24 Beyond the Usability Lab Another approach that will help mitigate these effects is to have the study tool pull subsets of tasks for each version so that the participant never performs the same task twice across all of the design versions. But again, you would have to make sure that each task is performed on each design version a sufficient number of times Tip so that a clean statistical comparison can be made. If you want to take a Some vendor services can support these kinds of between-subjects approach for requests with sophisticated scripting, but it can the study but still obtain some get very complicated very quickly and may cost comparative feedback from users, extra for customization. another option to consider is supplementing the online study with Because of these factors, we recommend stick- a lab-based within-subjects study. ing with a between-subjects design whenever The lab-based study will be easier possible! If you have access to a large enough to counterbalance order as it will sample of participants, between-subjects is the be a smaller, fixed number cleanest and easiest approach. Of the several hun- of participants. dred online studies we've done where we compared alternatives, over 90% have been between-subjects.