Is it a quasi-experiment or something else?
I plan to conduct a research whereby the respondents would be asked to read the first ad, and complete a questionnaire attached to it, before they read the second and third ads, and complete the respective questionnaire in a sequential manner. As these three ads are different, can it be considered as intervention? The purpose is to assess whether the third ad would yield greater/different responses compared to the first and second ads. Please advise. Thank you.
In an experiment, we (researchers) randomly assign participants to conditions. Then we observe if our experimental manipulation causes an effect. A quasi-experiment is when participants are presumably randomly assigned to conditions by happenstance. The experimental manipulation happens for some reason beyond our control. And we analyze the result in hopes that this naturally-occuring experiment can tell us about cause and effect. You’re study is definitely not a quasi-experiment, because you control everything. I would call your study a within-subject experimental design, because each participant receives all 3 ads. You didn’t say explicitly, so I’m inferring. But I think your experimental manipulation is random assignment to 1 of 6 counter-balancing conditions for the 6 combinations of ad orders. Then your hypothesis is saying that we can predict something about participants’ responses to the ads by knowing the order in which they saw the ads. Best wishes with your project Hiram. ~ Kevin
I second Kevin's answer, but I would describe this as a single-factor within-subjects design with three levels (and not a quasi-experiment). Your "intervention" is your three ads.
And as Kevin indicates, you should counterbalance or risk the possibility that your ad 3 effect is really an effect like fatigue. There are 3x2x1=6 possible orders and you should randomly assign participants to these orderings.
You don't have to investigate the order effect. After counter-balancing, all ads will have been in each position an equal amount of times, so any ordering effect is equally distributed across the three ads. But if you ignore it, then it goes into the error term.
Thank you Kevin and Alan. I really appreciate your responses.
Do you happen to have any good articles (or links) on these designs?
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics
- Survey Design
- Questionnaire