Analysis of data of a Randomized Control trial

[drakshmamc2811]

I had conducted a RCT on the effect of zinc Supplementation on diarrhoea and growth. My study included a follow up of five months in which i collected data for "Number of Episodes of Diarrhoea" every month and measured the Height and Weight of the study subjects every month.
Now finally I have a data were i have number of episodes per month for five month for each subject and the height and weight of each subjects for five months.  
I had recorded the baseline height and weight and baseline for diarrhoea episodes was considered to be zero.
Can anyone please help on how i should proceed with the analysis of this data and how should i go about it in SPSS

Looking forward for a speedy response
Regards
Dr Akash Malik

[Maguin, Eugene]

I'm going to assume that you are going to analyze each variables separately. I'm going to assume your sample sizes are 'sufficient' and your data are appropriate to the assumptions of the following analysis methods.

You have two options. One is a repeated measures analysis using GLM and which makes the standard assumptions regarding repeated measures data. The other is using Mixed. Within mixed, you have two suboptions (and maybe more). One is a growth model in which you assume that each person's data can be fit with some sort of curve, which might include a straight line. The other suboption is to fit a repeated measures model. Certain others are more knowledgeable commentators than I am.

I don't know if you can obtain this book but Judith Singer's recent book on mixed/multilevel/growth models would be very helpful reading for working your analyses.

Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of drakshmamc2811
Sent: Sunday, March 25, 2012 12:27 PM
To: [hidden email]
Subject: Analysis of data of a Randomized Control trial

I had conducted a RCT on the effect of zinc Supplementation on diarrhoea and
growth. My study included a follow up of five months in which i collected
data for "Number of Episodes of Diarrhoea" every month and measured the
Height and Weight of the study subjects every month.
Now finally I have a data were i have number of episodes per month for five
month for each subject and the height and weight of each subjects for five
months.
I had recorded the baseline height and weight and baseline for diarrhoea
episodes was considered to be zero.
Can anyone please help on how i should proceed with the analysis of this
data and how should i go about it in SPSS

Looking forward for a speedy response
Regards
Dr Akash Malik

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Analysis-of-data-of-a-Randomized-Control-trial-tp5593483p5593483.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Ebru Elif Gokcek (Risk Yonetimi)]

27-28 Mart tarihlerinde ofis dışında olacağım.

Acil konulardaki maillerinizi [hidden email] adresine iletmenizi rica ederim.

 

[Poes, Matthew Joseph-2]

I can't add to Gene's comments below without more information.  I can only echo his recommendation to the Singer and Willett book, and also the Bryk and Raudenbush HLM book.  I've taken courses from both of these people on HLM, and have learned quite a bit from both.  Singer's approach is pragmatic to the limitations of SPSS and SAS in doing mixed models, as compared with the purpose built software HLM from Raudenbush.  I'll note that they use different notation from each other, but both write in a very approachable manner.

While setting up your file and writing the program to analyze the data can be a bit daunting at first, it's worth noting that the basic repeated measures RCT analysis couldn't be more basic.  If the RCT experiment involved perfect baseline equivalence between treatment and control, then no adjustments are needed in the covariates and you end up with a model that is effectively a treatment by time factorial design (accounting for time nested within subject correlation).

I'd push for an HLM approach over RMANOVA due to its many benefits and ability to work around common assumptions rarely met.  If by chance you do meet all the assumptions, the end results would be nearly identical anyway, so you lose nothing, and if you don't meet them, then the HLM approach will have superior power to detect effects.

Matthew J Poes
Research Data Specialist
Center for Prevention Research and Development
University of Illinois
510 Devonshire Dr.
Champaign, IL 61820
Phone: [phone]
email: [hidden email]



-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Maguin, Eugene
Sent: Monday, March 26, 2012 4:38 PM
To: [hidden email]
Subject: Re: Analysis of data of a Randomized Control trial

I'm going to assume that you are going to analyze each variables separately. I'm going to assume your sample sizes are 'sufficient' and your data are appropriate to the assumptions of the following analysis methods.

You have two options. One is a repeated measures analysis using GLM and which makes the standard assumptions regarding repeated measures data. The other is using Mixed. Within mixed, you have two suboptions (and maybe more). One is a growth model in which you assume that each person's data can be fit with some sort of curve, which might include a straight line. The other suboption is to fit a repeated measures model. Certain others are more knowledgeable commentators than I am.

I don't know if you can obtain this book but Judith Singer's recent book on mixed/multilevel/growth models would be very helpful reading for working your analyses.

Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of drakshmamc2811
Sent: Sunday, March 25, 2012 12:27 PM
To: [hidden email]
Subject: Analysis of data of a Randomized Control trial

I had conducted a RCT on the effect of zinc Supplementation on diarrhoea and growth. My study included a follow up of five months in which i collected data for "Number of Episodes of Diarrhoea" every month and measured the Height and Weight of the study subjects every month.
Now finally I have a data were i have number of episodes per month for five month for each subject and the height and weight of each subjects for five months.
I had recorded the baseline height and weight and baseline for diarrhoea episodes was considered to be zero.
Can anyone please help on how i should proceed with the analysis of this data and how should i go about it in SPSS

Looking forward for a speedy response
Regards
Dr Akash Malik

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Analysis-of-data-of-a-Randomized-Control-trial-tp5593483p5593483.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD