Dear Listers,
I'm involved in an industry study on online research quality. I have an
elapsed time field (time stamp in minutes : seconds) for each question. One
theory I have is that response quality is inversely correlated with speed
(interview length), so if someone races through an interview, I'd expect
greater levels of response inconsistency, more pattern response, mismatches
versus proven benchmarks, etc. This result may or may not be normally
distributed; in fact, it may follow more of a traditional Poisson
distribution. Other than an overall elapsed time measure, or time to
complete a section or attribute grid, what else might you think about
looking at as a way to identify poor data quality (solely using time for
your predictive variable)? TIA,
Bob Walker
Surveys & Forecasts, LLC
www.safllc.com
E-mail message checked by Spyware Doctor (6.0.0.386)
Database version: 5.11580
http://www.pctools.com/en/spyware-doctor-antivirus/=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
