULAR(0.000000000001) HCONVERGE(0, ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE) /FIXED=Epoch Location SemanticCondition Epoch*Location Epoch*SemanticCondition Location*SemanticCondition Epoch*Location*SemanticCondition | SSTYPE(3) /METHOD=REML /PRINT=CPS DESCRIPTIVES /RANDOM=PatientNum | COVTYPE(VC) /EMMEANS=TABLES(OVERALL) /EMMEANS=TABLES(Epoch) COMPARE ADJ(LSD) /EMMEANS=TABLES(Location) COMPARE ADJ(LSD) /EMMEANS=TABLES(SemanticCondition) COMPARE ADJ(LSD) /EMMEANS=TABLES(Epoch*Location) /EMMEANS=TABLES(Epoch*SemanticCondition) /EMMEANS=TABLES(Location*SemanticCondition) /EMMEANS=TABLES(Epoch*Location*SemanticCondition) . All I want to do now is see whether the difference in the means of certain groups are signficant at the 5% level. Specifically I want to test whether For the Left Side of the Brain during the T2 Time period, whether the difference in the mean for the 'same' 'different' is significant. I know I need to use the /TEST command, but I don't know how to compare different means of one factor at the SPECIFIC level of another factor. Also - can anyone please point me towards a good explanation of the /TEST command, I've been using the SPSS Statistics 17.0 Command Syntax Reference - but Its not helping me out too much. |
|
|
What exactly is the depdendent variable?
Ryan
On Mon, Aug 30, 2010 at 2:03 AM, thrillhouse86 <[hidden email]> wrote: Hi All, |
Re: custom hypotheses in MIXED model
|
|
Sorry I should of made that clearer - basically for each patient for each side of the brain for each question we take a voltage time-history of the EEG before, during and after the question is asked and we turn that into a single scalar value based on the variance.
In my Syntax its called POWER_EEG Thanks, Thrillhouse |
|
|
Thrillhouse,
The code you present needs work. But before I can begin to address your code, I need you to provide an illustration of what your data set looks like in long format.
Ryan
On Mon, Aug 30, 2010 at 8:56 PM, thrillhouse86 <[hidden email]> wrote: Sorry I should of made that clearer - basically for each patient for each |
Re: custom hypotheses in MIXED model
|
|
OK I've just copied across 3 chunks of the data in long code format (there is a lot of it) - don't worry about "True_PatientNum" I never ended up using it.
"PatientNum" "True_PatientNum" "Epoch" "Location" "SemanticCondition" "QuestionNum" "Power_EEG 32 1 1 1 1 14 -1.18 32 1 1 1 1 17 1.68 32 1 1 1 1 21 -3.11 32 1 1 1 1 22 1.38 32 1 1 1 1 23 -0.19 32 1 1 1 1 26 -0.55 32 1 1 1 1 27 -1.60 32 1 1 1 1 3 -0.73 32 1 1 1 1 5 0.39 38 2 2 2 2 11 -0.17 38 2 2 2 2 20 0.06 38 2 2 2 2 25 -0.10 38 2 2 2 2 8 0.08 38 2 3 2 2 11 0.37 38 2 3 2 2 20 0.30 38 2 3 2 2 25 0.29 69 5 1 2 1 17 0.27 69 5 1 2 1 18 0.52 69 5 1 2 1 21 -0.14 69 5 1 2 1 26 0.51 69 5 2 2 1 12 0.60 69 5 2 2 1 14 0.92 |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |