By lovetolearn - 8/3/2017
Hi There,
We've run an online study based on AgeIAT.iqx. This study had four conditions and a 2X2 between-subject design. We ran this study through Inquisit Web and only changed the original pictorial and verbal stimuli in AgeIAT.iqx to four different sets of stimuli, one for each condition. Now we are revising a manuscript for this study and would like to report any potential measurement error that might occur due to the data being collected online through Inquisit Web. Hence, we wonder if we may please seek advice on what component of the latency to the IAT could be measurement errors caused by individual online respondents' computers or internet speed being different from each other.
Thanks so much!
|
By Dave - 8/4/2017
+xHi There, We've run an online study based on AgeIAT.iqx. This study had four conditions and a 2X2 between-subject design. We ran this study through Inquisit Web and only changed the original pictorial and verbal stimuli in AgeIAT.iqx to four different sets of stimuli, one for each condition. Now we are revising a manuscript for this study and would like to report any potential measurement error that might occur due to the data being collected online through Inquisit Web. Hence, we wonder if we may please seek advice on what component of the latency to the IAT could be measurement errors caused by individual online respondents' computers or internet speed being different from each other. Thanks so much!
Internet connection speed does not play a role; Inquisit Web performs all measurements locally on the participant's device, the code is literally identical to Inquisit Lab. Measurement error due to individual device characteristics (e.g. make and model of the particular keyboard attached) is unknowable, but would be within what's described in the relevant literature (see e.g. Plant & Turner, 2009; https://link.springer.com/article/10.3758%2FBRM.41.3.598 ). Note that the IAT is a within-subjects paradigm, i.e. measurement error is not confounded with condition (compatible vs incompatible trials), so measurement error introduced by a participant's hardware will affect both conditions equally. Moreover, the outcome measure -- the D-score -- does not reflect absolute response latency, but is a difference score normed by SD, i.e. there is no reason to think that typical hardware-related measurement error would significantly affect a participant's D-score.
|
|