Foot-in-the-door and problematic requests: A field experiment

Abstract

The “Foot-in-the-door” (FITD) is a compliance technique that consists of making a small initial request to a participant, then making a second, more onerous request. In this way greater compliance with the second request is obtained than under a control condition where the focal request is not preceded by the initial request. Most of the studies using this paradigm have tested prosocial requests. So the generalization of this compliance technique to other types of requests remains an open question. The authors carried out two experiments in which the FITD effect on deviant behaviors was tested. Results showed that the FITD technique increased compliance with the focal request, but only among male participants.

Keywords:

• Foot-in-the-door
• Deviant behavior
• Gender
• Compliance strategies

Freedman and Fraser (1966) found that 53% of a group of housewives agreed that a team of five to six investigators would visit their homes for 2 hours to make an inventory of all the products they used for cleaning and cooking. Three days before this visit the women were asked to fill out a brief questionnaire containing eight questions concerning their product consumption. Without this initial request only 22% of the women agreed to the visit by the investigators. This technique, which consists in preparing the participant with a small initial request before submitting a second one later on, was called the “Foot-in-the-Door” (FITD) technique by Freedman and Fraser (1966). Five meta-analyses of FITD have shown its efficacy on compliance with various requests (Beaman, Cole, Preston, Klentz, & Steblay, 1983; Burger, 1999; Dillard, Hunter, & Burgoon, 1984; Fern, Monroe, & Avila, 1986; Pascual & Guéguen, 2005). It is known that this technique is particularly effective for inciting people to respond positively not only to prosocial requests but also to demands of a more commercial nature.

A certain number of studies have shown that this technique could effectively be used, for example, to incite women to undergo screening for breast cancer (Dollin & Booth-Butterfield, 1995), or to incite students to carry a card designating them as organ donors (Carducci, Deuser, Bauer, Large, & Ramaekers, 1989). Recent studies show that this technique has a positive effect on the compliance rate to a request as well as on the involvement of the participant. For example, Guéguen and Fischer-Lokou (1999) showed that stopping someone in the street and asking them the time before asking them for money predisposed the participant to accept this request more readily (43%) than after a direct demand (28%). Moreover, these same authors observed that the donations from the first group were larger than those from the control group. In the literature various theoretical explanations for the FITD effect can be found. The most popular explanation for the effectiveness of the FITD technique is based on the self-perception theory (Bem, 1972). This theory asserts that people develop their attitudes by observing their behavior and concluding which attitudes might have caused them. The theory is counterintuitive in nature, as the conventional wisdom is that attitudes come prior to behaviors. In this way Freedman and Fraser (1966) considered that the initial request makes the participant feel that they are helping other people, and once this perception is activated it favors compliance with the second request.

Surprisingly, no publications have tested the FITD technique with problematic target requests. However, it does seem plausible that FITD may be a commonly used strategy in social interaction to get someone to escalate deviant or illegal behavior (in a gang or sect situation for example). Furthermore, according to the self-perception theory it might be possible that this technique works for this kind of request. After agreeing to the first request, participants might feel they were the type of person who helps other people, and this feeling might favor compliance to the second request, even if complying with it involves deviant and illegal behaviors—because these behaviors would still be helping behaviors. Therefore two experiments were conducted in a natural setting in order to evaluate the effect of the FITD technique for obtaining deviant and illegal behavior. In the first study a prosocial act was followed up with a deviant request, whereas in the second study two deviant acts were requested. According to self-perception theory we hypothesized that helping someone in complying with a small first request (illegal or not) will encourage participants to help with a second larger illegal request.

Study 1

Method

Participants

The participants were 74 pedestrians (33 men and 41 women) who were walking alone in a street in the city of Bordeaux (France).

Procedure

According to a random assignment, a male confederate (32 years old) acted in two different ways. In the FITD condition the confederate made the first request to the participant: “Excuse me, Sir/Madam, I’m a sociology student, would you mind answering a few questions for a public opinion poll about the referendum concerning the European constitution?”1 After the participant had complied, the confederate thanked him/her. Immediately he then asked, “Could you please help me again. I would like to take this road sign home [pointing his finger]. Would you please hold my step-ladder while I unscrew the sign?” (The step-ladder was nearby.) After the participant's acceptance or refusal, the confederate explained the real purpose of the experiment. Under control condition the participant was not asked the first question. After the interaction participants were personally debriefed by the experimenter in order to explain the real goal of the experiment. This debriefing lasted between 20 and 60 minutes and included participants’ questions. This point was important because of the nature of the target request, which was clearly deviant.

Results

Compliance with the final request was the single dependent variable. A 2 (experimental conditions) × 2 (participant gender) Log-Linear analysis was performed with the compliance of the request as the dichotomous dependent variable. Data in the FITD condition were analyzed using the total number of participants tested and not only participants who agreed with the first request. So, in the FITD condition, one man and one woman refused the first request but were included in analysis as non-compliant participants.

Neither the main effect of the experimental condition, χ²(1, 74) = 0.45, p = .50, φ = .08, nor the main effect of gender of the participants, χ²(1, 74) = 3.01, p = .08, φ = .20, was significant. However, an interaction effect between experimental conditions and gender was found, χ²(4, 74) = 10.51, p = .001, φ = .38. Under the control condition men were less compliant (37.5%) than in the FITD condition (70.6%). This difference is statistically significant, χ²(1, 33) = 3.64, p = .05, φ = .33. On the other hand, under the control condition women were more compliant (55%) than in the FITD condition (14.3%). This difference is also statistically significant, χ²(1, 41) = 7.55, p = .006, φ = .43.

STUDY 2

Method

Participants

The participants were 160 adults (80 men and 80 women) who were standing in a bookshop and stationer's shop, consulting various magazines when they were solicited.

Procedure

A young male confederate (12 years old) was used as the request solicitor in this experiment.2 According to a random assignment the confederate acted in two different ways. In the FITD condition the confederate made the first request to the participant: “Excuse me, Sir/Madam, would you please help me to get that magazine because I’m not tall enough to reach it.” During the formulation of the request the confederate pointed to a pornographic magazine that had been placed in a corner two meters up. The confederate then told the participant, “It's the red magazine. Can you see it?” When the participant agreed, the confederate thanked him/her. Immediately afterwards the confederate then added: “Could you please help me again? I can’t buy this magazine because I’m under age. Could you please buy this magazine for me?” While formulating this request, the confederate took out a 200 FF banknote (this study was conducted in 2001 before France used Euro). Then he added: “I’ll wait for you outside so no one can see what you’re doing for me.” If the participant refused, the confederate began to debrief him, and the experimenter came and explained the purpose of the experiment. If the participant complied, the confederate left the shop and waited outside until the participant came out. Then the experimenter came up to the participant and explained the real purpose of the experiment.

Under the control condition the first request was not made. The confederate only said to the participant: “Excuse me, Sir/Madam, could you please help me? I can’t buy that magazine, [pointing his finger] the red one. Can you see it? Because I’m under age. Would you please buy it for me?” Again, when formulating this request, the confederate took out a large banknote. Then he added: “I’ll wait for you outside so no one can see what you’re doing for me.”

As in the first study, participants were personally debriefed by the experimenter, considering the deviant nature of the target request. Similarly, the experimenter paid careful attention to debriefing the young confederate.

Results

Compliance with the final request was the single dependent variable of our experiment. A 2 (experimental conditions) × 2 (participant gender) Log-Linear analysis was performed with the compliance of the request as the dichotomous dependent variable. In the FITD condition all participants accepted the first request. A main effect of the experimental condition was found, χ²(1, 160) = 7.59, p = .006, φ = .21, revealing that, overall, more participants agreed with the confederate's request in the FITD condition (21.3%) than in the control condition (6.3%). A main effect of gender of the participants was found, χ²(1, 160) = 7.59, p = .006, φ = .21, revealing that men complied more frequently with the final request (21.3%) than women (6.3%). An interaction effect between experimental conditions and gender was found, χ²(4, 160) = 16.32, p < .005, φ = .32. Further analysis revealed that the foot-in-the-door effect was only obtained with men, who complied by 32.5% in the FITD condition and 10.0% under the control condition, χ²(1, 80) = 6.05, p = .014, φ = .27, while no difference between the FITD condition (10%) and the control condition (2.5%) was found with women, χ²(1, 80) = 1.92, p = .17, φ = .15. Again, it seems that only men's behavior was influenced by the technique, whereas under the two conditions the behavior of women remained the same.

Discussion

In the two experiments it was found that the FITD technique appeared to be an effective technique for increasing the probability that men would produce deviant and illegal behavior. However, the FITD technique was ineffective in producing deviant behavior in women. Such unexpected moderation by gender effect is interesting because a lot of studies have been carried out where no gender effect was found. Various meta-analyses (Beaman et al., 1983; Burger, 1999; Dillard et al., 1984; Fern et al., 1986; Pascual & Guéguen, 2005) confirm that the FITD technique led to an increase in both male and female compliance. Furthermore, there was a significant difference between previous research on the FITD technique and our experiments. In the studies cited above the requests used by the experimenters were essentially prosocial (i.e., a donation of money to a humanitarian organization). In our experiment the target requests were problematic and cannot be entirely considered as prosocial requests, even if they concerned helping behaviors (helping to steal a road sign in Study 1; purchasing the pornographic magazine for a child in Study 2). So, observing that women are not affected by the FITD technique with deviant requests could possibly mean that women are clearly more reluctant to produce deviant behavior, and the absence of effect of the FITD technique is evidence of this reluctance. Indeed, researchers have shown that, in general, boys are more antisocial than girls (Coie & Dodge, 1998). Duke (1978) concluded that girls committed fewer acts of misbehavior in schools than boys, and that different models for boys and girls are required to understand the origins and development of antisocial behavior during childhood and adolescence. According to Duke, parental control towards boys in the education process was more permissive than towards girls, and might explain the gender difference in our experiment. According to sex-role socialization theory women have been socialized to obey to the rules, whereas the socialization of men is less insistent in this regard. So, concerning antisocial behavior, Eagly and Steffen (1986) concluded that social norms tend to encourage some forms of transgression in boys but not in girls. In Denmark, Bonde, Obel, Nedergard, and Thomsen (2004) found that, in 3-year-old children, boys are more likely to produce deviant behavior. In a study of the antisocial behavior of adolescents living in Hong Kong it was found that boys were significantly more antisocial than girls (Ma, Shek, Cheung, & Lee, 1996; see also Ma, 2005). Moreover, some studies found that higher representation of women in government reduced corruption (Dollar, Fishman, & Gatti, 2001; Swammy, Knack, Lee, & Azfar, 2001).

More precisely, in Study 1 it is possible that men perceived themselves as being a “helping person” after the acceptance of the first “prosocial” request, which could have led them to accept the deviant target request because it was still a helping behavior, and moreover because according to sex role socialization theory men are more inclined to produce this kind of behavior. So their self-perception of being prosocial in their initial helping did not interfere with their compliance with the follow-up deviant request. On the other hand, we have seen that women are less susceptible to producing deviant behavior because social norms tend to encourage some forms of transgression in boys but not in girls. So perhaps, concerning women, the self-perception of being prosocial in their initial helping interfered with the compliance with the follow-up deviant request. This could explain the reverse effect of the FITD observed in the first study.

In the second study the first request and the target request were both deviant and illegal. As in Study 1, men could have perceived that these two requests consisted in helping someone. In this way the illegal characteristic of the two requests is probably less problematic for them, unlike women according to the sex role socialization point of view. Concerning women, after agreeing to the first small deviant request (all accepted), women could be more reluctant to comply with another deviant and illegal request because, according to sex-role socialization theory, it could be difficult to accept. So the control condition did not differ in the FITD condition for women.

Conclusion

Such findings have some methodological and theoretical interest. It will be interesting for social scientists to test attitudes and behavior using such compliance techniques. It has been found in numerous studies that FITD is a good technique for influencing people's behavior. Furthermore, the negative effect (Study 1) or absence of effect (Study 2) of such a technique on female participants in our experiments could mean that some categories of people are clearly more reluctant than others in agreeing to help someone steal a road sign or to buy a pornographic magazine for a teenager. Finally, such results can be interpreted by the self-perception theory and the sex-role socialization theory. However we can point out two limitations in our studies. First, it is possible that the results stem from the specific acts chosen for the studies. In Study 2, for example, the large request consists of purchasing a pornographic magazine. It seems plausible that, on average, this would be an easier act for a man than for a woman. So future research should test a wide variety of target requests to validate our first results. The second problem concerns the fact that we only used male experimenters in the two experiments. Future research should control this variable to ensure that it was not causing a bias in our results.

Notes

¹The experiment took place in May 2005, just before a referendum vote on the European constitution.

²The young confederate was the brother of the experimenter who observed the sequence of events from a distance.