Tobacco Documents Online

I .+, I •:-t+ PuUTiOCAL aTUDiFS/ INSTtTUTE FOR' SOCIAL RESEARCH /'THE UNIYEH.odTY' OF MICHIGANI/ ANN ARBUF, MICHiGAN a8t08 August 23, 1985' Prof essor Harry Perlstad t De, ar'tment of Sociology Berkey ilall tlichigan State University East Lansing, MI 48824 Dear Professor Perls'tadt: I am writing'as a fo1!lovup to ou!r recent conversation in which you asked for a description of the methodology used in the Mli'chi'gsn Sur'veys vt:ich we conduct and which included questions on smoking in public places last October. The basic design involves a stratifiedlclus'ter sample of persons 18 and older'resi'ding in, Michigan householde with telephones. Three regions of the state - the exchanges covering the city of'Detroit, the exchanges in remaining suburbs'of the tri~coun'ty area, and'ex'changes in, the rest of, the state - are sampled to achieve equal expected sample sizes and then weighted to reflect their appropriate: propo'rtion of the state's telephoae household population. The purpose of this is to provide for statieticall'y reliable comparisons between groups of individual!s living In these regions. The interviewing is done using Computer Assi'sted Telephone I'nterviewing (CY+TI) procedures. Each respondent is randomly selected from a listing which is made of all adults residing in the household. This listing is made at the start of the interview, an& the selection is peiformed by a computer program which contains the interview schedule and processes this information upon eollection. Only the randomly selected respondent can be intervieved„ and'repe'ated callbacks are made In order to: contact the designated, resp'ondent across the extended intervieving period necessary to conduct surveys'in this f ashion. The October survey was part of a panel study conducted during the fa1i1 election campaign. Of the total o'f 750 interviews conducted N-23

Professor Harry Perlstadt August 23,,1985 Page2 betveu~ October 1!2 and 29, 401 respondcuts hnd' been contacted previr.usly and 349 involved an inslependhnt random samplL, of approximately half the normali size. The response rate for these 349 interviews was 61 percent, which is in the normal range of 60 to 65'percent which vee obtain for surveys conducted under these conditions. The recontact ratefor' the respondentspreviousliy, interviewed in~ S+eptembervas8'2' percent, uhich, also is in the range of our experience in recontacting respondents under these conditions. I: hope this letter respond's to the quesiions which you raised. If I can be: of any further assistnnce,, please feul free to contact me. Sincerely, ~J1 Michael W. Traugott Senior Study Director N-24 Ci i