Date:         Sat, 12 Oct 2002 17:05:36 +0200
Reply-To:     Craig Kolb <craig@bmi-t.co.za>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Craig Kolb <craig@bmi-t.co.za>
Organization: BMI-TechKnowledge
Subject:      Re: Honesty and Missing Income
Content-Type: text/plain; charset="iso-8859-1"

Hi,
I did a study for a large microlending company and took the opportunity to
compare reported income with their database (salaries had to be confirmed to
grant a loan). I will have to look the data file up again to see if there is
anything atypical about non responders - if you can wait a bit I will get
back to you next week.
Craig Kolb
Research Analyst - Enterprise Solutions
BMI-T / IDC Africa
Your knowledge partner in IT, telecoms and the Internet.
Tel: +27 11 540 8000
Fax: +27 11 540 8001
Physical: Building 9 - Pebble Beach; Fourways Golf Park; Roos Street;
Fourways; Gauteng; South Africa
Postal: P.O. Box 4603; Rivonia; 2128; Gauteng; South Africa (for the
interim)
__________________

Visit BMI-T online at http://www.bmi-t.co.za
Confidentiality Warning
=======================
The contents of this e-mail and any accompanying documentation
are confidential and any use thereof, in what ever form, by
anyone other than the addressee is strictly prohibited.

----- Original Message -----
From: "Jeff A." <lists3@jp.pair.com>
Newsgroups: bit.listserv.spssx-l
To: <SPSSX-L@LISTSERV.UGA.EDU>
Sent: Saturday, October 12, 2002 1:08 AM
Subject: Honesty and Missing Income


> ...not directly a spss question, but this is likely quite relevant to many
> here.
>
>
> Is anyone aware of any research that would suggest what the relationship
is
> between the refusal to provide household income on a survey and the
overall
> honesty of the respondent or the accuracy of their survey answers?
>
> We have some analyses where there is a good deal of missing income
> measures.  When we simply delete these cases our effect sizes increase,
but
> our other analyses do not seem indicative of any real interactions between
> income and other variables on the outcome.  In other words, if we use
> alternative measures of socio-economic status, or impute income using
> different techniques, we are finding no interactions of income with our
> other variables.  A few other findings seem to suggest that the
individuals
> in this particular survey who refuse to give their income may simply be
> providing other responses less accurately also.
>
> ...now wondering if there is any prior research that would support or
> refute our assumption.  Maybe something in public opinion quarterly?
>
> Thanks in advance.
>