Date:         Mon, 24 Jan 2005 08:04:13 +1100
Reply-To:     paulandpen@optusnet.com.au
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Paul Dickson <paulandpen@optusnet.com.au>
Subject:      Re: Sponsor Bias
Comments: To: Tiffola@aol.com
Content-Type: text/plain

Tiffany,

Is there some objective way to compare % for all of the stores that are selected,
while taking into account the bias toward the sponsor?

You would need at least two conditions, one would be maybe 50% where the sponsor
is identified, and another 50% where the sponsor is not identified.  That way you could
compare proportions/percentages of responses using chi-square.

If the data are already collected, I would see if there is any research done in the past
with similar actual testing of biases in place and justify your work from there.  The
closer past research replicates your work, the better justified will be your inferences.
That way, at least you would have some estimates of the amount of bias in the
sampling process.

Regards Paul


> Tiffany S.  Perkins <Tiffola@aol.com> wrote:
>
> Has anyone heard of a rule of thumb regarding how much scores
> attributed to the sponsor of a survey should be "discounted?"
>
> For example, if an insurance company does a survey of its clients via
> telephone interviews and identifies itself as the sponsor and at some
> point during the interview, asks the participants to state the 2
> insurance companies they feel are #1 and #2, it is probable that the
> sponsor will get the highest #1 percentage.
>
> Is there some objective way to compare % for all of the stores that are
> selected, while taking into account the bias toward the sponsor?
>
> Any suggestions or references would be greatly appreicated.
>
> Thank you.
>
> Tiffany Perkins-Munn, Ph.D.