Science Communication and the Rationality of Public Opinion Formation

Abstract

Because research indicates that people’s value orientations significantly affect their opinions about advanced technologies some risk scholars argue that technology policy should reflect the recommendations of experts rather than the opinions of a value-driven public. This suggested approach might be bolstered to the extent that people use substantive (value-driven) rather than formal (reason-driven) rationality when assessing the recommendations of experts. We found evidence that people’s opinions are indeed influenced strongly by their substantive rationality. We nevertheless argue that technology policy formation should rely more rather than less upon public opinions, thereby encouraging further efforts at improving science communication theory and practice.

Keywords

consumer trust recreancy theorem substantive rationality U.S. food system Cultural Cognition Project

Introduction

If people rely primarily upon their value orientations to evaluate the recommendations of experts, perhaps in the process hindering the rational development of beneficial technologies used in societal systems, then to what extent should a well-meaning but relatively uninformed public participate in developing technology-related policies? This question has been central to contemporary debates about the extent to which technology-related policy should be guided by expert rather than public opinion (e.g., Cultural Cognition Project, 2011; Sunstein, 2005, 2006). This debate might be further intensified to the extent that people use substantive (i.e., value-driven) rather than formal (i.e., reason-driven) rationality in forming their opinions about expert recommendations. That is, if people tend to rely upon their value orientations rather than their formal reasoning to evaluate the assessments of experts, then does not this finding support claims that public opinions should be secondary to expert opinions in forming reason-based, technology-related policy? This article explored the extent to which people express substantive rationality in evaluating expert recommendations regarding the U.S. food system, where agricultural production policy formation represents an arena of intense controversy regarding the role of public opinion. The results indicate that people indeed rely strongly upon substantive rather than formal rationality in evaluating experts’ recommendations. Although these results seemingly support arguments for less emphasis on science communication in its efforts to encourage public participation in technology policy formation, we argue that the findings support increasing such efforts.

We begin by describing social science perspectives on rationality. Next, we establish a theoretical foundation for empirically testing for expressions of formal and substantive rationality. We then analyze data gathered from two nationwide surveys of U.S. consumers to explore the extent to which they utilize substantive rationality in evaluating expert opinions about the U.S. food system. Next, we briefly summarize the current debate about the extent to which the public should be involved in guiding technology policy. Then, we describe how our findings—which indicate that people rely strongly upon substantive rationality—inform this debate and support increased efforts at improving science communication.

Social Science Perspectives on Rationality

Weber’s (1921/1968) treatises on rationality distinguish between zweckrational motivations—ones aimed at attaining “rationally pursued and calculated ends” (pp. 24-26)—and wertrational motivations—ones guided by a “conscious belief in [a] value [orientation] for its own sake” (p. 24). To Weber, wertrational expressions have importance to the actor “independently of [their] prospects of success” (p. 24). Wertrational expressions thereby are neither irrational nor nonrational but instead are “induced by immanent or transcendental (as opposed to instrumental) values and thus by intrinsic rather than extrinsic motivation” (Zafirovski, 2005, p. 91). In distinguishing between zweckrational (i.e., formal) and wertrational (i.e., substantive) motivations, Weber posed a complementary ontological position to that offered in other social sciences, particularly to that offered in economics as the foundation of rational choice theory (e.g., Boudon, 1981; Coleman, 1990; Elster, 1989; Hechter & Kanazawa, 1997; Sen, 1977; Simon, 1982). Rational choice theory posits that individuals in their decision making seek to maximize their benefits and minimize their costs related to some form of goal attainment (e.g., Sen, 1977). “Thin” forms of rational choice assume that economic benefits define the first-order ends and motives for goal attainment, whereas “thick” forms of rational choice broaden the definition of potential ends to include noneconomic goals, such as prestige, power, influence, and the like, including expressions of value orientations (Elster, 1989; Hechter & Kanazawa, 1997). In either its thin or its thick versions, the essential element of rational choice is that behavior in some manner is oriented toward achieving an extrinsically identifiable goal, whether this goal be defined by the maximization of economic or noneconomic utility (Boudon, 1981; Sen, 1977).

As might be expected, the teleological determination inherent within thick rational choice theory, wherein it is assumed that the actor is pursuing “some goal,” in addition to its assumptions about full knowledge, intentionality, and transitivity have engendered debate within the social sciences regarding its falsifiability and eventual pragmatic usefulness (e.g., Ackerman, 1997; Sen, 1977; Smelser, 1992). We pass on this debate to explore instead the distinction drawn by Weber (1921/1968) between rational behavior oriented toward extrinsically defined goals, whether these goals be economic or noneconomic ones (i.e., formal rationality), and rational behavior reflecting intrinsic self-expressions of values that present merits to the actor in and of themselves, ones that have importance to the actor independently of their prospects of success (i.e., substantive rationality). Weber’s typology has proved valuable to social scientists in understanding human motivation and in informing approaches to incorporating such motivations within public policy formation (Ritzer, 2010). In considering sound public policy formation, for example, policy makers might wonder about the extent to which public opinions offered only for the sake of expressing value-based orientations (i.e., an expression of substantive rationality) should be incorporated into policy deliberations. Hence, from an applied perspective, empirically validating this distinction might highlight the significance of such expressions as a mechanism by which people support experts’ recommendations (e.g., Kahan & Slovic, 2006; Kahan, Slovic, Braman, & Gastil, 2006; Sunstein, 2005, 2006). And conceptually, such an evaluation would support Weber’s (1921/1968) typological distinction between zweckrational and wertrational motivations. We wondered, therefore, about whether expressions of values for the purpose of obtaining an extrinsically identifiable goal might empirically be differentiated from expressions of values for their own sake.

Such an empirical investigation requires identifying expressions that (a) seemingly do not reflect zweckrational, or formal rationality, (b) reflect expressions of value orientations, and (c) reflect value orientations for their own sake independently of their prospects of success. This third criterion is critical to an empirical investigation of zweckrational and wertrational behavior because from the perspective of thick rational choice theory, zweckrational behavior includes expressions of values if such expressions are intended by the actor to further goal attainment. That is, one might express values in a manner that seemingly cannot be identified as instrumentally logical but might intentionally or unintentionally ignore logical inconsistency for the purpose of achieving a broader or longer term purpose of goal attainment. Explorations of individuals’ choices within decision theory (e.g., Anand, 1993) and prospects theory (e.g., Kahneman & Tversky, 1979), for example, indicate that actors sometimes break with logically consistent and transitive rules because they hold the prospect that by doing so they will attain an extrinsically identifiable goal. An empirical indication of substantive rationality therefore, must meet all three criteria specified above.

To be certain, such an evaluation cannot exhaust all possible alternative prospects of success an actor might be contemplating. So attempts to empirically evaluate substantive rationality will always be subject to claims that the prospect has not been identified or cannot be achieved due to the conditions of either the actor’s limited knowledge and imagination or external structural constraints (see discussions of bounded rationality; e.g., Jones, 1999; Simon, 1982)). Nevertheless, an investigation of possible expressions of substantive rationality might be undertaken within the bounds of a clearly identified external goal to determine if this goal seems to serve as a prospect for success that can explain seemingly nonrational behavior.

Theoretical Approach to Assessing Rationality

The Recreancy Theorem

Assessing expressions of substantive rationality requires a theoretical framework by which to logically identify such expressions. The recreancy theorem (Sapp et al., 2009), with some conceptual modifications as described below, meets this need. Furthermore, inasmuch as the theorem addresses public confidence in representatives of societal institutions it provides a valid framework for illuminating the importance of science communication within an arena of sometimes conflicting value orientations about public policy. While recognizing the achievements within psychometrics (e.g., Slovic, 2000), anthropology (e.g., Douglas & Wildavsky, 1982), sociology (e.g., Rogers, 1995), economics (e.g., Gollier, 2004), and social psychology (e.g., Fishbein & Ajzen, 1975) regarding public support for existing and alternative institutional arrangements, Freudenburg and colleagues (Alario & Freudenburg, 2003; Freudenburg, 1993; Freudenburg & Pastor, 1992; Rosa & Freudenburg, 1984) note several limitations they deem worthy of further conceptualization and empirical examination. These scholars assert that most approaches to understanding public perceptions of policy focus upon attributes of individual perceivers without paying sufficient attention to the actions taken by institutions responsible for policy development and management. Freudenburg (1993) expresses concern about these limitations in light of indications that within the past few decades there has been a dramatic increase in the potential for recreancy—“the failure of institutional actors to carry out their responsibilities with the degree of vigor necessary to merit the societal trust they enjoy” (p. 916). Related to this concern are misgivings about the language used by some risk assessors in classifying as irrational the skepticism expressed by citizens regarding public policy. For example, whereas some (e.g., Cohen, 1985; Starr, 1969) consider public rejection of new technologies deemed as sufficiently safe based on quantitative risk assessments to be irrational, Freudenburg et al. point out the inherent rationality of expressing skepticism toward societal institutions that too often neglect their obligations to conduct quantitative risk assessments in a scientifically objective manner, adequately inform the public about known risks, and manage risks with due regard to public safety. Drawing principally upon the work of Barber (1983), Freudenburg (1993) posits that institutional representatives require competence—the skills, abilities, and experience needed to successfully complete a task—and fiduciary responsibility—the integrity required to do the right thing. Freudenburg et al. thereby propose that assessments of consumers’ trust in and support for institutional arrangements should focus upon their assessments of these attributes of institutional actors. In accordance with this conceptual approach, Freudenburg (1993) asserts that citizens’ evaluations of institutional recreancy—and their related assessments of the competence and fiduciary responsibility of societal institutions—reflect elements of both formal rationality and substantive rationality. Freudenburg reasons that theoretical approaches to understanding public perceptions of recreancy that presume that these perceptions are based upon expressions of formal rationality only conceptually cannot account for the extent to which such perceptions are based upon expressions of value orientations.

Subsequently, Sapp and colleagues (Sapp & Downing-Matibag, 2009; Sapp et al., 2009) relied upon this perspective to posit the recreancy theorem as an approach to understanding public support for institutional arrangements, wherein this support presumably is based primarily upon individuals’ judgments of the competency and fiduciary responsibility of institutional actors. Quantitative investigations of support for selected institutional actors across five areas of the U.S. food system (Sapp et al., 2009) and public acceptance of irradiated food (Sapp & Downing-Matibag, 2009) found empirical support for the recreancy theorem. That is, these investigations found that public perceptions of the competency and fiduciary responsibility of institutional representatives explained a significant amount of the variance in trust in and support for them, whereas individual attributes such as social-demographic characteristics and perceived risk had only modest influences on trust and support. Finding empirical support for hypotheses specified by the recreancy theorem, however, does not imply support for the theoretical propositions from which these hypotheses are derived. Such reasoning represents the fallacy of affirming the consequent, or believing that a set of observed relationships necessarily lends support to the propositions used to specify these relationships (e.g., Sarkar & Pfeifer, 2006). The conceptual foundations of the recreancy theorem infer that people will exhibit both formal and substantive rationality in their evaluations of societal institutions. Therefore, to improve social science understanding of citizens’ trust in and support for societal institutions, we explored within the context of the recreancy theorem the extent to which expressions of formal rationality and substantive rationality might be identified in evaluations of these institutions. Earle and colleagues (Earle, 2010; Earle & Cvetkovich, 1995; Earle & Siegrist, 2006; Earle, Siegrist, & Gutscher, 2007; Siegrist, Earle, & Gutscher, 2003) describe theoretical linkages between rationality and trust that inform possible procedures for empirically examining expressions of formal and substantive rationality within the context of the recreancy theorem.

Rationality and Trust

Earle and colleagues posit the trust, confidence, and cooperation (TCC) model as an approach to understanding individuals’ expressions of trust in others. In relying principally upon the work of Luhmann (1979), they distinguish between within-group trust—exemplified by trust based either on limited information (i.e., “social trust”) or on repeated interaction (i.e., “interpersonal trust”)—and across-group trust (i.e., trust among strangers, or “general trust”). In contrast, Sapp et al. (2009) in relying upon the work of Luhmann (1979) and Barber (1983), refer to trust based upon repeated interaction as relational trust (e.g., “I trust my grocer”) and interpret the term interpersonal trust to mean “the perceived bond between a specific agent and the trustee” (e.g., “I trust grocers”; Sapp et al., 2009, p. 528). Thus, the TCC model, which addresses the determinants of social trust, and the recreancy theorem seem to refer to a similar expression of vulnerability—that is, one deferred to societal agents for whom the actor has limited information. Earle et al. (2007) define confidence “as the belief, based on experience or evidence, that certain future events will occur as expected” (p. 4). The TCC model then posits that the actor’s perceived performance of the agent to be trusted will have a direct effect on the evaluations of confidence. The recreancy theorem, in contrast, distinguishes between two beliefs that might influence actors to assume that certain future events will occur as expected: perceived competence—evaluations of the skills, abilities, and expertise of the agent to be trusted (i.e., can the agent successfully complete a desired task)—and perceived fiduciary responsibility—evaluations of whether the agent has the integrity to do the right thing (i.e., will the agent successfully complete a desired task). Thus, both the TCC model and the recreancy theorem focus upon expressed vulnerability to within-group agents for whom the actor has limited information and specify that actors’ performance evaluations of agents will affect their expressed trust in them. One noteworthy difference between the two approaches is the significance placed within the TCC on shared values. Earle et al. assert that perceived shared values represent an essential theoretical link between individuals’ general trust and their expressions of social trust in within-group agents. In that the TCC model specifies the critical nature of perceived shared values on evaluations of agents, it emphasizes a theoretical element that might influence evaluations of both competence and fiduciary responsibility within the recreancy theorem. It is this insight into the determinants of trust that lead us to posit an extension to the recreancy theorem, one posited for the purpose of exploring expressions of formal and substantive rationality within it.

The expanded recreancy theorem (Figure 1) specifies that perceived competence and fiduciary responsibility reflect the actor’s sense of shared values with the agent to be trusted (with respect to pursuing a particular agricultural production goal). In this sense, the expanded theorem incorporates insights offered by Earle and colleagues regarding actors’ trust in and support for institutional arrangements. Previously unspecified by Earle and colleagues but informative here is a consideration of the conceptual meanings of perceived shared values on judgments about both competence and fiduciary responsibility. The hypothesized relationship between perceived shared values and evaluations of fiduciary responsibility seems to conform with the description of formal rationality. That is, it seems like a logical means-ends relationship to assume that a perception of shared values with an agent would influence the actor to believe that the agent will “do the right thing.” On the other hand, the hypothesized relationship between perceived shared values and evaluations of competence does not seem to meet the requirements of formal rationality because a sense of shared values seems to provide neither a necessary nor a sufficient condition to infer competence. For example, Stephanie might believe that Natalie shares the same values as her but does not have the competency to successfully carry out her task-related responsibilities. Conversely, Stephanie might believe that Natalie does not share the same values as her but recognizes that Natalie nevertheless is highly competent at carrying out her task-related responsibilities. The specified path between shared values and competency, therefore, might reflect some type of rationality but inasmuch as it does not represent a logical connection between means and ends cannot in itself represent formal rationality. Nor is it likely that such an expression be deemed as irrational, at least within the scope of contemporary social science definitions of rationality (e.g., Zafirovski, 2005).

Figure 1.

The expanded recreancy theorem.

The question becomes, then, does this seemingly nonrational expression of values represent an extension of rational choice in that it is directed toward goal attainment (i.e., thick rational choice) or does it reflect an expression of values in themselves (i.e., substantive rationality)? Thick forms of rational choice theory might consider this relationship as an expression of formal rationality in that it reflects attributed competence for the prospects of attaining an extrinsically identifiable goal. For example, although Stephanie might believe that Natalie does not have the competence necessary to successfully complete a task, she nevertheless might attribute competence to her as a means of encouraging her to attain an extrinsically identifiable goal (i.e., “I desire for Natalie to be successful”). Alternatively, this expression might reflect substantive rationality in that it represents an assertion of values in themselves without regard for the prospects of success. For example, although Stephanie might believe that Natalie does not have the competence necessary to successfully complete a task, she nevertheless might attribute competence to her as a means of expressing her value orientation regardless of whether the task is successfully completed (i.e., “I wish to express my support for Natalie’s values”). We propose that empirically evaluating this distinction is an important undertaking as a means of gaining further understanding of the significance placed upon perceived shared values in people’s support for experts’ recommendations.

An Examination of Rationality Within the Recreancy Theorem

Samples

The data were collected in two nationwide surveys of adults, ages 21 to 65 years, conducted in 2011. Survey Sampling International provided access to their Internet panels, consisting of adults recruited through web-related contacts. The response rates for each survey equaled approximately 25% Although approximately 70% of homes in the United States have access to the Internet (U.S. Bureau of the Census, 2010), given that less affluent homes are less likely to have an Internet connection, one should take caution in generalizing to the population the descriptive results observed in these samples. To some extent, the degree to which social class might affect our investigation of expressions of formal and substantive rationality within the recreancy theorem can be evaluated with respect to how strongly measures of social class affect trust in and support for the U.S. food system.

Respondents in Survey 1 were asked for their opinions on the model variables with respect to the activities of “family” farmers, and respondents in Survey 2 were asked for their opinions about the activities of “commercial” farmers. Family farming was defined for respondents as follows: “This is a farming operation that is owned and operated by a family. All decisions on how to operate this farm are made by the family members and carried out by family members or employees.” Commercial farming was described as follows: “This is a farming operation that is owned by a company and operated by employee farmers. All decisions on how to operate this farm are made by managers of the company and carried out by employees.” Such definitions, as would others, present some ambiguity with regard to what is meant by the distinction between family and commercial farming. After all, much of family farming is conducted within the context of a family-owned corporation. Distinctions that might be drawn between the two types of farming are largely immaterial here inasmuch as we focus on expressions of rationality with regard to evaluations of representatives of them. Nevertheless, these definitions seem to have been sufficient to create statistically significant differences in expressed opinions about the U.S. food system across the two surveys, differences that might reflect the propositions advanced by Earle et al. (2007) regarding how cultural context can affect the interpretation of theories of rational expectations such as the recreancy theorem.

Measurement of Model Variables

Support for the recommendations of farmers was measured by responses to the following request: “Please tell us … the extent to which you are willing to support the actions of [family/commercial] farmers who [agricultural goal].” As in the measurement of all model variables, responses were recorded on an 11-point scale, where 0 = low (support, trust, etc.) and 10 = high (support, trust, etc.). We described trust as “a combination of how competent you believe farmers are (knowledge, skills, and experience) and how confident you are in relying on them to do the right thing.” Trust was measured by responses to the following request: “Please tell us how much trust you have in farmers who [agricultural production goal].” Perceived competence was measured by responses to the following request: “Tell us the degree to which you believe farmers have the competence to achieve a specific goal. Do they have the knowledge, skills, and experience to do a good job?” We defined fiduciary responsibility (i.e., confidence) as “doing the right thing.” To measure fiduciary responsibility, we asked respondents to tell us how much confidence they had in farmers who (agricultural production goal). Perception of shared values was measured by responses to the following request: “Tell us the degree to which you believe farmers share the same values as you do” regarding [agricultural production goal]. Conceptually, it is important to account for variance in perceived shared values that might be related to support for the agricultural goal being considered. Therefore, our model estimations included a statistical control for the correlation between perceived shared values and perceived competency.

We included in the model the following characteristics of respondents as statistical controls for their expressions of trust in and support for the goal orientations of farmers: age (in years), sex (0 = male, 1 = female), years of formal education (<high school, high school, vocational or technical school, undergraduate college degree, master’s degree, PhD degree or other doctorate degree), and income (<$25,000, $25,000-$49,999, $50,000-$74,999, $75,000-$99,999, $100,000-$249,999, and ≥$250,000). Given that psychometric research has identified perceived risk as a key determinant of trust and support (e.g., Frewer, Howard, Hedderley, & Shepherd, 1996; Slovic, 2000), we included a measure of concern about agricultural production goals. We asked respondents to “tell us how concerned they were about”: the rising cost of food, humane treatment of animals, having enough food to feed the world, environmental sustainability in farming, and food safety.

Procedure

Our procedure required that we know the extent to which respondents supported selected goals of production agriculture. Thus, in each survey we asked respondents to assign a total of 100 points to express their opinions about the extent to which farmers should pursue eight agricultural production goals: grow affordable food for me and my family, assure humane treatment of farm animals, grow enough food to help feed the world, assure environmental sustainability on the farm, assure food is grown safely, assure food is nutritious, assure farm profitability, and maximize farm productivity. We reasoned that one approach to empirically differentiating expressions of formal from substantive rationality within the context of the expanded recreancy theorem would be to examine the moderating effects of goal orientation on the relationship between perceived shared values and evaluations of competency. One might presume, for instance, that persons who strongly support, for example, environmental sustainability—as measured by the strength of their desire to attain this goal—would be more likely to attribute competence perhaps even without a sense of shared values. It is within this understanding that we assessed the extent to which formal and substantive rationality might be expressed within the recreancy theorem. Of course, such an evaluation will leave much to debate given the many alternative goals that might have been pursued by respondents in their evaluations of shared values and competency. This indeterminacy cannot be avoided in empirical evaluations of thick rational choice (e.g., Sen, 1977). We will, however, be able to examine expressions of formal and substantive rationality with respect to goals clearly identified to the survey respondents and to which their responses were directed. Furthermore, we believe that apart from the extent to which such an evaluation supports an empirical validation of expressed substantive rationality, it can reveal the importance of perceived shared values in individuals’ expressions of trust in and support for the U.S. food system.

Our procedure was to use structural equation modeling (Jöreskog & Sörbom, 1995) to estimate the causal relationships posited by the expanded recreancy theorem for each of the five agricultural production goals: (i.e., affordable food for me and my family, animal welfare, environmental sustainability, grow enough food to feed the world, and safe food). We explored, first, the extent to which the hypotheses specified by the recreancy theorem were supported by the data. Next, we investigated whether the specified relationships between perceived shared values and perceived competence and fiduciary responsibility were supported by the data. Next, we reasoned that finding a statistically significant parameter estimate for the relationship between perceived shared values and perceived competence that was significantly moderated by the actor’s degree of interest in attaining the goal under consideration would represent an expression of formal rationality. That is, such a finding would support the proposition that actors were expressing their values as encouragement for the prospects of success. We reasoned further that finding a statistically significant parameter estimate for the specified relationship between perceived shared values and perceived competence that was not significantly moderated by the actor’s degree of interest in attaining the goal under consideration would represent an expression of substantive rationality. This assertion rests upon the assumption that the latter finding would indicate an expression of values without regard to the prospects of success in attaining the goal under consideration. Thus, our central hypothesis is that the relationship between shared values and perceived competency will be moderated by the extent to which one supports the agricultural production goal under consideration (i.e., which would reflect an expression of formal rationality). A rejection of this hypothesis in favor of the null presumably would provide an indication of an expression of substantive rationality.

Results

The median category for respondents’ age was 35 to 44 years, with the distribution leaning toward older age categories, with approximately 25% of respondents being 55 to 65 years old (Table 1). Approximately 52% of respondents were males. The median category for respondents’ education was “vocational/technical.” Approximately 48% of the survey respondents had earned a college degree or higher, thereby reflecting a sample with higher educational attainment than that of the U.S. population, where about 27% report earning a college degree or higher (U.S. Census Bureau, 2007). For both samples, the median category for total household income before taxes in 2010 was $50,000 to $74,999, which corresponds with the 2009 median income category for U.S. households (U.S. Census Bureau, 2007).

Table 1.

Descriptive Statistics for the Two Samples.

	Sample assessing family farmers (n = 1,000)		Sample assessing commercial farmers (n = 1,004)
Variable	Frequency	%	Frequency	%
Age (years)
21-25	195	19.50	164	16.33
26-34	199	19.90	203	20.22
35-44	155	15.50	164	16.33
45-54	263	26.30	227	22.61
55-65	188	18.80	246	24.50
Gender
Male	487	48.70	523	52.09
Female	513	51.30	481	47.91
Education
<High school	19	1.90	9	0.90
High school	284	28.40	321	31.97
Vocational/technical	202	20.20	194	19.32
College degree	375	37.50	369	36.75
Master’s degree	96	9.60	97	9.66
PhD or other doctorate	24	2.40	14	1.39
Income (US$)
<25,000	207	20.70	215	21.41
25,000-49,999	292	29.20	282	28.09
50,000-74,999	278	27.80	281	27.99
75,000-99,999	109	10.90	106	10.56
100,000-249,999	104	10.40	112	11.16
≥250,000	10	1.00	8	0.80

The mean values for all model variables for both samples were above the response scale midpoint of 5.0 (figures available on request). The mean evaluation of the competency of family farmers to “feed the world” equaled just 5.123, whereas the mean evaluation for support of family farmers to produce safe food equaled 8.418. We found that the evaluations of commercial farmers with just three exceptions were for every topic and model variable statistically different from those observed for evaluations of family farmers. In some cases the evaluations for commercial farmers were higher and in some cases lower than the evaluations of family farmers. The numerical differences in evaluations typically were small ones, wherein the indication of statistical significance in part reflects the large sample sizes for the two surveys.

We estimated the model depicted in Figure 1 using the data gathered from both samples for those topics for which we had full information for potential goals (i.e., affordable food for me and my family, animal welfare, environmental sustainability, grow enough food to feed the world, and safe food). These 10 evaluations of the recreancy theorem showed support for its propositions (Table 2). That is, the explained variance in trust and support, the size and statistical significance of the standardized effects of perceived competence and fiduciary responsibility on trust, and the statistics regarding overall model fit indicated that the model has adequate explanatory power and construct validity to serve as a theoretical approach to understanding consumer trust in the U.S. food system. Across the ten evaluations, the R² values indicating explained variance in trust ranged from .62 to .69. The R² values for explained variance in support ranged from .40 to .55 regarding the recommendations of family farmers and from .24 to .43 regarding the recommendations of commercial farmers. As has been noted in previous evaluations of the recreancy theorem (Sapp et al., 2009), the effect of perceived fiduciary responsibility on trust outweighed that of perceived competence on trust—across the 10 models tested here—by an average margin of approximately 3 to 1.

Table 2.

Standardized Parameter Estimates and Model Statistics for the 10 Evaluations of the Recreancy Theorem.

	Affordable food		Animal welfare		Environmental sustainability		Feed the world		Safe food
	FF^a	CF^b	FF	CF	FF	CF	FF	CF	FF	CF
Model paths
Shared values → competence	.664	.522	.579	.498	.609	.535	.566	.529	.682	.573
Shared values → fiduciary responsibility	.618	.618	.614	.573	.547	.625	.543	.595	.644	.618
Competence → trust	.201	.178	.219	.133	.266	.187	.278	.263	.246	.154
Fiduciary responsibility → trust	.651	.698	.634	.738	.559	.669	.591	.621	.635	.705
Trust → support	.492	.422	.473	.362	.530	.408	.632	.543	.569	.412
Model statistics
R²: Support	.40	.28	.48	.32	.44	.33	.55	.43	.43	.24
R²: Trust	.67	.67	.65	.68	.62	.64	.69	.68	.69	.66
χ² (degrees of freedom = 28)	136.31	147.60	152.03	159.11	182.23	190.89	140.05	89.63	124.30	156.88
Adjusted goodness-of-fit index	.939	.934	.935	.931	.925	.970	.940	.960	.945	.932
Critical N	354.83	329.08	318.25	305.33	265.66	254.68	345.39	541.28	389.02	309.67

Note. All parameter estimates are statistically significant at p < .01.

Estimates and statistics for the respondent evaluations of family farmers (n = 1,000).

Estimates and statistics for the respondent evaluations of commercial farmers (n = 1,004).

The parameter estimates for the effects of perceived shared values on perceived competence and fiduciary responsibility all were statistically significant at p < .01, as was the case for the parameter estimates for all the modeled independent variables (i.e., shared values, competence, fiduciary responsibility, and trust). We found that the standardized parameter estimates for the effect of perceived shared values on perceived competence (i.e., a possible indication of substantive rationality) were for the 10 evaluations of the expanded recreancy theorem nearly as large or larger than the estimates for the effect of perceived shared values on perceived fiduciary responsibility (i.e., a presumed indication of formal rationality). The critical issue here is whether these statistically and substantively significant parameter estimates for the effects of shared values on perceived competence (i.e., ranging in size from .499 to .682) were moderated by how strongly respondents desired for the goal under consideration to be realized.

We calculated the product of perceived shared values and goal orientation to measure the extent to which the interaction between these two variables moderated the effect of perceived shared values on perceived competency. Our procedure was to conduct a chi-square difference test (Bollen, 1989) for the significance of the path between perceived shared values and perceived competency wherein the baseline model contained just the main effects of perceived shared values and goal orientation and the test model included the measure of the interaction between these two variables. We found for 9 of the 10 evaluations of the expanded recreancy theorem that the relationship between perceived shared values and perceived competency was not moderated by the extent to which the goal under consideration was desired by the respondent (Table 3). Therefore, as much as might be possible to ascertain within reasonable limits of empirically testing alternative goals, in 9 of 10 models we found an indication of expressed substantive rationality. It seems, therefore, for most applications of the model that respondents were expressing values for their own sake without necessarily doing so for the prospects of success—an indication of expressing substantive rationality.

Table 3.

Examination of the Moderating Effect of Goal Orientation on the Relationship Between Shared Values and Perceived Competency.

	Sample assessing family farmers (n = 1,000)		Sample assessing commercial farmers(n = 1,004)
Model	χ²	Degrees of freedom	χ²	Degrees of freedom
Affordable food
Baseline model^a	136.308	28	147.597	28
Test model^b	136.287	27	144.407	27
Animal welfare
Baseline model	152.025	28	159.113	28
Test model	150.329	27	155.406	27
Environmental sustainability
Baseline model	182.233	28	190.885	28
Test model	181.108	27	184.072**	27
Feed the world
Baseline model	140.047	28	89.627	28
Test model	140.047	27	89.597	27
Safe food
Baseline model	124.299	28	156.877	28
Test model	122.302	27	155.530	27

The baseline model is the expanded recreancy theorem (see Figure 1).

The test model is the expanded recreancy theorem testing for a potential moderating effect of goal orientation on the relationship between perceived shared values and perceived competency.

The difference in chi-square at 1 degree of freedom is statistically significant at p < .01.

The expanded model included five covariates specified as having a direct effect on expressions of trust in and support for farmers: age, sex, education, income, and concern about the goal under consideration. The mean percentage contribution to the explained variance in trust for perceived competence, perceived fiduciary responsibility, the combined effects of the social-demographic variables, and expressed concern were 10.34%, 88.17%, 0.40%, and 1.09%, respectively. Thus, the social-demographic variables and expressed concern about the goal under consideration had just small relative contributions to the explanation of expressed trust in farmers (analysis based upon an evaluation of Type III sum of squares). The mean contributions of trust and the combined social-demographic variables to the explained variance in support equaled 74.68% and 1.95%, respectively. The mean percentage contribution of concern about the goal under consideration to the explained variance in support for farmers equaled 23.37%, an effect size supported by psychometric studies of risk perceptions (e.g., Frewer et al., 1996, Slovic, 2000).

Discussion

Evaluation of the Recreancy Theorem

Because the findings here replicate previous studies indicating support for the recreancy theorem, we believe the theorem’s explanatory power is sufficient to warrant its application to understanding key determinants of trust in representatives of the U.S. food system. Additionally, we believe that the theorem’s parsimony can be appealing to practitioners in that it focuses attention upon just two determinants of trust: perceived competency and fiduciary responsibility, both of which are subject to change as a result of actions taken by food system representatives. The value of the recreancy theorem is that it addresses the extent to which the public believes that representatives of societal institutions have the necessary skills and integrity to adequately fulfill their responsibilities to the public. It should be noted, however, that the theorem provides no insight into the psychological mechanisms by which perceptions are formed, including the incorporation of the many heuristics people use to form their beliefs (see Slovic, 2000; Sunstein, 2005). Similarly, the expanded model examined here does not identify which values (e.g., Douglas & Wildavsky, 1982) might affect consumers’ evaluations of competency or fiduciary responsibility or their trust in and support for representatives of the U.S. food system. Nor does the model account for the sociological mechanisms that influence the social construction of perceptions (see discussions of social cascades and group polarization as reviewed, for example, by Sunstein, 2006; see also the literature regarding the diffusion effect, i.e., Rogers, 1995).

In summary, the recreancy theorem excels in its simplicity to identify the key determinants of trust. However, because the model provides little guidance into what factors influence these perceptions, one can be left wondering how to interpret estimations of it. This study sought to address this issue in a general sense by empirically examining the extent to which consumers express substantive rationality in their evaluations of expert recommendations regarding the U.S. food production system. We found, at least within the context of our procedures, evidence that consumers express both substantive and formal rationality in such evaluations. The significance of this finding can be illustrated by considering it within the context of contemporary civic discourse regarding the extent to which public values should be incorporated within social policy formation. We explore this line of inquiry specifically with regard to consumer perspectives of U.S. agricultural production goals.

Applied Implications of the Findings

Much of contemporary debate about the extent to which the public should be involved in technology policy formation centers on issues in agricultural production (e.g., Sunstein, 2005, 2006), where U.S. consumers express their opinions about agricultural practices through their adoption or rejection of new technologies (e.g., Israel & Hoban, 1992; Sapp & Korsching, 2004) and their advocacy for legislative changes in agricultural policies (e.g., Auld, 1990; Humane Society of the United States, 2011; Lovvorn & Perry, 2009; Lulka, 2011; Sierra Club, 2011). Within the past decade especially, consumers have successfully pursued legislative solutions, particularly with regard to addressing environmental and animal welfare issues, typically lobbying for what experts consider to be less rational technologies (e.g., Croney, 2010). As expected, however, citizens’ pursuit of legislative solutions raises questions about the extent to which social policy should be determined by people whose opinions are relatively uniformed and sometimes reflect ineffective use of decision-making heuristics (see especially reviews by Dietz & Stern, 2008; Renn, 2008). Sunstein (2005), for example, in assessing public responses to genetically modified food products, proposes that because people’s opinions always will be influenced by a lack of full knowledge and the use of heuristics that often mislead them regarding actual risks (i.e., as determined by expert risk analysis), a deliberative society is best served by having expert panels listen to citizens’ value expressions but develop social policies that conform with scientific findings. He thereby argues that expert panels should be the principal agents in forming social policy—policy that is guided by values but not by blunders. Of course, ample evidence exists to note that the enterprise of science itself is guided by politics, economics, religious beliefs, culture, and the like (e.g., Freudenburg, 1988; Shrader-Frechette, 1991); and sound scientific evidence might be superseded by the implementation of politically expedient social policies. For example, although civic leaders in the European Union might recognize that restrictions placed upon imports of genetically modified food arising from consumer-voiced safety concerns are unsupported by scientific evidence, they might nevertheless impose such restrictions to appease concerned consumers (e.g., Carter & Gruere, 2003). As a further potential limitation of the expert panel approach one wonders about the extent to which a skeptical public will accept the idea of social policy being influenced for the most part by panels comprising persons whom they mistrust (Croney, 2010). Fischhoff (1995), for example, notes that approaches that ignore public opinion, present just the facts, interpret the facts for the public, or rely upon other strategies that represent a one-way communication from risk experts to the public have in the past failed both to adequately inform public opinions and to achieve public support. Therefore, within a democratic society it might be unrealistic to presume that citizens will delegate social policy decision making to expert panels. Nevertheless, Sunstein (2005) echoes the sentiments of others (e.g., Goklanv, 2001; Graham, 2004; Hanekamp, 2006; Powell, 2010) in advocating for a greater influence of science-based expert opinion when designing social policies.

In contrast to this position, Kahan and colleagues (e.g., Cultural Cognition Project, 2011) contend that greater attention should be paid to the value-based expressions of citizens. They assert that the key determinants of trust in and support for social policies are the value orientations that citizens use to inform their policy preferences rather than the heuristics they use to interpret risks. Relying upon the grid-group typology advanced by Douglas and Wildavsky (1982), Kahan and colleagues propose that effective and acceptable social policies can be achieved when they elicit a sense of shared values among persons/groups with differing value orientations. It should be noted that value orientations do not necessarily predict social policy preferences. For example, a person with an egalitarian value orientation and a global vision might support the agricultural production goal to “grow enough food to help feed the world.” A person with an egalitarian value orientation and a vision limited to “America first,” on the other hand, might oppose this goal. Thus, as Sunstein (2006) notes, the culture cognition approach might simply represent one form of bounded rationality. Nevertheless, the findings of Kahan and associates support the claim that policies that give people a sense of shared values can be effective in gaining consensus support. Recent legislative action regarding egg production, for example, represents in part a reconciliation among institutional representatives who express differing value orientations (e.g., Humane Society of the United States, 2011).

It should be recognized that some persons, even if they hold strong opinions, do not necessarily wish an audience for them among policy makers. And to the extent that one believes that experts share their values one might feel comfortable with policy being guided by these experts. Yet it is not uncommon, especially within the arena of agricultural production policy formation, for expert opinions to differ substantially from those of a public that has only their value orientations to offer as justification for policy alternatives. For example, whereas experts in a recent policy debate strongly agreed that the current size of hen cages were scientifically justified, U.S. consumers successfully voiced their value-oriented opinions that cages needed to be larger (Humane Society of the United States, 2011). In this sense, although public opinions might be relatively uniformed, they nevertheless are important to policy formation and not necessarily irrational in content. To the extent that citizens wish to be heard, therefore, the question for a democratic society becomes to what extent should policy makers heed the advice of relatively uniformed citizens in lieu of the advice offered by experts?

Our findings can be interpreted to support either of the two perspectives described above. To the extent that the findings here demonstrate an intrinsic importance to value expression, they support social policy formation that facilitates reconciliations among groups/people with diverse value orientations. This approach, however, might be frustrated if citizens are expressing their values with no desire to attain an externally identifiable goal, as would be the case when these expressions represent instances of substantive rationality. In this case, pursuing such approaches will be ineffective to the extent that values are entrenched. Alternatively, in indicating that citizens to some extent express their values with no extrinsically identifiable rationale in mind the findings here support social policy formation that accentuates decisions made by expert panels. After all, if citizens express their values with no more rationale than their desire to do so then it seems prudent for a nation to provide expert guidance to social policy formation. However, along these same lines the findings indicate that expert panel approaches—ones that listen to but overrule citizens’ value orientations—might be difficult to implement because people strongly desire value expression. Also, implementing such approaches might foment distrust in societal institutions to the extent that they are perceived as being insensitive to citizens’ values.

In conclusion, finding that consumers seemingly use both substantive and formal rationality to evaluate the U.S. food system presents challenges to gauging the extent to which consumer opinions should be incorporated into social policy formation. If indeed it is unrealistic to implement expert panel approaches to social policy formation, then the most successful approach to achieving acceptable U.S. agricultural production policies, even if in some cases this approach might be either inefficient or ineffective, will involve actions taken by agricultural producers, agribusiness firms, and agriculture-related government institutions to develop goals and pursue agricultural production practices that are based upon mutual understandings and shared values with consumers. With these considerations in mind, we believe our results indicate that developing both well-reasoned and publically acceptable technology-related policy will require greater rather than less emphasis upon furthering effective science communication theory and practice.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Stephen G. Sapp is a professor of sociology at Iowa State University and specializes in issues of consumer trust in the U.S. food system.

Peter F. Korsching is professor emeritus of sociology at Iowa State University and specializes in issues of public involvement in social policy formation and change.

Charlie Arnot is the CEO of the Center for Food Integrity, and he works with companies and associations across the food system to develop and implement sustainable solutions in issues management, public relations, and strategic facilitation and marketing communication.

Jannette J. H. Wilson is the President of Gestalt, Inc. and conducts marketing research and offers consultation regarding strategic planning for firms engaged in national and global marketplaces.

References

Ackerman

(1997). Consumed in theory: Alternative perspectives on the economics of consumption. Journal of Economic Issues, 31, 651-664.

Alario

Freudenburg

(2003). The paradoxes of modernity: Scientific advances, environmental problems, and risks to the social fabric. Sociological Forum, 18, 193-214.

Anand

(1993). Foundations of rational choice under risk. Oxford, England: Oxford University Press.

Auld

M. E.

(1990). Food risk communication: Lessons from the Alar controversy. Health Education Research, 5, 535-543.

Barber

(1983). The logic and limits of trust. Rutgers, NJ: Rutgers University Press.

Bollen

(1989). Structural equations with latent variables. New York, NY: Wiley.

Boudon

(1981). The logic of social action. London, England: Routledge & Kegan Paul.

Carter

C. A.

Gruere

G. P.

(2003). International approaches to the labeling of genetically modified food. Choices, 18, 1-4.

Cohen

W. R.

(1985). Criteria for technology acceptability. Risk Analysis, 5, 1-3.

10.

Coleman

J. S.

(1990). Foundations of social theory. Cambridge, MA: Harvard University Press.

11.

Croney

C. C.

(2010). Should animal welfare policy be influenced by consumers’ perceptions? Retrieved from http://www.wcds.ca/proc/2011/Manuscripts/Croney.pdf

12.

Cultural Cognition Project. (2011). People. Dan M. Kahan. Retrieved from http://www.culturalcognition.net/kahan/

13.

Dietz

Stern

P. C.

(2008). Public participation in environmental assessment and decision making. Washington, DC: National Research Council.

14.

Douglas

M. T.

Wildavsky

A. B.

(1982). Risk and culture: An essay on the selection of technical and environmental dangers. Berkeley: University of California Press.

15.

Earle

T. C.

(2010). Trust in risk management: A model-based review of empirical research. Risk Analysis, 30, 541-574.

16.

Earle

T. C.

Cvetkovich

G. T.

(1995). Social trust: Toward a cosmopolitan society. Westport, CT: Praeger.

17.

Earle

T. C.

Siegrist

(2006). Morality information, performance information, and the distinction between trust and confidence. Journal of Applied Social Psychology, 36, 383-416.

18.

Earle

T. C.

Siegrist

Gutscher

(2007). Trust, risk perception, and the TCC model of cooperation. In Siegrist

Earle

T. C.

Gutscher

(Eds.), Trust in cooperative risk management (pp. 1-49). London, England: Earthscan.

19.

Elster

(1989). Nuts and bolts for the social sciences. Cambridge, England: Cambridge University Press.

20.

Fischhoff

(1995). Risk perception and communication unplugged: Twenty years of progress. Risk Analysis, 15, 137-145.

21.

Fishbein

Ajzen

(1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley.

22.

Freudenburg

W. R.

(1988). Perceived risk, real risk: Social science and the art of probabilistic risk assessment. Science, 242, 44-49.

23.

Freudenburg

W. R.

(1993). Risk and recreancy: Weber, the division of labor, and the rationality of risk perceptions. Social Forces, 71, 909-932.

24.

Freudenburg

W. R.

Pastor

S. K.

(1992). Public responses to technological risks: Toward a sociological perspective. Sociological Quarterly, 33, 389-412.

25.

Frewer

L. J.

Howard

Hedderley

Shepherd

(1996). What determines trust in information about food-related risks? Risk Analysis, 17, 473-500.

26.

Goklanv

I. M.

(2001). The precautionary principle: A critical appraisal of environmental risk assessment. Washington, DC: Cato Institute.

27.

Gollier

(2004). The economics of risk and time. Cambridge: MIT Press.

28.

Graham

J. D.

(2004). The perils of the precautionary principle: Lessons from the American and European experience (Report No. 118). Washington, DC: Heritage Foundation.

29.

Hanekamp

J. C.

(2006). The precautionary principle: A critique in the context of the EU food supplements directive. Environmental Liability, 2, 43-51.

30.

Hechter

Kanazawa

(1997). Sociological rational choice theory. Annual Review of Sociology, 23, 191-214.

31.

Humane Society of the United States. (2011). Farm animal protection. Retrieved from http://www.humanesociety.org/issues/

32.

Israel

G. D.

Hoban

T. J.

(1992). Concern about eating genetically engineered food. Southern Rural Sociology, 9, 23-43.

33.

Jones

B. D.

(1999). Bounded rationality. Annual Review of Sociology, 2, 297-321.

34.

Jöreskog

Sörbom

(1995). LISREL 8 [Computer software]. Chicago, IL: Scientific Software International.

35.

Kahan

D. M.

Slovic

(2006). Cultural evaluations of risk: Values or blunders? Harvard Law Review, 119, 166-172.

36.

Kahan

D. M.

Slovic

Braman

Gastil

(2006). Fear of democracy: A cultural evaluation of Sunstein on risk. Harvard Law Review, 119, 1071-1109.

37.

Kahneman

Tversky

(1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263-291.

38.

Lovvorn

J. R.

Perry

N. V.

(2009). California Proposition 2: A watershed moment for animal law. Journal of Animal Law, 15, 149-169.

39.

Luhmann

(1979). Trust and power. New York, NY: Wiley.

40.

Lulka

(2011). California’s Proposition 2: Science, ethics, and the boundaries of authority in agriculture. Culture, Agriculture, Food and Environment, 33, 29-44.

41.

Powell

(2010). What’s the harm? An evolutionary theoretical critique of the precautionary principle. Kennedy Institute of Ethics Journal, 20, 181-206.

42.

Renn

(2008). Risk governance: Coping with uncertainty in a complex world. London, England: Earthscan.

43.

Ritzer

(2010). Sociological theory (8th ed.). New York, NY: McGraw-Hill.

44.

Rogers

E. R.

(1995). Diffusion of innovations (5th ed.). New York, NY: Free Press.

45.

Rosa

E. A.

Freudenburg

W. R.

(1984). Nuclear power at the crossroads. In Freudenburg

W. R.

Rosa

E. A.

(Eds.), Public reactions to nuclear power: Are there critical masses? (pp. 3-37). Boulder, CO: Westview Press.

46.

Sapp

S. G.

Arnot

Fallon

Fleck

Soorholtz

Sutton-Vermeulen

Wilson

J. J. H.

(2009). Consumer trust in the U.S. food system: An examination of the recreancy theorem. Rural Sociology, 74, 525-545.

47.

Sapp

S. G.

Downing-Matibag

(2009). Consumer acceptance of food irradiation: A test of the recreancy theorem. International Journal of Consumer Studies, 33, 417-424.

48.

Sapp

S. G.

Korsching

P. F.

(2004). The social fabric and innovation diffusion: Symbolic adoption of food irradiation. Rural Sociology, 69, 347-369.

49.

Sarkar

Pfeifer

(2006). The philosophy of science: Vol. 1. A-M. London, England: Taylor & Francis.

50.

Sen

(1977). Rational fools: A critique of the behavioral foundations of economic theory. Philosophy & Public Affairs, 6, 317-344.

51.

Shrader-Frechette

K. S.

(1991). Risk and rationality: Philosophical foundations for populist reforms. Berkeley: University of California Press.

52.

Siegrist

Earle

T. C.

Gutscher

(2003). Test of a trust and confidence model in the applied context of electromagnetic field (EMF) risks. Risk Analysis, 23, 705-716.

53.

Sierra Club. (2011). Sierra club news releases. Retrieved from http://action.sierraclub.org/site/PageNavigator/E-Newsletters/Pressroom

54.

Simon

(1982). Models of bounded rationality. Cambridge: MIT Press.

55.

Slovic

(2000). The perception of risk. London, England: Earthscan.

56.

Smelser

(1992). The rational choice perspective. Rationality and Society, 4, 381-410.

57.

Starr

(1969). Social benefit versus technological risk. Science, 165, 1232-1238.

58.

Sunstein

C. R.

(2005). Laws of fear: Beyond the precautionary principle. New York, NY: Cambridge University Press.

59.

Sunstein

C. R.

(2006). Misfearing: A reply. Harvard Law Review, 119, 1110-1125.

60.

U.S. Census Bureau. (2007). American community survey. Retrieved from http://www.censusgov/acs/

61.

U.S. Census Bureau. (2010). Reported internet usage for households, by selected householder characteristics: 2009. Washington, DC: Current Population Survey.

62.

Weber

(1968). Economy and society ( Roth

Wittich

, Eds., Vol. 1). Berkeley: University of California Press. (Original work published 1921)

63.

Zafirovski

(2005). Is sociology the science of the irrational? Conceptions of rationality in sociological theory. American Sociologist, 36, 85-110.