Is the Door-in-the-Face a Concession?

Abstract

The Door-in-the-Face (DITF) sequential message strategy was investigated in a three-study analysis of existing experimental findings. The current study predicted there would be a positive relationship between concession size and compliance rates in DITF studies. Study 1 included 25 comparisons where size of concession was quantifiable as measured by percentage reduction from initial to target request in the DITF condition. Study 2 data relied on a panel of undergraduate students to provide an index of concession size in 12 additional observations. A third study validated the panel procedure of rating concession size and also provided 9 additional independent observations from the pool of published studies on DITF. Results from each study indicated a positive relationship between concession size and effect size (r = 0.35, 0.55, 0.68, respectively). Study findings provide support for reciprocal concessions explanation for DITF effects.

Keywords:

• Communication Theory
• Compliance
• Social Cognition
• Social Influence

Notes

^[1] Feeley et al. (2012) identified requester variation and type of sample as significant moderators of the relationship between request strategy and compliance. While delay between requests, beneficiary variation, and prosocial cause did not emerge as significant moderators in their analyses, confidence intervals associated effect sizes from scenarios employing requests on behalf of the same beneficiary, prosocial causes, and no delay between requests were statistically significant.

^[2] Situations have been documented, however, where a small concession actually produces even less compliance than merely presenting the target request-only (see e.g., Even-Chen, Yinon, & Bizman, 1978).

^[3] Fern et al. (1986) failed to report in tabular form the studies and comparisons used for meta-analysis nor did they report how magnitude of concession (i.e., small, moderate, large) was coded for each study. Thus, it was not possible to replicate their procedures.

^[4] The criteria used to differentiate quantifiable and non-quantifiable requests in the current analysis were slightly more stringent than those employed by Feeley et al. (2012) in coding their moderator entitled “Reduction in Original Request.”

^[5] Feeley et al.’s original meta-analysis included an ES of r = −0.164, n = 607 derived from Hayes, Dwyer, Greenwalt, and Coe’s (1984) study of DITF requests in the context of blood donation. In re-considering Hayes et al.’s (1984) work, the authors of the current article noted that ES was calculated using requests made to current and prior blood donors, as well as non-donors. In the current study, the ES derived from Hayes et al. was revised to reflect the impact of the request strategy on the non-donor group only. This decision was made to be more consistent with extant work employing the DITF strategy that generally relies on message recipients who do not have extensive experience with the requested target behavior.

^[6] Due to the manner in which the first two measures of difference in request size were calculated, some subjects obtained negative values, indicating that initial requests were perceived as smaller than target requests. The individuals who rated the second request as larger (and hence negative value) were eliminated from analyses. As the third measure employed only a single horizontal response slider ranging from “0” to “100,” negative response values were not possible on that response scale.

^[7] For the 12 comparisons for Study 2 and the 9 comparisons use for Study 3, the average reliability coefficient for concession size index was 0.69 and 0.66, respectively. These reliability coefficients were used to disattentuate coefficients due to measurement error, as recommended by Spearman (1904).

^[8] In Feeley et al. (2012) meta-analysis, the study by Bell, Abrahams, Clark, and Schlatter (1996) contributed four values for ES pertaining to conditions based on respondents’ approval motivation and exchange orientation. As manipulation of approval motivation and exchange orientation was not possible in the current study, a single ES was calculated by collapsing across those variables to represent the overall difference in compliance between the DITF and control conditions. Sample size was increased to reflect this alternate calculation.

^[9] Confirmatory Factor Analyses were performed for Studies 2 and 3 to examine if ratings of legitimacy and concession size were independent factors. Findings indicated little evidence the two factors represent a single unidimensional factor.

^[10] The concession size and reliability coefficients were as follows: Abrahams and Bell (1994; 44.92, 0.75), Conner (2005; 28.95, 0.82), Crano & Sivacek (1982; 37.12, 0.56), Gamian-Wilk & Lachowicz-Tabaczek (2009; 35.02, 0.56), LeCat et al. (2009; 49.93, 0.82), Miller (1974; 37.10, 0.59), Miller et al. (1976; 40.53, 0.74), Reeves et al. (1991; 36.88, 0.65), Snyder & Cunningham (1975; 14.13, 0.55).