I made a web-based experiment that is the exact copy of a lab-based experiment.
How can I compare the data from these two experiments?
I mean, statistically speaking, which test can I use?
At first I thought to use a "simple" t test to see if, for example, the condition Y-online is not equal the condition Y-laboratory. The problem is that the result I wish to find is "Y-online results=Y-laboratory results". So the usual "H0: u1=u2; H1: u1 ≠u2" now is "H0: u1 ≠u2; H1: u1 = u2".
I think your a lab-based experiment data is more valid rather than a web-based experiment. A lab-based experiment data is the original one.
The researcher wants to find out data used research instruments. Research instruments items can help the researcher to differentiate which data is more valid, a lab-based experiment or a web-based experiment are. This is one of strategies to trace back the data validity, by re-reading research instruments items.
Thanks
I think the strongest design would use Structural Equation Modelling to compare whether a single underlying model can fit both datasets, or whether you need to specify two separate sets of coefficients.
Technically, this is known as a "moderation" design where your mode of data collection would be the moderating variable. This is also equivalent to testing for interaction effects. So, if you have only a single independent and dependent variable, then the mode of administration would be the source of the interaction effect.
Agreeing with David L Morgan, you can do a comparison of an experiment and its replication using Structural Equation Modeling. You might find the following study helpful.
Stapleton, L. M. (2008). Variance estimation using replication methods in structural equation modeling with complex sample data. Structural Equation Modeling: A Multidisciplinary Journal, 15(2), 183–210. https://doi.org/10.1080/10705510801922316.
Article Variance Estimation Using Replication Methods in Structural ...
Good luck,
- Badges
- Science topic
- Similar topics
- Cognitive Science
- Thinking