GLM repeated measurements with very large database


New Member
I am trying to perform a multivariate regression analysis in SPSS. The problem is the database is very large (one variable contains over 200.000 measurements) and not every subject has the same amount of measurements. The original database was sorted like this (the dependent variable is blood pressure; independent variables not shown here);

Patientnumber Dependent variable Date etc...

The range of measurements per patients is from 156 to almost 2000. Because this is too much for GLM repeated measurements in SPSS (so I've heard) I medled the measurements per patient per month (so every measurement done in January, unregarded the year, was pooled together under one variable called "January");

Patientnumber January February etc. to December

Now I want to perform a GLM repeated measurements to determine if some climaticological (like temperature) and between-subject factors/covariates (sex, age, use of medication, etc.) have an effect on blood pressure. What is the best way to sort my data for this? Should I keep it like this; 1 case = 1 patient? If so, how can I assign the right climaticological value to the right variable (since every variable is the mean blood pressure of a different month, every variable/month has a different mean temperature too; this means I can't make a variable with Temperature). Or should I turn it back into 1 case = 1 measurement (one mean bloodpressure of that month for a specific patient; so every patient has 12 measurements of blood pressure in one column/variable. This means I can make a variable with Temperature, with the right temperature in the same row as the month). But in that case I can't select the "Within-subjects variables" (since there is only one within-subject variable if i sort the data in this way). I hope someone understands my problem. I should warn you that I'm not used to working with SPSS or statistic problems, so I might not understand some things that appear to be simple.
Last edited: