Hi all   I have two datasets addressing essentially the same question about ideal body shape and body composition.   In the first, between subjects dataset (BETWEEN), participants are assigned to one of three experimental groups: personal ideal body shape (PERS), cultural ideal body shape (CULT), and most attractive body shape (ATT). In each group, participants judge images of body composition that fit the requirements of that condition. Then, the average muscle and adipose are calculated from their decisions. We also gather information about 2 covariates: the age and actual body mass index (BMI) of the participant.    In the second, within subjects dataset (WITHIN), each participant makes judgements about all  three conditions: personal ideal body shape (PERS), cultural ideal body shape (CULT), and most attractive body shape (ATT). The average body composition for the bodies in each of the three conditions is computed for each participant, and the same covariates, age and BMI obtained.    It is relatively straight forward to carry out appropriately structured MANCOVA style analyses using PROC GLM, separately for the BETWEEN and WITHIN datasets. But this means that I cant directly assess the potential effect of sample type - i.e. WITHIN versus BETWEEN. There are reasons to do with psychological biases that I would like to estimate this.    So, does anyone know if I can build a single model for the multivariate output (each decision about body composition renders both an adipose a skeletal muscle value), which  incorporates sample type as a fixed effect, along with condition (PERS, CULT, and ATT), and the two covariates. I imagine that at least one difficult bit is structuring the random effects correctly.    Bottom line, I would be extremely grateful to anyone who can provide expert advice on this problem.   Many thanks in advance   Piers   ... View more

Hello,   I have individual time series data from 32 EEG channels. In the data, time runs from -200ms to 800ms. Zero is where a stimulus is presented. Two groups of subjects (10 patients v 20 controls) carried out four experimental conditions. My data have been averaged across trials, separately for each group, participant and condition. In other words, for each participant I have a separate time series for each condition, which has been averaged across trials.   For individual EEG channels I want to overplot the average EEG signal for both groups (on the y-axis) as a function of time (from -200 to 800ms, in 10ms increments), and indicate where in time the two time series are statistically significantly different from each other - i.e. effectively a set of multiple pairwise comparisons, one for each time point.   So far, to deal with the fact that we have a repeated measures design, I have used PROC MIXED as follows:   ods output diffs = diff; PROC MIXED DATA=mlm2 COVTEST METHOD=ML;  CLASS sub group time;    MODEL EEGdata = group time group*time / SOLUTION                                                                         OUTP=pred;   RANDOM INTERCEPT / SUBJECT=sub TYPE= un;   LSMEANS group*time / diff; run;   data subset; set diff;  if group = _group then delete;  if time ^= _time then delete;  timepoint_id = group||_group||time1;  raw_p = Probt;   keep timepoint_id raw_p; run;   proc multtest inpvalues=subset holm hoc fdr; run;   The general idea was to use LSmeans to get a set of pairwise comparisons which would give me a complete set of uncorrected t-tests. Then select the ones I want and run those p-values through proc multtest to do the statistical corrections. I would have liked to use the slice option, but could not find a way to obtain adjustments for the p-values, directly, so I settled on this rather clumsy route. I know you can use the adjust option in the lsmeans statement, but the only way I could think of using this to prevent computing all pairwise comparisons (thereby obtaining a very conservative adjustment as a result), was to use the lsmestimate route - but coding that looked horrific with 100 time points.   In summary:   1) Does anyone see an objection to the  use of proc mixed to run the basic model in this way? 2) I am aware that taking account of autocorrelation in the data would be ideal, but including a random effect like this:     random intercept time / subject=sub type=ar(1)     caused the models to run forever, even when I resampled the timeseries to a much coarser scale. This also made me think that I          might be doing something wrong. 3) Are there better / smarter / more elegant ways of obtaining the pairwise comparisons I am seeking.     Bottom line, help needed please ....   Thanks   Piers C  ... View more