Avoiding fallacies when analysing individual data at group level Let us say that you are running an experiment comparing a number of conditions. You have a database of individual scores, nested in groups, which are then nested by condition. Are there any issues/problems with aggregating individual data by summing the total score for each group and then dividing to provide a score for each group?
Obviously, you would not want to make inferences about individual differences based on the group aggregates, but aside from that are there any problems/issues?
 A: You are correct about not wanting to make inferences about individual differences based on the group aggregates (you are avoiding ecological fallacy). 
Also, there are no problems with aggregating individual data to generate group means ("summing the total score for each group and then dividing to provide a score for each group"). Means could not be calculated if we didn't do this with individual data. Good questions to ask though!
A: As mentioned, from the way you frame your question, you are aware and try to avoid ecological fallacy.
A "problem" I can think of relates to sample size. Imagine your individual data has 100 people in both experiment and control groups. Say you only have one independent variable, then your degree of freedom (df) will be 100-1 = 99.
Now you compress everyone in each corresponding group, and end with two only two summary scores, one belongs to experiment group and the other belongs to control group. Now you sample size is only 2, and your df is 2-1 = 1.
Power, the probability to reject null hypothesis when the alternative hypothesis is true, depends on statistical significance criterion used in the test, the magnitude of the effect size, and sample size used to detect the effect. Compressing into group will reduce final sample size, which in turn, will reduce the power.
