repeated measures

gill

New Member
#1
Which analysis is appropriate if you have repeated measures of continous explanatory variables and one measure of a continuous outcome variable?
 

JohnM

TS Contributor
#2
gill,

That actually depends more on the questions you want to answer, rather than the nature of the data.

I'm leaning toward just doing a correlation first, but I'd have to hear more about the purpose of your study....

JohnM
 

gill

New Member
#3
Sorry I didn't give much explanantion of my data.

Its looking for a relationship between growth and milk yield in cows. So I have weight, height and girth measurements of the cows when they were calves at birth, 3 months, 6 months, 9 months and pre-service.

My outcome variable is 305 day milk yield during first lactation.

I figured its not much use to relate the actual size of a cow with its milk yield but better to look for an assocaition between growth rate and milk yield.

So I have growth rate (daily rate of change in height / weight / girth) for the periods birth to 3 months, 3 to 6 months, 6 to 9 months and 9 months to preservice.

Its not really suitable to calculate an overall daily growth rate for the whole period as their growth rate is quite different for the different periods.

So I was wondering if there is an analysis in which I could treat these as repeated measures of an explanatory variable and relate them to the outcome - milk yield.?

I have tried a linear mixed model but you cant use this unless the outcome measure and explanatory measures are repeatedly measured at the same time - as far as I know?

I have also analysed it treating each growth period as an independent explanatory variable but I think I am losing some of the power of the study by doing this?

Does anyone have any suggestions?

Thankyou!
 

JohnM

TS Contributor
#4
Sounds like you have a really good handle on the research problem - have you computed correlation coefficients (and tested their significance) between each growth period and the milk yield?

Repeated measures designs usually involve taking multiple measurements of the dependent variable, not the independent variable, as in your study.

You simply have 1 dependent variable, milk yield, and several growth time periods, and you're trying to see if the growth rate in any given period is a strong predictor of milk yield.

Also - did you try taking growth rates across 2 time periods?