Search results

In response to Peverill Squire's claim (see SA 36:4/88T6952) that no empirical research had been conducted to determine why the Literary Digest 1936 poll failed in its election prediction, 1937 research is cited (see Cahalan, Don, "Factors Affecting Validity of a Mail Ballot Public Opinion Poll [1936 Literary Digest presidential poll]," Master's thesis, Dept of Psychology, State U of Iowa, Iowa City, 1938). This analysis drew on completed interviews, with results suggesting that a Republican bias in the sample interacted with a much stronger bias in return in 1936 causing the error in the Digest prediction. Also considered is how the Digest was able to successfully predict the three preceding presidential election outcomes. It is concluded that Squire's findings & the 1937 findings are identical. 1 Table, 5 References. C. Grindle

The primary focus of this reanalysis of data from the 1949 'Denver Validity Study' (initially reported by Hugh J. Parry & Helen M. Crossley in 1950) was on relative accuracy of R's of diff age, sex, & SES. The survey entailed 920 interviews in a probability sample of adults drawn from a new city directory, with 45 interviewers given randomly-chosen equivalent assignments. Data obtained in the survey, which was represented as a community survey on issues of primarily local concern, included 15 issues of fact (telephone, home ownership, valid Colo driver's license, valid library card, voting in various elections over the preceding 5 yrs, & R's personal contribution to the latest Community Chest drive) which were checked against official records. Findings include: (1) women were more accurate than men on 6 of 8 items compared, but no diff's were signif to the .05 level; (2) those 21-29 who claimed to have voted had a greater tendency to overclaim their voting (signif at .05 for 4 of the 5 elections) than did older persons; & (3) those of lower SES were consistently lower in the ratio of performance to claims, particularly with respect to Community Chest contributions. (4) Re interviewer effects, in 6 of 8 instances there was a higher proportion of performance in relation to claims (ie, less over-claiming) on the part of Ls R's when interviewed by Ls interviewers, but diff's were not signif at the .05 level. Of the many possible explanations for these generally slight diff's, preference is expressed for the influence of soc norms upon the R's self-image, in which memory may be distorted in order to reinforce a favorable perception of oneself as a good citizen. The implication is that 'getting the truth' is not so much an issue of good rapport or persistence as it is of presenting the issue so that it is not threatening to the R's selfimage. (See also SA 0104/D9158 & 0104/D9163.) AA.

An attempt was made to establish which method of self-administered questionnaires should best be adopted: (1) the field self-administered method, in which names of specific individuals are selected from probability samples, or (2) the by-mail self-administered method, offering great savings in personal & travel costs. D. Cahalan's experience with by-mail sampling is cited. The results show that both methods of surveying have high response rates. But the experiment showed that the by-mail method was likely to be superior for military surveys covering drinking problems & other sensitive issues. It is recommended that care be used in utilizing this method to ensure R's anonymity & to impress him with the need for making a response. 1 Table. S. Karganovic.

The 1st comprehensive conference on PO res was organized by Harry H. Field of NORC & held at Central City, Colorado in the summer of 1946. Backgrounds of the registrants & the topics discussed are described. As a result of the 1946 meeting & 2nd International Conference on PO res was held at Williams Coll, Sep, 1947. The formation of 2 associations also resulted: the Amer Assoc for PO Res (AAPOR) & the World Congress for PO Res (the latter became the World Assoc for PO Res). Crucial issues faced by AAPOR are discussed, especially its position on the enunciation & enforcement of standards. The affiliation of the AAPOR with the Public Opinion Quarterly is noted. The activities of AAPOR in promoting res are described as well as analysis of membership & content of annual meetings. S. F. Fava.