Dear Mike,
For an analysis it would usually be better if you have an equal number of trials for the different conditions - the more trials, the less noisy the estimate should be. 30 trails seems to be a good guiding value, although this might well depend on the particular design (also, if subjects perform very poorly on one task, you might have to increase the no. of trials if you want to look at correct trials only).
If one task/condition is less frequent it could also affect subjects' behavior negatively, e.g. overall reaction times might be biased as they don't e.g. "learn over time"/"familiarize" to the same extent. However, if it's mainly for confirmatory purpose (having some baseline task instead of just implicit baseline against which to plot the other conditions), then a lesser no. of trials should be okay probably.
Best
Helmut
|