Abstract
A random-effects meta-analysis was undertaken to examine the effectiveness of the Door-in-the-Face (DITF) persuasive message strategy on compliance. Results indicate an overall significant effect of the DITF strategy on verbal compliance (k=78, r=.126), but an insignificant effect for behavioral compliance (k=39, r=.052). In terms of verbal compliance, the DITF strategy works significantly better than controls for different samples, across varied communication media, and for prosocial causes. Additionally, the DITF technique is more successful than controls for volunteering/research than other target behaviors (e.g., monetary donation). For both verbal and behavioral compliance outcomes, the toughness (measured as amount of baseline compliance) of the donation context negatively predicted the magnitude of the DITF effect. It is argued social responsibility theory best accounts for observed moderator factors.
Notes
1. According to Lipsey and Wilson (Citation2001, p. 201), the following formulas were used to compute effect sizes based on a 2×2 table with cell frequencies of a, b, c, and d.
In instances where the total sample size and the χ2 statistic with df=1 was available, an alternative formula was also utilized:
2. For example, Ebster and Neumayr (Citation2008) had four message conditions—a standard DITF condition, a DITF condition plus verbal emphasis on concession, a DITF condition plus credibility manipulation, and a DITF condition plus credibility manipulation and verbal emphasis on concession—compared to a single, target request-only control group. In this case, the standard DITF condition was considered most representative of the DITF technique and compared to the control group.
3. For example, Even-Chen, Yinon, and Bizman (Citation1978; Experiment 1) employed a moderate initial request and a large initial request DITF condition. The large initial request was randomly selected to represent the DITF condition in the current analysis.
4. Our own re-analysis of the O'Keefe and Hale (Citation1998) data identified the I 2 or percentage of variance explained for their meta-analysis. Our re-analysis replicated all study findings reported by O'Keefe and Hale lending credibility to this value.
5. The current data were also analyzed using Hunter and Schmidt's method to examine for potential differences due to the analysis procedure (see Anker, Reinhart, & Feeley, Citation2010). Overall findings differed little with HS procedure producing overall r for verbal effects of .107 (.099, .114) and for behavioral effects of .038 (−.061, .050).