Subject ID as a Fixed Covariate in Doubly Repeated Measures Design
Locked
Subject ID as a Fixed Covariate in Doubly Repeated Measures Design
 
|
Hello. I am currently working with a data set that is a doubly repeated within subjects design. Sixteen subjects received 1 treatment and a control (control always first and served as own control). I realize the conditions were not randomized, which is not ideal. I am working with a data set for a client and the data are what they are…During data collection, 45 observations at 2-min intervals were recorded. The dependent variable is transcutaneous oxygen partial pressure (tcPO2). It is a very sensitive measure which results in sizeable inter- and intra-subject variability. Hence, the standard deviations are quite large. Because of the within subjects variability, I would like to apply a blocking factor (e.g., subjects; variable name Subject ID) to reduce the within subject variance. In the case of my data, I would like to apply this as a categorical, fixed covariate, not a random effect. For this approach, d.f. would be n-1 (16-1=15). When I apply Subject ID as a fixed covariate (numerical value), d.f. = 1. Not what I want. SPSS lets me use Subject ID (a numerical value) as a random effect, and I do get the appropriate d.f., but I am perplexed as to code Subject ID as a categorical variable to use as a fixed covariate, which would reduce variance even further. BTW: SAS can do this quite easily (I’m told by my colleague who does quite a lot of work in phase 3 clinical trials). SPSS is not as cooperative, it would seem. Unfortunately, I do not know the SAS syntax for this procedure. Any help you can provide would be greatly appreciated. Michael Coyle |
Re: Subject ID as a Fixed Covariate in Doubly Repeated Measures Design
|
|
What procedure are you using? What's your current syntax? Alex
Hello. I am currently working with a data set that is a doubly repeated within subjects design. Sixteen subjects received 1 treatment and a control (control always first and served as own control). I realize the conditions were not randomized, which is not ideal. I am working with a data set for a client and the data are what they are…During data collection, 45 observations at 2-min intervals were recorded. The dependent variable is transcutaneous oxygen partial pressure (tcPO2). It is a very sensitive measure which results in sizeable inter- and intra-subject variability. Hence, the standard deviations are quite large. Because of the within subjects variability, I would like to apply a blocking factor (e.g., subjects; variable name Subject ID) to reduce the within subject variance. In the case of my data, I would like to apply this as a categorical, fixed covariate, not a random effect. For this approach, d.f. would be n-1 (16-1=15). When I apply Subject ID as a fixed covariate (numerical value), d.f. = 1. Not what I want. SPSS lets me use Subject ID (a numerical value) as a random effect, and I do get the appropriate d.f., but I am perplexed as to code Subject ID as a categorical variable to use as a fixed covariate, which would reduce variance even further. BTW: SAS can do this quite easily (I’m told by my colleague who does quite a lot of work in phase 3 clinical trials). SPSS is not as cooperative, it would seem. Unfortunately, I do not know the SAS syntax for this procedure. Any help you can provide would be greatly appreciated. Michael Coyle |
Re: Subject ID as a Fixed Covariate in Doubly Repeated Measures Design
|
|
If you are using GLM Univariate, you can specify a fixed blocking factor by putting it in the Fixed Factors box in the GUI. SPSS will treat it as categorical with n-1 df, even if its a numerical id number. Garry Gelade From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Alex Reutter
|
| Free forum by Nabble | Edit this page |