Keeping focus on the consumer is critical for the fast food industry as well as other customer service firms. In 2014, the average U.S. consumer spent $2,390 on food [1] . Consumers are increasingly demanding more from their food, and fast food companies in particular have been trying to meet customers’ growing demands for social responsibility. McDonald’s [2] is claiming to make sustainable practices the “new normal” for their business while Panera Bread [3] is focusing on “food as it should be” with clean ingredients, menu transparency, and an initiative to make a positive impact on the food system. Despite the efforts of these companies, to our knowledge, no studies have thoroughly examined consumers’ perceptions of the social responsibility of fast food companies.

Corporate Social Responsibility (CSR) is not a new topic of discussion. Yet a consistent, objective definition and set of corresponding public expectations remain elusive. Corporate Social Performance (CSP) and the element that is the social responsibility of firms have been examined since the mid-1950’s [4] . However, some suggest that it has been practiced in the U.S. since the late 1890’s [5] . One early definition of CSR is “the firm’s consideration of, and response to, issues beyond the narrow economic, technical, and legal requirements of the firm (to) accomplish social benefits along with the traditional economic gains which the firm seeks” ( [6] , pp. 312-313). [4] breaks CSR into three principles: legitimacy, public responsibility, and managerial discretion applied at the institutional, organizational, and individual level.

The current movement towards CSR may be influenced by the belief that companies can be more influential and effective in addressing social and environmental issues and the general argument that “good corporate citizenship is also good business” ( [7] , p. 11). The impact of CSR in the fast food sector may be tied to the ability of restaurants to further differentiate themselves through cause-related marketing [8] . Despite differences in cuisine, the overall quality, price, and service offered by fast food restaurants is often similar. Thus, Bronn and Vrioni [8] warn firms pursuing cause-related marketing as a means of promoting their CSR initiatives that they must first ensure that they are genuine and have a full understanding of their consumers’ knowledge, understanding, and potential skepticism.

The customer-perceived value is comprised of tangible attributes and is related to product quality and processing. However, this is increasingly related to less tangible attributes, like brand image and CSR [9] [10] . Key aspects of these intangible attributes (i.e. animal welfare and ethically sourced ingredients) are often important to consumers although they cannot be verified through consumption of the food. Schroeder and Mc Eachern [10] therefore argue that there then become two audiences for CSR initiatives: the consumer of the product and the broader public.

One mixed-methods study from the UK examined the awareness and perceptions of CSR for McDonald’s and KFC [10] , focusing on students from an English and Scottish university. The majority of fast food purchasing decisions were made on impulse. Exploratory factor analysis was used to identify four central attitudes towards fast food purchase behavior: brand value, nutritional value, ethical value, and food quality [10] . General findings were a high level of awareness of McDonald’s CSR initiatives and expectation of the majority of respondents that fast food companies should be involved in healthy eating initiatives, animal welfare, and community activities [10] .

The aim of this study was to contribute to knowledge on perceptions of CSR by investigating U.S. resident’s perceptions of the “most” and “least” socially responsible of prominent fast food restaurants. The present situation in the U.S. with regard to perceptions of CSR for fast food restaurants is largely unstudied. Nearly all food service establishments portray to their customers aspects of their business which may be interpreted as related to CSR. Information shared can range from nutritional information to procurement strategies (i.e., local or from farms using certain practices) and treatment of employees. However, the presentation of CSR information is such that comparisons across restaurants can be difficult. For example, the language used to describe different aspects of CSR may differ significantly. Further, there is not a central source of information that summarizes CSR activities for all restaurants in an accessible way. The objective of this study was to examine the tradeoffs amongst perceptions of social responsibility of prominent fast food restaurants by utilizing a best-worst scaling estimation of consumer perceptions of CSR of fast food restaurants. Previous studies have used best-worst analysis to investigate perceptions; Cummins et al. [11] sought to understand consumer preferences for “most” and “least” important pork attributes, while Wolf and Tonsor [12] investigated producer’s preferences for “most” and “least” important dairy policies. Erdem et al. [13] also used this methodology to examine tradeoffs made by both consumers and producers for relative responsibility in ensuring food safety in the food supply chain. This study sought to investigate the relationships amongst individual preferences for social responsibility of prominent fast food restaurants, participant demographic factors, and consumption practices.

An online survey was distributed to collect information from 1201 U.S. residents from March 31^st through April 4^th, 2015. The sample was targeted to be representative of the U.S. national population according to the 2012 U.S. Census. Researchers at Purdue University created the survey, and then Light speed GMI distributed a link to potential respondents in their large proprietary opt-in panel. The questionnaire was designed to collect basic demographic information (gender, age, pre-tax income, education, region of residence) in addition to information about monthly food expenditures, fast food consumption habits, and the nature and frequency of their restaurant visits. Respondents were required to be at least 18 years of age to participate in the survey.

Within the survey, a randomly selected subset of participants, n = 302, was shown a best-worst choice experiment in which they indicated their preference for the “most” and “least” socially responsible fast food restaurants over a series of best-worst scaling (BWS) choice scenarios. Preferences of those respondents (n = 302), elicited through the BWS choice experiment were analyzed. All analyses were completed using the sample of respondents who completed the BWS questions (n = 302).

In addition to the BWS, this analysis used cross-tabulations to investigate relationships between monthly fast food consumption and demographic factors, including gender, households with/without children, age, household income, and education level.

In BWS methodology, participants are forced to make tradeoffs over a series of choice scenarios offering multiple options [14] . Developed by Jordan Louviere in the late 1980’s, BWS originated in random utility theory [15] . Also called maximum-difference scaling, BWS outcomes represent the maximum difference between a respondent’s most and least preferred options [16] . This study investigated individual perceptions of the “most” and “least” socially responsible fast food restaurants. The eleven fast food restaurants that were presented in this study were (in no particular order): McDonald’s, Subway, Panera Bread, Starbucks, Wendy’s, Burger King, Taco Bell, Dunkin’ Donuts, KFC, Chick-fil-A, and Chipotle.

In this experiment, participants could choose any one fast food restaurant up to five times. Given that eleven attributes (fast food restaurants) were presented (j), J = 11 in the experiment, the total possible choice combinations were calculated as: J(J − 1) = 110. In short, there were 110 aggregate possible best-worst choice combinations. Participant selections of the “most” and “least” socially responsible fast food restaurants were used to determine the relative social responsibility of each fast food restaurant presented in this study. These two selections represent the maximum difference between two attributes on the underlying continuum of importance [17] . Similar in form to Lusk and Briggeman’s [17] study, participant’s latent unobservable level of importance is denoted as: I_ij = λ_j + ε_ij, in which λ_j is used to represent the location of importance for each attribute, j on the continuum of importance, in which the random error term is ε_ij.

The probability that a respondent chooses j and k, respectively, as the best and worst, or “most” and “least,” socially responsible fast food restaurants, is the probability that the difference between I_ij and I_ik is larger than all other J(J − 1) − 1 possible choice combinations [17] , thus represented by the maximum difference between a respondent’s two chosen attributes. As in the experiment outlined by Lusk and Briggeman [17] , the error term is assumed to be independently and identically distributed. The probability of choosing one attribute as most socially responsible and another attribute as least socially responsible was represented in the multinomial log it (MNL) form:

However, the MNL model was hypothesized to not accurately represent participant preferences as it assumes homogenous preferences across attributes. Participant’s perceptions of CSR amongst fast food restaurants were hypothesized to be heterogeneous. Studies such as Shwartz [18] and Auger et al. [19] , have shown that individual people, even within the same society, can have unique preferences. Therefore, the random parameter logit (RPL) model, which assumes heterogeneous preferences amongst respondents for the presented attributes, was estimated in addition to the MNL. Adjustments from (1) for the RPL model include the unobservable level of importance for respondent i and attribute j in population λ_j, in which the mean is represented as, the standard deviation, and the random term μ_i. Adjustments for the RPL model were then specified as:. The random term, within the RPL model, was normally distributed with mean zero and unit standard deviation; thus the level of importance of CSR attribute j is distributed according to a normal distribution curve [17] .

The probability that each fast food restaurant was picked as most important across all eleven restaurants necessarily sums to one. Shares of preference represent the probabilities and were calculated as:

Shares of preference for all eleven fast food restaurants were calculated using the RPL model. The perceived CSR (relative to all other fast food restaurants presented) for individual-specific RPL parameter estimates were found through maximum likelihood estimation (MLE) using the individual-specific coefficients from the RPL model. Normalizing MLE to 0 prevents the dummy variable trap [17] . Relative perceived importance or individual-specific preference shares were estimated for each of the eleven fast food restaurants for each respondent (using individual-specific parameter estimates). The individual-specific parameter estimates equipped further analysis, specifically correlations between preferences shares and demographic or consumption factors.

Summary demographics for the sample are displayed in Table 1. Also shown in Table 1 are the census figures, which were targeted proportions for the sample obtained. Thirteen percent of survey participants were between the ages of 18 and 24, 35% were 25 to 44, 35% were aged 45 to 64, and 17% were 65 years and older. For analysis purposes, respondent household incomes were grouped into low (less than $25,000 to $34,999), medium ($35,000 to $99,999), and high ($100,000 to $150,000 or higher) income categories; in this sample, 33% classified as low, 46% medium, and 21% high income earning households. Information about geographic region of residence was also collected. In total, 18% of respondents were from Northeast, 41% from the South, 20% from the Midwest, and 21% from the West. In terms of age, household income, and region of residence, the proportion of the sample in each bracket or category was extremely close to those targeted from the U.S. Census.

Information was also collected from respondents with respect to their education levels, revealing that 57% of respondents graduated with a higher education degree (Associates, Trade, Bachelors, Masters, or PhD). The sample in this study was more educated than the U.S. population overall, which could be due to a number of factors including internet access, available time online, ability to read and/or comprehend, and interest in voluntary participation. The majority of respondents, 68%, did not have children in their household. The information regarding household composition is particularly pertinent when investigating food consumption habits, as having children in the household is believed to influence attitudes toward food and consumption practices. Information regarding respondent/household vegetarian and vegan dietary preferences was also collected. The majority of respondents claimed neither that they nor any member of their household was vegan or vegetarian (78% and 79%, respectively).

The knowledge of respondents with respect to fast food operations was of interest, given the focus of this analysis on perceptions of fast food restaurants’ perceived responsibility. Responses on questions pertaining to fast food, knowledge of the related business practices, and consumption habits and summary statistics are presented in Table 2. With respect to business practices in fast food, 45% of respondents self-reported being unknowledgeable, compared to 31% of respondents who indicated they were knowledgeable.

In addition to an individual’s perceived level of knowledge, the consumption behaviors of respondents were of interest, specifically, individual perceptions of fast food with respect to food consumption practices. Respondents were asked about average household weekly food expenditures; 23% spent less than $50/week, 34% spent $51 - 100/week, 30% spent $101 - $200/week, while the remaining 12% spent more than $200/week. On average, respondents indicated spending 75.8% of their food budget on food consumed at home and the remaining 24.2% on food consumed away from home. To gain further insight into spending on food away from home,

respondents were asked the average number of times they visited different restaurant types (Figure 1). Restaurant types presented to respondents were: fast food/sit down in restaurant, fast food/take out or drive through, fast casual, casual dining, and fine dining. Most pertinent to this study was people’s fast food consumption habits. With respect to fast food/sit down in restaurant eating, the largest portion of respondents, 39%, claimed they did not eat

at any such establishments in a given month, while 32% claimed to eat at such restaurants 1 - 3 times monthly, 12% claimed 4 - 6 monthly visits, 3% indicated 7 - 9 visits, and 9% ate fast food more than 10 times per month. Fast food/take out or drive through figures were not practically different as 31% claimed not to eat any take out fast food in a given month, 34% ate take out fast food 1 - 3 times per month, 19% ate takeout fast food 4 - 6 times per month, 4% ate take out 7 - 9 times per month, and 12% ate at such establishments more than 10 times per month. Nearly half of participants, 47%, claimed to eat at fast casual restaurants 1 - 3 times per month, while 35% did not do so at all, 12% ate at such restaurants 4 - 6 times monthly, 2% partook 7 - 9 times per month, and 4% ate fast casual food more than 10 times per month.

Few differences were observed between males and females for monthly restaurant visits (Table 3). A significantly larger percentage of women visited a fast food restaurant for take-out or went through the drive through 1 - 3 times per month compared to men, who had a larger percentage of fast food restaurant or drive through visits for take-out 4 - 6 times per month. Not surprisingly, households with children frequented fast food take out or drive through options more frequently than households without children. The percentage of respondents who reported no visits in a month was 37.1% for households without children compared to 18.6% with children. The percentage of those that frequented 10 or more times was 7.8% for households without children and 19.6% for those with children. Respondents of different ages (which was often related to having children at home or not) also varied in their visits to fast food restaurants. Over 52% of respondents 65 years and older had no monthly

visits to a fast food restaurant to sit down, for takeout or drive through, or a fast casual restaurant, and less than 2% of respondents frequented a fast food restaurant 10 or more times in a month. Comparatively, approximately 20% of respondents in the 25 - 44 year age group reported no visits to a fast food restaurant and almost 20% of these respondents had 10 or more visits.

1Significant differences are denoted by *, **, 2Significant differences are denoted by I, II, 3Significant differences are denoted by a, b, c, d, 4Significant differences are denoted by A, B, C, 5Significant differences are denoted by α, β

The experiment in this study forced participants to make tradeoffs in perceived social responsibility amongst the different fast food options, thus, resulting in relative rankings of CSR (Table 4). Panera Bread was selected as the restaurant that was most socially responsible, given its relative size of mean preference share (19.8%) followed by Subway (14.2%) and Chick-fil-A (12.3%). The restaurants perceived the least socially responsible with the lowest mean preference shares were McDonald’s (4.7%), KFC (4.6%), and Taco Bell (4.1%). While the

specific preference share estimates are of interest for further analysis, the relative rankings of restaurants (as measured by preference shares) are themselves of practical significance. Given the vast array of marketing materials and consumer messaging being employed by restaurants, it is likely of interest within the fast food industry to understand the relative ranking of fast food restaurants (according to perceived CSR) by respondents.

The relationships amongst perceived social responsibility of the different fast food restaurants are shared in Table 5. Amongst those relationships, generally the sizes of preference shares for restaurants perceived to be “most” socially responsible negatively correlated with the sizes of preference shares for those restaurants that were perceived to be “least” socially responsible. This was the case with the sizes of preference shares for Panera Bread, Chipotle, Chick-fil-A, and Starbucks when correlated with the sizes of preference shares for all other restaurants, including each other. Restaurants perceived to be least socially responsible typically positively correlated with each other as did the relationships amongst sizes of preference shares for McDonald’s, Wendy’s, Burger King, Taco Bell, Dunkin’ Donuts, and KFC.

Responses to perceived social responsibility with respect to sample demographics are presented in Table 6. Being female was positively correlated with the size of preference share for Chipotle. Women more often gave higher preference share to Chipotle (relative to all other fast food options presented). Interestingly, respondent’s membership in the 18 - 24 year age group positively correlated with the sizes of preference shares for Burger King and Chipotle, and negatively correlated with the size of preference share for Panera Bread. Being a

Statistical significance to the 1%***, 5%**, and 10%* levels.

Statistical significance are indicated as 1%***, 5%**, and 10%* levels.