The leadership effect was described earlier by social psychologists as a “modeling effect” (Bryan & Test, 1967; Lincoln, 1977; Reingen, 1982 in Bekkers & Wiepking, 2011). The modelling effect describes the effect on people that see others give to a charity; they can take this as a signal that others have confidence in the organization (Bekkers & Wiepking, 2011). The modeling effect has been studied in diverse settings: help offered at a car breakdown scene and donations to The Salvation Army at the exit of a department store (Bryan & Test, 1967), donations to organ grinders in Amsterdam (Lincoln, 1977), fund-raising efforts by the Heart Association and asking students to donate blood (Reingen, 1982), direct mail fund raising for a charity institution (Vriens, Scheer, Hoekstra & Bult, 1998) and in a laboratory where one participant is induced to have higher status than the other (Kumru & Versterlund, 2005).
In the above mentioned literature, behavior of the respondents who are conditioned to donate by a model, is compared with that of respondents who are not conditioned to donate. The appearance of the model, the duration of his/her appearance and the number of models differs per study. Consequently, the situations where conditioning is presumed vary across studies.
Appearance of the model
In studies by Bryan and Test (1967), Lincoln (1977) and Kumru and Vesterlund (2005) the model physically attends the scene that is observed. On the scene, in sight of the respondents, (s)he helps or donates, conditioning the respondents. In the study by Reingen (1982) models appear textually on a list with names. This list is read aloud to respondents. The study by Vriens et al. (1998) features models in a letter, in this case represented by a signature and a name (CEO of a health care institution or a professor in health care).
Duration of attendance by the model and application of the ‘condition’
On Bryan & Test’s (1967) car breakdown scene, the model is continuously present. The drivers of passing cars are considered conditioned. In their experiment with donations to The Salvation Army the model passes the scene on a round trip every minute, he donates then walks along for 20 seconds, repeating this every minute. Everyone leaving the department store in the 20 seconds of the model’s presence is considered conditioned. Lincoln (1977) considers a passerby on the organ grinder scene conditioned if the precursing passerby donates. Respondents in Reingen’s 1982 study, who a shown and read to aloud a listing of models, are considered conditioned when they are, after a 2 seconds pause, requested to donate themselves. Vriens et al. (1998) consider every respondent conditioned, as they are all presented with the letter signed by the model with induced status.
Number of models
Bryan and Test (1967), Lincoln (1977), Vriens et al. (1997) and Kumru and Vesterlund (2005) each assign and present one model per scene. On the other hand, Reingen (1982) presents a list with multiple models. Bryan and Test (1967) and Lincoln (1977) report the possibility that respondents are conditioned by someone else than the designated model. For example, in the Amsterdam organ grinder scene, it is possible that the precursor’s precursor is, by chance, being observed by the respondent. Lincoln (1977) presumes this is not problematic for the validity of the tests, as the effect of the designated model can only be understated if no action is taken to compensate for this effect. Bryan and Test (1967) take action by designing their tests more conservatively.
Research design implications
Studies regarding the modelling effect show significant diversity in operationalization. Assignment of model and conditioned status appears to be determined situationally. Who functions as a model, when and for how long a respondent is considered conditioned depends strongly on the research design. When designing a research design for testing the model effect it is important to make sure that the assumption holds that these seemingly random factors do not lead to an overestimation of the effect of conditioning; the analyses should be designed conservatively. As long as these criteria are met, models and conditioning status may depend on situational factors. Therefore, in this study, the duration of the modelling effect is assumed to be 24 hours, for the practical reason that lending behavior is also measured per day.