[Home Page] [Previous Page]
[Old Roundtable Archives] [New Roundtable]
Posted by Stan at BeginSurvey.Net on February 02, 2000 at 17:02:16:
In Reply to: Re: We will now ask you some screening questions... posted by Daniel Schulman on February 02, 2000 at 00:21:12:
I don’t know why there’s this impression that allowing non-targeted respondents to proceed through an online survey has no discernible negative impact – either economically or from a research standpoint.
For those proposing that the incremental cost for including non-targeted respondents is miniscule, you are incorrect. From a programmer’s perspective, there would be additional programming required in order to make a professional online survey for several classes of respondents. Using the unlicensed driver example, how would an unlicensed driver answer any questions about driving conditions? … favorite route to destination questions? … etc.
How do you envision the survey proceeding when both a licensed driver and unlicensed driver come upon questions only applicable to persons legally able to drive? If it’s with branching or skip patterns, the programming for these things doesn’t happen magically. And if the majority of a targeted survey is applicable to only the “qualified” group, then there will be a lot of additional code needed because someone thinks he might want “nice-to-know” information.
If there is a suggestion that to bypass additional programming, all one needs to do is put a message window on every screen with something like “Unlicensed drivers … this question is not applicable to you … please skip to the next question”, then I believe this corrupts the results in a way similar to position, color, or format bias on any self-administered survey instrument
Finally, there are additional costs in project director labor in composing an “everyone’s included” questionnaire, as well as additional costs in the analyses because of cleaning or filtering data of non-targeted respondents.
::::slapping Spain around for agreeing so easily and not seeing the extra programming time needed to accommodate non-targeted respondents::::
Stan
Subject: