A Framework For Explaining Multiple Request Effectiveness: the Role of Attitude Towards the Request

Rohini Ahluwalia, The Ohio State University
Robert E. Burnkrant, The Ohio State University
ABSTRACT - This article proposes a conceptual framework, based on the availability valence hypothesis, for explaining multiple request effectiveness (foot-in-the-door and door-in-the-face). We hypothesize that an important factor determining compliance to the target request is the attitude towards the request (AR). Its position on the Attitude-Nonattitude continuum, would determine whether AR would have a main effect or interact with other variables in this framework (i.e. request behavior and own behavior) in influencing compliance. We propose that the two techniques are influenced differentially by the AR because their underlying processes are different.
[ to cite ]:
Rohini Ahluwalia and Robert E. Burnkrant (1993) ,"A Framework For Explaining Multiple Request Effectiveness: the Role of Attitude Towards the Request", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 620-624.

Advances in Consumer Research Volume 20, 1993      Pages 620-624


Rohini Ahluwalia, The Ohio State University

Robert E. Burnkrant, The Ohio State University


This article proposes a conceptual framework, based on the availability valence hypothesis, for explaining multiple request effectiveness (foot-in-the-door and door-in-the-face). We hypothesize that an important factor determining compliance to the target request is the attitude towards the request (AR). Its position on the Attitude-Nonattitude continuum, would determine whether AR would have a main effect or interact with other variables in this framework (i.e. request behavior and own behavior) in influencing compliance. We propose that the two techniques are influenced differentially by the AR because their underlying processes are different.


How can you induce a person to comply with reasonable requests? This is a question of great importance for non-profit organizations, requiring donations to maintain their operations, for market research companies desiring to increase questionnaire response rates, and for salespersons wanting to induce compliance to their requests without coercion. Indeed it is an important question for marketers in general.

Freedman and Fraser (1966) published two articles that demonstrated strong support for a compliance technique they labeled, the "foot-in-the-door" (FITD) phenomenon. Their data indicated that once a person has complied with a small request, he or she will be more likely to comply with a subsequent larger request. Thus, the FITD technique starts with a small request, which has a high probability of acceptance, and is followed by a larger request which by itself would induce little compliance.

A second technique that has been found to enhance complianceBthe "door-in-the-face" (DITF) was introduced by Cialdini et al. (1975). An extreme request is first made, which is almost certain to be rejected, and this is followed by a more moderate second request, the one which was actually desired.

While reviews by DeJong (1979), Beaman et al. (1983), and Dillard, Hunter and Burgoon (1984) have shown FITD and DITF to be reliable, controversy exists with regard to their underlying processes. Self-perception theory (Bem, 1972) has been employed to explain the FITD phenomenon. Numerous explanations have been put forth to account for DITF: the reciprocal concessions model (Cialdini et al., 1975); the perceptual contrast explanation (Miller et al., 1976); and the worthy person hypothesis (Foehl and Goldman, 1983). Finally, Tybout, Sternthal and Calder (1983) offered and tested a single explanation for both the FITD and DITF: an availability explanation.

The purpose of this paper is to extend the framework provided by Tybout et al. (1983) to account for some of the findings in this literature, that have not been dealt with in previous work. We also attempt to explain more thoroughly the processes underlying the two multiple request techniques and highlight the differences between them. The purpose of this review is to offer a conceptual framework for future research and to propose some testable postulates.


According to the Tybout, Sternthal and Calder (TSC) framework, which is based on the availability valence hypothesis, information about one's "own behavior" (OB) and information about the behavior of the requestor or "request behavior" (RB) determine compliance in multiple request situations. Compliance is enhanced when favorable information about either one's own behavior or the request behavior is available: compliance is undermined when unfavorable information about either own behavior or request behavior is available. Further, there should be no systematic effect when favorable information of one type is as available as unfavorable information of the other type.

In both multiple request strategies there are two requests, and the second one is the "critical" one i.e. the one which is actually desired. In the case of FITD, the first request is small and has a high likelihood of acceptance, thus OB is favorable (due to acceptance of the first request). However, in the case of DITF the first request is very large and has a high likelihood of being refused, implying that OB is unfavorable.

The second request in FITD represents an escalation in the request size (i.e. unfavorable RB) whereas in the case of DITF it represents a concession (i.e. favorable RB). Thus, if RB is more available, the subject's response to the second request would be negative for the FITD and positive for the DITF. However, if "own behavior" (OB) to the first request is more available, then the subject's response to the target request would be favorable for FITD and unfavorable in the case of DITF.

Fern, Monroe and Avila (1986) in their integrative review of the multiple request phenomena present seven theoretical predictions derived from the TSC framework, and provide a synthesis of previous multiple request research results in terms of these predictions. Fern et al. (1986) found some evidence in favor of the TSC framework. Three of the seven hypotheses were supported by the data at p<.10, and there was some directional support for three of the other theoretical predictions.

The TSC framework not only provides a parsimonious explanation for both the multiple request phenomena, but is also able to accommodate a number of the contradictory explanations proposed in this literature. For instance, self-perception theory (Bem 1972) and the concessions explanation (Cialdini et al. 1975) are inconsistent with each other. Self-perception theory would anticipate a lower level of compliance with the DITF technique, and the concessions explanation would predict that FITD will be totally ineffective (Dillard et al. 1984). The TSC framework resolves this inconsistency by stating that, depending on what is available, both the explanations can be supported. In the studies where favorable OB information is available many of the self-perception predictions are met. In studies where favorable request behavior is available many of the concession and the contrast model predictions are supported.

However, the TSC framework is not able to explain some of the findings reported in this literature. For instance, it is frequently found that FITD works in business or non-charity contexts (Goldman et al. 1981; Katzev and Johnson 1984; Patch 1986; Reingen and Kernan 1979), whereas DITF is effective with pro-social requests (Dillard et al. 1984; Foehl and Goldman 1983; Goldman et al. 1981; Mowen and Cialdini 1980). According to Tybout et al. (1983) both the multiple request techniques are more likely to be effective in contexts where requests pertain to favorable issues. They do not anticipate FITD to perform better than DITF in commercial settings, and don't explain why and how the processes underlying the two techniques are different.

Further, some research has shown that even when the issue is favorable and request is pro-social, and thus favorable information is available, multiple request techniques may still not be effective (see Goldman et al. 1984; Kilbourne 1988; Mowen and Cialdini 1980). The TSC framework does not deal with this possibility as well.

The TSC framework proposes that the favorability of the information available would influence the compliance behavior. However, it does not specify very completely what is available and is not able to account for some of the results obtained in this literature.

We need to consider what other information, in addition to RB and OB, may be available and may have an influence on the compliance outcome. The role of the attitudes related to the request i.e. towards the source of the request, the issue in the request and the task being requested has not been considered by the literature in this area. We argue that these are important determinants of the compliance outcome and the TSC framework should be extended to include them.

Attitude towards the source would be the respondent's evaluation of the person or the organization making the request. The source of a request could be a public oriented citizens' organization, or a profit oriented private company or a friend of the respondent etc. The attitude towards the issue deals with how the respondent evaluates the issue in the request which could be traffic safety or electricity conservation, or the effect of television on children etc. The attitude towards the task deals with the task being requested e.g. a 15 minute interview, or donation of money or escorting teenagers to the zoo etc.

In this paper we will refer to the combined effect of these three attitudes relevant to the request as Attitude towards the Request (AR). If one or more of these attitudes relevant to the request is very accessible then AR would be highly accessible. If at least one of these attitudes is moderately accessible then AR would be termed as moderately accessible, and only when all of these are low in accessibility will AR be considered as falling on the nonattitude end of the attitude-nonattitude continuum discussed in the following section.

The Attitude-Nonattitude continuum:

We could view the attitude towards the request (AR) in terms of an attitude-nonattitude continuum C one that focuses on the accessibility of the attitude from memory (Fazio 1986, 1989). At the lower end of the continuum is the nonattitude which implies that no a priori evaluation of the attitude object exists in memory. As we move along the continuum, an evaluation does exist and the strength of the association between that evaluation and the object and, hence, the chronic accessibility of the attitude, increases. In the case of a weak association, the attitude can be retrieved via an effortful, reflective process but is not capable of automatic activation. At the upper end of the continuum is a well-learned, strong association that is likely to be activated automatically upon mere observation or mention of the attitude object.

The effectiveness of persuasion strategies may be limited to attitudes that occupy a relatively low position along the attitude-nonattitude continuum (Fazio 1989; Wood, Kallgren, & Priesler 1985). More accessible attitudes are more likely to be activated upon mere mention or observation of the attitude object. These attitudes are apt to serve as a filter through which the available information is viewed, and hence are more likely to influence behavior (Fazio, 1986). If the relevant association is too weak to be activated, then behavior will follow from a definition of the event that is not attitudinally based. The behavior may be determined by whatever features of the situation and the attitude object are sufficiently salient to influence the individual's immediate perceptions. Therefore, persuasion strategies are more likely to be effective in the latter scenario.

We propose that both the availability and valence of AR will influence a person's compliance to the target request. But, the nature of this influence will depend on where the AR is located on the attitude-nonattitude continuum.

Scenario 1: High Accessibility of Attitude towards the Request

When the attitude towards the request is located at the higher end of the attitude-nonattitude continuum, then it is very likely that the AR will guide the compliance behavior, and leave very little or no scope for the multiple request techniques to work. The high accessibility of AR will automatically activate it upon mention of the request. This highly accessible AR, would then, instead of RB and OB, be more likely to guide the subject's response to the target request. In the case where the AR is positive and very accessible, it is very likely that the target request will be accepted, with or without the multiple request techniques being used. However, if the AR is negative and highly accessible, then it is likely that this attitude will influence behavior and result in the rejection of the critical request. This would also hold for scenarios in which the AR is not chronically accessible, but is made accessible due to some situational manipulations. Therefore, our first postulate is as follows:

Postulate 1: The multiple request techniques will not be effective in inducing compliance when the attitude of the respondent towards the request is highly accessible. The AR will, in this case, guide the compliance behavior.

Several of the FITD and DITF studies have failed to find any significant effects (see Beaman et al. 1983; and Fern et al. 1986). One reason for this could be the high accessibility of the AR which would dominate the RB and OB. A positive and highly accessible AR would be very likely to guide the compliance behavior and push up the compliance level to its ceiling. We will discuss below some published studies from the literature that lend support to our hypothesis.

Kilbourne (1988) anticipating that social norms of friendship may result in atypical FITD effects (without hypothesizing the direction of these effects), used two different types of sources for the same FITD requests: friends and strangers. According to our framework, the respondent's attitude towards a friend is expected to be positive and accessible and hence guide the compliance behavior, whereas towards a stranger it would tend to lie closer to the non-attitude part of the continuum. Therefore, we would not expect a FITD effect in the condition where a friend was making the request, because the control group compliance would be pushed up to a ceiling level. But in the case where the requestor was a stranger, a FITD effect would be expected. Kilbourne's results confirm our hypothesis C when the FITD experimental and control conditions comprised of friends as requestors, there was no FITD effect since the control group compliance was very high, but a reliable effect was obtained in the stranger-requestor condition.

Mowen and Cialdini (1980) tested several ways of increasing target request compliance in the DITF context. One of their manipulations comprised emphasizing to the subjects, how much their compliance with the request would "help out" the company. The authors anticipated this manipulation would enhance the legitimacy of the request and hence the compliance with DITF. However, our framework would predict results in the opposite direction. In this study (survey about home and dorm safety), the AR is likely to be positive. The "help" emphasis or the stress on the altruistic aspects of the request in this manipulation may increase the accessibility of the positive AR which may now guide the compliance levels. This would render the DITF manipulation ineffective. The results of this study are consistent with our expectations. The data indicate that when help was emphasized, control group compliance reached a ceiling level, wiping out the DITF effect.

Scenario 2:Moderate Accessibility of Attitude towards the Request

In this situation, the attitude towards the request is not highly accessible, and therefore it is not automatically activated upon presentation of the request. It appears reasonable to assume that a moderately accessible attitude is not very likely to have an extreme valence, since a number of researchers have reported a significant relationship between attitude accessibility and attitude extremity (Berger and Mitchell 1989; Fazio and Williams 1986; Powell and Fazio 1984). The compliance with the target request will then be determined by an interaction between the AR and the situational variables (i.e. RB and OB) and their relative accessibility and diagnosticity. Much of the published research in the multiple request techniques corresponds to this scenario. For instance a request for information by a new radio station would be considered as falling in this scenario.

Foot-In-the-Door: When the respondent is confronted with the second request, the OB from past acceptance of the first request is expected to be positive and the RB, since it is an escalation, is expected to be negative. The subject would have consented to the first request mainly due to it's small size.

As we have already discussed in the previous section, when one of the three components of the model (i.e. AR, OB or RB) is very accessible it will dominate the other components and determine the compliance outcome. In the FITD scenario, past compliance with the initial request is likely to make OB accessible. This is because OB deals with own behavior and reactions which are related to the self, a highly accessible schema (Rogers, Kuiper and Kirker 1977). This would be more so when the respondent not only agreed to the request but performed it, since direct experience (Fazio and Zanna 1981) and elaboration (Petty and Cacioppo 1986) are expected to increase the accessibility of the response information. Thus, in this case, the accessibility of the OB will dominate the AR, which is only moderately accessible. So, FITD may even be effective with requests in which the AR is neutrally or negatively valenced, since there is a low likelihood of it being retrieved and considered while responding to the target request. This leads us to our next postulate:

Postulate 2: When the accessibility of Attitude towards the Request (AR) is moderate then FITD is not likely to be very sensitive to the valence of AR.

Door-In-the-Face: In this case the first request is rejected due to its large size. Subsequently, a concession is made on the size dimension. The respondent then reconsiders the request which now seems more reasonable in size due to it's contrast with the previous request. Here OB is not likely to dominate the short-term memory since a behavior was not performed in response to the first request, and hence unlike the FITD scenario the OB information is not highly accessible. Since the respondent does not have a past experience of compliance easily accessible, and needs some relevant information to make a decision, he/she is likely to use the AR, which is relatively more accessible than the OB. Pressure from the concessionary RB would lead the subject to comply, given that the valence of the AR is positive or neutral. However, if the AR is negatively valenced, the respondent is likely to refuse the request.

Postulate 3: When the accessibility of Attitude towards the Request (AR) is moderate then DITF is likely to be sensitive to the valence of AR.

Thus, in the DITF condition, the AR is considered when making a compliance decision, because OB is not accessible. However in the FITD condition, AR will most likely not come into the picture, due to the dominance of the positive OB. This difference in the relative availability of OB in the two techniques, leads to their differential sensitivity to the valence of the AR. The need for the source and issue to be positive is more crucial for the success of the DITF than the FITD. This reasoning implies that in a business context, there would be a greater likelihood for the FITD to work than DITF.

The literature in this area lends support to our proposed postulates. FITD has been shown to work in cases where the AR was not positive and the compliance of the control group was quite low (Dillard et al. 1984; Katzev and Johnson 1984), and it is expected to be successful across a broader spectrum of requests than the DITF (Fern et al. 1986). But DITF appears to be successful mainly in pro-social contexts. Foehl and Goldman (1983) proposed and found evidence for the Worthy Person Hypothesis. According to it, DITF is likely to be effective only when the source is positive or "worthy" and/or the issue in the request is a pro-social or worthy one. For instance, a charity organization, or a cause benefiting the public.

Patch (1986) hypothesized that since FITD promotes pressure toward internal consistency and DITF puts normative pressure on the respondent, source legitimacy would be more likely to influence the success of the latter than the former. He tested the FITD and the DITF under conditions of low and high source legitimacy. In the critical request subjects were asked to respond to a survey about television viewing preferences. The source of the request was manipulated as either a public interest group C Parents for Good Television Programming (a source which is likely to be rated as positive by the subjects but not likely to be very highly accessible for most of the respondents), or a private consulting firm (a profit oriented organization, towards which the attitude of the respondents is not likely to be positive or accessible). For each condition, the source was constant across the requests. We would predict that FITD would be successful with both the sources because the valence of AR is not likely to be important for its success. This is because with FITD, OB should dominate this relatively inaccessible AR. But the DITF would only be likely to be effective when the AR is positive. Patch's results are consistent with these predictions C only FITD was effective when the source of the request was the private consulting firm, whereas both the techniques were effective when the organization conducting the survey was the public interest group.

Goldman, Gier and Smith (1981) randomly selected people from a telephone directory and asked them to answer questions for a new radio station C a profit oriented request towards which the attitude of the respondents is not likely to be positive. They found that only FITD worked in this context, and even when they manipulated the difficulty level of the requests, the DITF technique failed to enhance compliance with the target request.

Scenario 3: Low Accessibility of the Attitude towards the Request

Finally, as we move towards the lower end of the continuum, we come to a situation in which either the respondent does not have an attitude towards the request (a non-attitude) or has a very weak attitude which is not easily accessible. In this case, the AR would not be expected to play a role in compliance behavior due to it's extremely low accessibility and somewhat neutral valence. Both the FITD and DITF would be expected to work, provided the other conditions necessary for their success, as proposed by the TSC framework, were satisfied. So, FITD would enhance compliance as long as the "own behavior" was very accessible and "request behavior" was not accessible. DITF would work, when the "request behavior" was more accessible than the "own behavior". This is the scenario in which the basic framework with RB and OB, would guide the compliance behavior without any input from AR. Our final postulate about multiple request effectiveness is as follows:

Postulate 4: When the AR is located closer to the nonattitude side of the continuum, then the relative availability of RB and OB will guide the compliance behavior.


This manuscript argues that when the attitude towards the request is highly accessible or located towards the attitude end of the attitude-nonattitude continuum, AR will guide compliance, and the situational variables in the multiple request techniques i.e. OB and RB will have little role to play.

The second case, where the AR is moderately accessible, is the most interesting one, and perhaps the one that is most often observed in the literature. The relative accessibility and valence of OB, RB and AR determines compliance. In the case of FITD the positive and accessible OB dominates the moderately accessible AR, and therefore this technique is not very sensitive to the valence of the AR. However, in the DITF technique OB is not dominant and AR considered in the decision making situation, making it more sensitive the valence of AR.

The AR does not have any role to play in the last scenario. Thus, as we keep moving towards the nonattitude part of the continuum, AR's role keeps on decreasing and towards the non-attitude end of the continuum, it does not play any role in the compliance process. At this point in the AR continuum, only RB and OB are needed to explain the multiple request effectiveness.

Based on our review, we can make some suggestions about the suitability of these multiple request techniques. When the AR is highly accessible, it may not be very effective to use the multiple request techniques. If the AR is positive, but not very accessible, instead of using two requests, it may be more efficient to increase the accessibility of the AR, and use just one request i.e. if it is possible to increase its accessibility (e.g. Mowen and Cialdini 1980). FITD would be the best technique to pick for a business and/or a non-humanitarian request when the AR of the respondents is likely to be moderately accessible. But, DITF would only be suitable if the respondents are likely to have a positive or neutral attitude towards the request.

Although we have reviewed the results of previous studies and found them to be consistent with the predictions of our framework, these results have not been empirically verified. Further evidence would be provided in support of this framework by demonstrating that the valence and accessibility of the requests in these studies confirms to our assumptions. A stronger test of this framework would be provided by actually manipulating attitude valence and accessibility and examining the effects of these manipulations on compliance using FITD and DITF techniques. It would also be important to measure the accessiblilty and valence of RB and OB, in order to provide support for the entire framework.


Beaman, Arthur L., C. Maureen Cole, Marilyn Preston, Bonnel Klentz and Nancy M. Steblay (1983), "Fifteen Years of the-Foot-in-the-Door Research C A Meta-Analysis", Personality and Social Psychology Bulletin, 9 (2), 181-196.

Beaman, Arthur L., Nancy M. Steblay, Marilyn Preston and Bonnel Klentz (1988), "Compliance as a Function of Elapsed Time Between First and Second Requests", Journal of Social Psychology, 128(2), 233-243.

Bem, Daryl J. (1972), "Self-Perception Theory", in Advances in Experimental Social Psychology, ed. L. Berkowitz, New York: Academic Press, 1-62.

Berger, Ida E. and Andrew Mitchell (1989), "The effect of advertising on attitude accessibility, attitude confidence, and the attitude-behavior relationship", Journal of Consumer Research, 16, December, 269-279.

Cantrill, James G. and David R. Seibold (1986), "The Perceptual Contrast Explanation of Sequential Request Strategy Effectiveness", Human Communication Research, 13 (2), Winter, 253-267.

Cialdini, Robert B., J. E. Vincent, S. K. Lewis, J. Catalan, D. Wheeler and B. L. Darby (1975), "A Reciprocal concessions Procedure for Inducing Compliance", Journal of Personality and Social Psychology, 31, 206-15.

DeJong, William (1979), "An Examination of Self-Perception Mediation of the Foot-in-the-Door Effect", Journal of Personality and Social Psychology, 37 (12), 2221-2239.

Dillard, James P., John E. Hunter and Michael Burgoon (1984), "Sequential Request persuasive StrategiesC Meta-Analysis of Foot-in-the-Door and Door-in-the-Face", Human Communication Research, 10 (4), Summer, 461-488.

Fazio, Russel (1989), "On the power and functionality of attitudes: The role of attitude accessibility", in Attitude Structure and Function, Eds. Pratkanis, Breckler and Greenwald, Hillsdale, NJ: Earlbaum, 153-179.

Fazio, Russel (1986), "How do attitudes guide behavior?", in The Handbook of Motivation and Cognition: Foundations of Social Behavior, Eds. Sorrentino & Higgins, New York: Guilford Press, 204-243.

Fazio, Russel and Carol Williams (1986), "Attitude accessibility as a moderator of the attitude-perception and attitude-behavior relations: An investigation of the 1984 presidential election", Journal of Personality and Social Psychology, 51, 505-514.

Fazio, Russel and Mark P. Zanna (1981), "Direct experience and attitude-behavior consistency", Advances in Experimental Social Psychology, 14, 161-202.

Fern, Edward F., Kent B. Monroe and Ramon Avila (1986), "Effectiveness of Multiple Request Strategies: A Synthesis of Research Results", Journal of Marketing Research, XXIII, 144-52.

Foehl, J. C. and Morton Goldman (1983), "Increasing Altruistic Behavior by using compliance techniques", Journal of Social Psychology, 119, 21-29.

Freedman J. L. and S. C. Fraser (1966), "Compliance Without Pressure: The Foot-in-the-Door Technique", Journal of Personality and Social Psychology, 4, 195-202.

Goldman, Morton, Joseph Gier, and D. E. Smith (1981), "Compliance as Affected by Task Difficulty and Order of Tasks", Journal of Social Psychology, 114, 75-83.

Goldman, Morton, James F. McVeigh and J. L. Richterkessing (1984), "Door-in-the-Face Procedure: Reciprocal Concession, Perceptual Contrast, or Worthy Person", Journal of Social Psychology, 123 (2), August, 245-251.

Katzev, Richard D. and Theodore Johnson (1984), "Comparing the effects of monetary incentives and foot-in-the-door strategies in promoting residential electricity consumption", Journal of Applied Social Psychology, 14(1), 12-27.

Kilbourne, Brock K. (1988), "A New Application of the Foot-in-the-Door Technique: Friend or Stranger?", Psychological Reports, 62, 31-36.

Miller, Richard, Clive Seligman and Malcolm Bush (1976), "Perceptual contrast versus Reciprocal Concession as Mediators of Induce Compliance", Canadian Journal of Behavioral Science, 8, 401-409.

Mowen, John C. and Robert B. Cialdini (1980), "On Implementing the Door-in-the-Face Compliance Technique in a Business Context", Journal of Marketing Research, XVII, 253-58.

Patch, Michael (1988), "Differential perception of source legitimacy in sequential request strategies", Journal of Social Psychology, 128 (6), 817-823.

Patch, Michael (1986), "The role of source legitimacy in sequential request strategies of compliance", Personality and Social Psychology Bulletin, 12 (2), 199-205.

Petty, Richard and John Caccioppo (1986), "Communication and persuasion C central and peripheral routes to attitude change", NY: Springer-Verlag.

Powell, Martha and Russell Fazio (1984), "Attitude accessibility as a function of repeated attitudinal expression", Personality and Social Psychology Bulletin, 10, 139-148.

Reingen, Peter H. and Jerome B. Kernan (1979), "More evidence on interpersonal yielding", Journal of Marketing Research, 26(November), 365-69.

Rogers, T. B., N. A. Kuiper and W. S. Kirker (1977), "Self-reference and the encoding of personal information", Journal of Personality and Social Psychology, 35(9), 677-688.

Tybout, Alice M., Brian Sternthal and Bobby J. Calder (1983), "Information Availability as a Determinant of Multiple Request Effectiveness", Journal of Marketing Research, XX, (August), 280-90.

Wood, Wendy, Carl A. Kallgren, R. M. Preisler (1985), "Access to attitude-relevant information in memory as a determinant of persuasion: The role of message attributes", Journal of Experimental Social Psychology, 21, 73-85.