Kimberly,

Sorry, I inadvertently placed an asterisk between the VAR and TIME effects on the REPEATED statement. The proper specification of the multivariate repeated measures covariance structure is

repeated var time / subject=unit type=UN@CS;

The two effects which are required when you employ one of the multivariate repeated measures covariance structures are 1) the indicator of the response variable (which is always assumed to have an unstructured covariance), and 2) the time frame for the repeated measures. UNIT is specified as the subject on the REPEATED statement. UNIT is NOT an effect.

As to the problem of over 200 time values, I trust that the time variable is constructed with TIME=0 for each units first observation. That is the first thing to confirm. If you are measuring units on roughly a daily schedule, but some units are not started until a half year has passed from the start of the first unit, you would end up with a lot of time values if time were the DAY on which observations were obtained. Moreover, in an experiment in which units starting at different times are expected to have a common variance structure for their starting values, then employing a time variable which is the starting date does not allow you to estimate that common variance structure.

Now, assuming that observations on units are obtained over a period of months with observations intended to be on a monthly schedule but deviating from that with some observations made a few days before you would like and some observations made a few days after you would like, it would certainly be permissable to aggregate observations clustered around the one month period into a single value as long as each unit has at most one observation in any aggregation period.

Of course, no good guidance can be provided without really knowing what the problem is that you are investigating. What is the scientific question which motivates the data collection in the first place? What are the units and the response variables? For any given unit, how do you obtain the repeat measures? Without knowing any of these specifics, it is very difficult to state methods for addressing the problem. It is possible that your simplistic approach of analyzing average responses would be acceptable. Probably not, but possible.

Dale

--- Kimberly Austin <austinkimberly@YAHOO.COM> wrote: > Greetings All, > > And again thank you for your responses to my repeated measures MANOVA > problem. > > I have run into a few difficulties with fitting a repeated measures > MANOVA model in proc mixed. First, when using the repeated statement > in the following code: > > proc mixed; > class unit var treatment time; > model y=var var*treatment var*cov / noint; > repeated var*time / subject=unit type=UN@CS; > run; > > results in the error "two repteated effects are required with this > covariance structure". I believe that both sources of correlation, > the mutliple responses, var, and the repeated measurements on the > experimental units, unit, need to be placed immediately after the > "repeated": > > repeated var*time unit/ type=UN@CS > > but this model will not coverge on parameter estimates. > > However, for the data set I am working with, measurements were taken > over a period of many months. Treating "time" as a class variable > results in a variable with over 200 levels, making the interaction > var*time too complex. > > My question remains, how to properly model both the correlation among > the multiple responses, "var", and the correlation among repeated > measurements on the same experimental units, "unit", which were taken > at unequal intervals and are not balanced among the units. > > It seems the further I venture into this problem the more I am prone > to take a more simplistic approach, perhaps collasping repeated > measurements on units into one mean value for each unit on which to > base the analysis? > > Any suggestions to this problem would be greatly appreciated. > > Much thanks, > > Kimberly Austin > > > "...it is silly to design a study for which the tools to analyze the > resulting data properly are not available." > -Hamer and Simpson. 1998. SUGI:23

===== --------------------------------------- Dale McLerran Fred Hutchinson Cancer Research Center mailto: dmclerra@fhcrc.org Ph: (206) 667-2926 Fax: (206) 667-5977 ---------------------------------------

__________________________________ Do you Yahoo!? The New Yahoo! Search - Faster. Easier. Bingo. http://search.yahoo.com