Dear Jieru Chen, 1) What you actually want when you look at response rate is an estimate of non-response error. So rather than worrying too much about what a "minimum acceptable response rate" might be you should be concerned with "what impact non-responses have on my estimates?". Should the impact be near zero - that is should be there no significant difference between those sampling units who responded to the survey and those who did not - then you really have only a larger than expected sampling error and can proceed with sampling unit substitution. The matter is quite different if you ascertain - or suppose - that there might be a substantial biasing factor due to responding behavior.
1) You may also try searching the SRMSNet on http://www.amstat.org/sections/srms/ and http://listserv.umd.edu/cgi-bin/wa?A0=srmsnet to get a point of view from survey statisticians. Best regards, Mr. Luca Meyer consumer research advisor http://www.lucameyer.com/en/ > -----Messaggio originale----- > Da: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] Per conto di Chen, Jieru > Inviato: mercoled� 7 gennaio 2004 16.26 > A: [EMAIL PROTECTED] > Oggetto: [edstat] Response rates in published articles > > > Hi. > > I am very interested in knowing the "minimum acceptable > response rates" for a national RDD survey. What are the RDD > response rates in the published articles these days? > > Thank you. > > Sincerely, > Jieru Chen > . > . ================================================================= > Instructions for joining and leaving this list, remarks about > the problem of INAPPROPRIATE MESSAGES, and archives are available at: > . http://jse.stat.ncsu.edu/ . > ================================================================= > . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
