Next Article in Journal
A Bi-Band Binary Mask Based Land-Use Change Detection Using Landsat 8 OLI Imagery
Previous Article in Journal
Agricultural Productivity Growth and the Role of Capital in South Asia (1980–2013)
Article Menu
Issue 3 (March) cover image

Export Article

Sustainability 2017, 9(3), 478; https://doi.org/10.3390/su9030478

Article
Factors Affecting Current and Future CSA Participation
1
Department of Agricultural Sciences, Clemson University, Clemson, SC 29634, USA
2
Food and Resource Economics Department, University of Florida, Gainesville, FL 32611, USA
*
Author to whom correspondence should be addressed.
Academic Editor: John P. A. Lamers
Received: 30 January 2017 / Accepted: 17 March 2017 / Published: 22 March 2017

Abstract

:
Community Supported Agriculture (CSA) is one of the widely used direct marketing strategies for small- and midsized farmers. CSA programs are an important option for sustainable production and consumption. It helps growers generate income (improve financial security) and consumers obtain fresh local foods. Sustaining and growing CSA participation is critical in order to continue enjoying these benefits. We used a national online survey in conjunction with discrete choice models to investigate the impact of demographic characteristics, lifestyle preferences, and different information outlets on the probability that a consumer is or will become a CSA member. The results indicate that the factors affecting current and future CSA participation differ substantially. While none of the demographic characteristics has a significant impact on current CSA participation, some of them significantly affect the probability that a consumer will become a CSA member in the future. Lifestyle preferences have a significant impact on current and future CSA participation. Although none of the information outlets examined affect current CSA participation, word-of-mouth and online sources significantly influence the probability that a consumer will join a CSA program in the future. These findings may have important implications for policy makers’ and CSA farm managers’ efforts to sustain future CSA development.
Keywords:
community-supported agriculture (CSA); direct marketing; local foods

1. Introduction

Consumer preferences for food have drastically changed over the last decades. Besides the nutrients provided by food, consumers increasingly care about the impact of food production on the environment and society [1,2,3]. To satisfy the demand of this increasingly sophisticated group of consumers, producers more extensively utilize different forms of direct marketing, including farmers markets, on-farm sales, roadside sales, U-pick operations, and community-supported agriculture (CSA) [3,4,5]. Other than farmers markets, CSA has become one of the most widely used direct marketing strategies over the last twenty years to satisfy the increased demand for sustainable production and consumption [6,7,8]. The number of U.S. farms that market their products through CSA has increased from an estimated 500 farms in 1996 to more than 12,000 farms in 2012—approximately 0.6% of U.S. farms [9,10,11,12,13]. Under a CSA arrangement, consumers purchase “shares” of a farm’s expected yield before the planting period and obtain a portion of the produce later during the harvesting season [14,15,16]. The most common CSA agreements are: (i) subscription-based, a type that is farmer driven; and (ii) shareholder-based, a type that is primarily consumer driven [17].
The growth of the CSA marketing scheme can be explained by the multitude of benefits enjoyed by CSA consumers and producers alike. For instance, CSA improves the financial security of producers through advanced payments [15,18] and the elimination of the middleman [18,19,20]. Moreover, consumers join CSA to obtain local produce for a healthier lifestyle, to reduce the negative impact on the environment, and to financially support local producers and the community [10,15,20,21]. As CSA arrangements become more popular, the model adapts to increase its customer base and further improve the experience of the members [7]. For instance, some CSA farm managers adopt alternative payment schemes (e.g., installment payments) and accept SNAP benefits (the USDA Supplemental Nutrition and Assistance Program for low-income families), thus increasing their potential customer base [22,23]. Furthermore, some CSA farms collaborate in marketing and distribution strategies to reduce the seasonality and limited product variety barriers [24].
The growing popularity of CSA marketing arrangements has triggered considerable research endeavors. A common theme of this literature includes efforts to identify the economic impact of CSA arrangements on consumers and producers. Brown and Miller [14] provide a comprehensive review of papers relevant to the aforementioned topic. Their findings indicate that CSA membership leads to healthier eating habits, and, in some cases, to lower cost for consumers. They also highlight the importance of the manager’s experience and of different marketing options on the viability of CSA farms. A related strand of the literature focuses on the dietary/nutritional impacts of CSA [10,20,25]. The consensus of these studies is that CSA membership leads to improved diet and greater consumption of fresh produce.
Furthermore, a plethora of studies has examined the effect and influence of several factors on consumers’ CSA participation and CSA membership satisfaction [12,22,26,27,28]. The majority of these studies indicate that CSA members are primarily female, middle aged, with higher education and income [5,22,29]. However, divergences remain. For instance, Kolodinsky and Pelch [9] and Onianwa et al. [30] found that income has no effect on consumers’ interest in local foods. Zepeda and Li [19], and Peterson et al. [21] highlight that generally demographic characteristics do not have a statistically significant effect in predicting CSA membership. Among the large body of literature on CSA, only one study examined the factors affecting an individual’s future CSA participation. Kolodinsky and Pelch [9] have shown that education level and lower cost per share have a positive and statistically significant impact on consumers’ decision to join the CSA program in Vermont. However, identifying the factors that influence the probability an individual will join a CSA arrangement in the future, at the national level, is a relatively unexplored topic in the literature because the majority of the previous research focus on limited geographical areas.
As the competition for local food expenditures intensifies, it is important to secure the sustainability and future growth of CSA participation for both growers and consumers to enjoy the multiple benefits of CSA. Therefore, it is crucial to identify: (i) the types of consumers who are likely to participate in CSA arrangements in the future; (ii) the impact of different information outlets on the decision to participate in CSA arrangements; and (iii) aspects of CSA agreements that help retain current CSA members. The answers to these questions will provide valuable insights to farm managers in order to better understand CSA marketing, promote the sustainability of CSA marketing, retain current CSA members, and attract new CSA members. The present study aims to fill this void in research using a 2015 national online survey of regular grocery shoppers.
The present study contributes to the current literature by revealing more about the factors that influence a consumer’s decision to join a CSA arrangement in the future. Rather than binary responses (Yes or No), we use multilevel response (Yes, Not Sure, or No) to capture the potential uncertainty in individuals’ future participation in CSA. Most importantly, rather than using samples from a particular state or a small number of CSA farms, this study uses a national sample, thus helping us draw more general conclusions that may be applicable to a broader audience. The national sample also allows us to investigate the influence of region of residence on the probability that a consumer will join a CSA arrangement in the future.

2. Survey Design and Data Collection

An online survey was distributed by Survey Sampling International (SSI) to its national representative consumer panels in May 2015. SSI is a premier global provider of data solutions for consumer and business-to-business survey research. SSI operates from 40 offices in 20 countries, reaching participants in 90+ sample countries via Internet, telephone, mobile/wireless and mixed-access offerings. For this study SSI used a quota sampling method by releasing the survey to its consumer panels in several rounds. After each round, the demographics of respondents who finished the survey were compared with the census data. If some of the demographics deviated too much from the census (e.g., too many females) in a round, the survey was released to a sample with the adjusted demographics (e.g., more males). This procedure continued until the quota was met and the demographics of the final sample were as close as possible to the national population. After removing respondents with missing values, there were 768 respondents in the final analysis. The sample size of 768 is sufficient for this study. Based on the formula that calculates the sample size, a sample size of 384 is good enough for a population of one billion, with 95% confidence interval and ±5% marginal errors.
In the survey, respondents were asked about: (i) whether they were current CSA members (Yes, or No); (ii) whether they were planning to join a CSA program in the future; and (iii) the reasons for their being current CSA or non-CSA members. Demographic and lifestyle variables were also collected in the survey. Consumers’ intention to join a CSA arrangement in the future was measured by their answers to the question, “Are you planning to join a CSA in the future?” Respondents were provided with three choices: (i) No; (ii) Not sure; and (iii) Yes. The verbal statement format was preferred over a Likert scale or probabilities option because it could be used even if responders were unable to fully differentiate among the probabilities [31,32]. Not including a “probably yes/no” option as a possible answer avoids the problem of the framing effect that refers to the difficulty of interpreting how responders distinguish between the middle responses [33]. Furthermore, this approach was preferred over a binary option (yes/no) approach because it could provide additional insights regarding consumers’ intentions.

3. Empirical Framework

Because the responses to the question regarding current CSA membership or non-membership are dichotomous, a probit model was utilized to evaluate the impact of the selected explanatory variables on the probability that a survey respondent was a member of a CSA arrangement during the time of the survey. However, as the responses to the question regarding future CSA participation have more than two discrete outcomes, two models (ordered probit and multinomial probit) were used to examine the effect of selected explanatory variables on the probability that a consumer will join a CSA arrangement in the future. The ordered probit formulation assumes that the three responses (“No”, “Not sure”, and “Yes”) indicate a willingness of future participation in CSA ranging from weak to strong. The multinomial probit model assumes that each of the three responses is a unique category, and there is no particular order in these responses. Including both models helped draw more robust conclusions regarding the impact of factors affecting an individual’s participation in future CSA arrangements. Marginal effects were estimated for all the models to gain a more meaningful interpretation of the results.
Four groups of explanatory variables were examined in the study: demographic variables (age, gender, race, income, and education level), lifestyle variables (vegetarian, gym membership, hours spent cooking, frequency of recycling, grocery expenditures, and past CSA membership), information sources (web, word of mouth, news, and road signs), and regional variables (South, West, Midwest, and Northeast). Among demographics variables, age and income were treated as continuous variables; gender, race, and education were included as dummy variables, with male, non- Caucasian, and high school degree or less used as base category, respectively. Among lifestyle variables, vegetarian, gym membership, and past CSA membership were included as dummy variables, while all other variables were treated as continuous variables. All the information sources variables were treated as continuous variables. Regional variable was coded as dummy variable with Midwest region used as base category (Table 1). Following Zepeda and Li [19], Curtis [34], and Peterson et al. [21], our a priori expectation was that responders with a more active and environmentally friendly lifestyle would be more likely to join a CSA arrangement. In line with Bond et al. [4], we expected that consumers in western states would be more likely to participate in CSA arrangements. Considering the lack of consistency in findings regarding demographic variables [19,21], our initial hypothesis was that demographic variables would not be helpful in explaining the likelihood of joining a CSA arrangement. The present section briefly discusses the econometric techniques used in the study.

3.1. Probit Model for Current CSA Participation

Assume a rational, utility-maximizing consumer, indexed by i, and let Ui1 and Ui0, respectively, be the consumer’s utility from participation or nonparticipation in a CSA arrangement. The utility-maximizing individual is a CSA member if
U i * = U i 1 U i 0 > 0
The latent variable U i * is defined as
U i * = X i β + ε i
where X’ is a vector of explanatory variables, β represents the coefficients associated with these variables, and ε is the error term following normal distribution. The observed dependent variable Yi equals one if U i * > 0 , or 0 if U i * 0 [35]. Under the probit formulation, the probability that the individual i is a CSA member is given by
P i = Φ ( X i β ) = x β φ ( z ) d z
where Φ() is the standard normal cumulative distribution function (CDF). The marginal effects are calculated as
P i X i j = φ ( X i β ) β j = φ ( Φ 1 ( P i ) ) β j

3.2. Ordered Probit and Multinomial Probit Models for Future CSA Participation

If we assume that the three responses to the question regarding the willingness to participate in CSA in the future measure an intention from weak to strong, then the three discrete responses can be ordered as follows: no (y = 0), not sure (y = 1), and yes (y = 2) [36,37]. Thus, an ordered probit formulation may be utilized. Following Cameron and Trivedi [35], we first introduced a latent variable y*, defined as
y i * = X i β + ε i
where β is a vector of parameters to be estimated, X is a vector of explanatory variables, and ε is the error term. The observed response categories (y) that are tied to the latent variable satisfy the following model:
y i = { 0 ,   i f   y i *     A 1   1 ,   i f   A 1   y i *     A 2 2 ,   i f   y i *   >   A 2
where A1 and A2 are unknown cutoff parameters to be estimated with β. The probability that consumer i will belong in group j [35] is given by
P r o b   ( y i = j ) = F ( A j   X i β ) F ( A j 1 X i β )
The marginal effect of the rth explanatory variable is calculated as
P r o b   ( y i = j ) X i r = { F ( A j 1   X i β ) F ( A j X i β ) }   β r
where F is the standard normal CDF.
The ordered probit model assumes that the responses are in ordinal scale and that the influence of explanatory variables remains consistent across the range of the dependent variable (the parallel regression assumption) [38]. Because the ordering of the responses to future CSA participation (No = 0, Not Sure = 1, Yes = 2) may be questioned as “crude”, a likelihood ratio test was implemented to test for the parallel regression assumption [39,40]. If the assumption were rejected, other models such as multinomial probit models that do not have the parallel regression assumption would have to be considered [41,42].
Under a multinomial probit model, the probability of selecting alternative j, out of m possible options is given by
P r ( y i = j ) =   V j 1 V j 2 V j M Φ M 1 ( η 1 j , η 2 j , . η m i ) d η 1 j d η 2 j .. d η m i
where Φ() stands for a M-1-variate normal distribution and ηji = εjει.

4. Results

Summary statistics for the sample are reported in Table 1. Among all participants, 53% were female, 77% were Caucasian, and the median age was 37 years old. These numbers compare favorably with the U.S. population demographics of 50.8% female, 77.4% Caucasian, and a median age of 35.3 years old [43]. Our sample is slightly biased toward higher income families. Specifically, the median household income for the sample is US$62,500, compared to the US median of US$53,657. Lastly, 46 responders (6%) were members of a CSA arrangement when the survey was conducted (Table 1).
A standard t-test for comparing means of variables with unequal variances was utilized to gauge differences in characteristics between CSA members and non-members (Table 2). In line with previous studies [12,44,45,46,47], our findings indicate that, on average, CSA members are younger, are more educated, and have higher incomes than do non-members. Furthermore, consistent with [34], our results highlight that CSA members spend more on grocery shopping and have more active lifestyles than do non-members. A potential explanation for the increased expenditures is that CSA members, in contrast to non-members, are more likely to eat more meals at home [19,34]. There was no significant difference between CSA members and non-members in race and time spent in cooking. With the exception of the Northeast, there was no difference between members and non-members (Table 2).
A consistent theme within the CSA literature is that women constitute the majority of CSA members [5,12,48,49,50,51]. However, in line with the results of a recent Food Demand Survey, which indicated that people who shopped at farmers markets were more likely to be male [52], males represent a larger percentage of CSA members in our sample (Table 2). A potential reason for this discrepancy lies in the sample selection. Specifically, the majority of the previous research focuses primarily on CSA members, whereas our survey focuses on a national representative sample of U.S. consumers who may or may not be CSA members. Alternatively, our findings may indicate a shift in the demographics of CSA members.
Table 3 reports summary statistics by responders’ intention to join a CSA arrangement in the future. About 48% of the responders indicated that they were not interested in joining a CSA arrangement in the future, 40% were not sure, and 12% responded that they would join a CSA arrangement in the future. The results also show that the respondents’ intention for future CSA participation may vary by age, gender, income, education, number of children, lifestyle, and region. For instance, respondents who said “Yes” to the future CSA participation question seemed to have more children, were more likely to be vegetarian, and were more likely to live in the South. The impact of these variables will be further discussed when presenting the results of regression analysis.

4.1. Reasons for Participation and Non-Participation in CSA

The importance of the various reasons in the respondents’ decision on whether or not to participate in CSA is reported in Figure 1 and Figure 2, respectively. Consistent with previous research [20,27,51], the findings indicate that: (i) supporting local farms; and (ii) purchasing organic foods are among the strongest motivations for joining a CSA. Specifically, more than 82% of the respondents indicated the aforementioned reasons as somewhat important or very important factors influencing their decisions. Conversely, approximately 39% of the survey respondents indicated that one of the main reasons for not joining a CSA is their preference for farmers markets (Figure 2). This finding implies that other direct marketing schemes such as farmers markets may provide similar products and services as CSA, therefore reducing the need to join a CSA. It also highlights the importance of differentiating CSA from other direct marketing schemes for future CSA development. The limited variety and the cost of CSA membership were the second and third most important factors that discouraged survey respondents from CSA participation.

4.2. Factors Affecting Current CSA Participation

Table 4 reports the estimates of the coefficients and marginal effects for the probit model. The McFadden adjuster R2 is 0.37, and the model correctly classified 96% of the observations (8 observations were misclassified as yes, when the correct classification was no, and 22 observations were misclassified as no when the correct value was yes), indicating a good fit. Consistent with previous research [19,21], the demographic variables did not exert statistically significant effects on the probability that a respondent is a CSA member (Table 4). We selected the variables included in the model based on previous studies. It is not uncommon for studies related to CSA marketing to have coefficients that are not significant. Furthermore, even the insignificant results can provide valuable information. For example, our findings provide further support for the argument that demographic variables are not a significant predictor to identify current CSA membership. However, we still need to include the demographics variables to avoid the model misspecification of “omission of relevant variables”.
Consistent with the findings of Kolodinsky and Pelch [9], our findings indicate that households that recycle more often are more likely to be CSA members (Table 4). A potential explanation for this finding is that CSA members are more likely to exhibit pro-ecological behavior [5]. A number of lifestyle variables had statistically significant and positive impacts on the probability that a survey respondent is a CSA member. For example, respondents who are members of fitness clubs are 4 percentage points more likely to participate in CSA (Table 4). These findings may help CSA managers design more efficient and targeted marketing strategies. However, it may be challenging and expensive to identify this type of health-conscious consumers [21,53]. Respondents who were CSA members in the past are 23 percentage points more likely to be current CSA members. This finding provides further support for the hypothesis that satisfaction substantially influences the probability of CSA membership [9,10,26].
Moreover, the findings indicate that consumers who spend more on grocery shopping are more likely to be CSA members. A potential explanation for this finding is that CSA members tend to eat more meals at home [34]. Contrary to the findings of [53], none of the regional variables had a statistically significant impact on the probability that a responder was a CSA member during the time of the survey.

4.3. Factors Affecting Future CSA Participation

Table 5 and Table 6 report the estimates of the coefficients and marginal effects for the ordered probit model. The McFadden adjusted R2 for the model is 0.14. Moreover, the threshold parameters are statistically different from each other (Table 5).
Consistent with previous research [9,21,54] and our initial hypothesis, most demographics such as age, gender, race, and the number of children did not have a statistically significant impact on the probability that a consumer will join a CSA (Table 5). The most discernable differences between the probit and ordered probit formulations are related to the impact of the income, education, and information variables. For example, the results in Table 5 indicate that responders with higher incomes are less likely to join a CSA arrangement in the future (Table 5). This finding is consistent with Zepeda and Li [19], who highlighted that more affluent consumers are less likely to buy local foods. A couple of reasons justify this result. First, households with higher incomes may eat out more often than do households with lower incomes [19]. Second, higher income may translate to higher opportunity cost and busier schedules, leaving less time to visit a CSA farm [19]. Our results also indicate that, compared to high school graduates, obtaining a graduate degree translates to a lower probability of future CSA participation. More educated consumers are 3.1 percentage points less likely than their less educated counterparts to join a CSA in the future (Table 5). Potential explanations for this finding are that more educated consumers are more likely to have busier schedules [55] or do not value buying local products [56].
Consistent with the findings of the probit analysis, the marginal effects from the ordered probit indicate that respondents who are already CSA members are 27% more likely than are non-CSA members to participate in CSA in the future (Table 6). This finding further highlights the importance of having satisfied members. Consumers who consider CSA food production techniques better for the environment are approximately 4% more likely to join a CSA (Table 6). Consumers who identify themselves as non-vegetarian are 13% less likely to participate in a CSA, which is consistent with the fact that most CSA farms sell no or limited meat products (Table 6).
Regarding the impact of information sources on future CSA participation, the results highlight the importance of communication through family members/word of mouth for the probability of becoming a CSA member, which is in line with the findings of Kolodinsky and Pelch [9]. Specifically, respondents who place a higher value on this information source are approximately 4.5% more likely to join a CSA program in the future. Under the same token, respondents who place a higher importance on websites and newspapers are more likely to join a CSA by 2.3% (Table 6). On the other hand, advertising through road signs did not have a statistically significant impact. These results can provide useful information to CSA managers as they design their marketing campaigns to maintain current memberships and attract new members.
As we discussed in the empirical framework section, the ordered probit model assumes parallel regression assumption. In this study, the likelihood ratio rejected the parallel regression assumption at the 1% significance level. Consequently, other models, such as the multinomial probit model, should be considered. The results of the multinomial probit model are reported in Table 7 (Marginal effects are not reported to save page space because the results from the multinomial probit and the ordered probit models are generally consistent.). The adjusted R2 for the model is 0.314, indicating a better fit compared to that of the ordered probit model. Overall, in line with the ordered probit findings, the results indicate that most demographic characteristics are a weak predictor of responders’ intention to join a CSA arrangement in the future. Consumers, who are currently CSA members, are vegetarians, believe that CSA farm production has a lower environmental impact than does non-CSA production, and recycle more frequently, are more likely to participate in CSA arrangements in the future (Table 7). The results also indicate that the web and word of mouth are the most important information sources for respondents’ future CSA participation (Table 7). The overall consistency in the results of the ordered probit and multinomial probit model indicates the robustness of our conclusions regarding factors affecting consumers’ future CSA participation.

5. Conclusions

A plethora of empirical studies, using primarily region specific surveys, have examined the characteristics of CSA members, and investigated factors affecting CSA membership satisfaction [12,22,26,27,28]. However, as competition for local food expenditures intensifies, a better understanding of the factors that can increase consumers’ future CSA participation is crucial for the continuous success of this direct marketing outlet. The present study tries to fill this gap using a national sample to gain a better understanding of the factors that influence consumers’ decisions to join CSA.
Consistent with previous research endeavors [44,45,46,47], our results show that CSA members are younger, more educated, and with a higher income than non-members. Supporting local farms and purchasing organic foods are the primary motivations for consumers to join a CSA, which is in line with previous research [20,27,51]. Conversely, the ability to buy local foods from farmers’ markets, the high CSA cost, and the limited variety are the main reasons for not joining a CSA. Furthermore, consumers who are more likely to join a CSA in the future appear to be younger, more educated and with a more active lifestyle (e.g., have a gym membership).
The results from a probit, an ordered probit, and a multinomial probit model further demonstrate the impact of consumers’ demographic characteristics, lifestyle preferences, and information outlets on the probability of an individual’s current and future CSA participation. Our results demonstrate that the factors affecting current and future CSA participation differ significantly. Despite the incompetence of all demographic characteristics in predicting the probability that a consumer is currently a CSA member, some of them such as income and education have a significant impact on the probability of the future CSA participation. Consumer lifestyle such as being a vegetarian, pro-environment perception of CSA, recycling behavior, and experience of being a CSA member have a statistically significant positive impact on both the probability that a consumer is currently a CSA member or will become a CSA member in the future. Moreover, although none of the information outlets examined in this study significantly affect an individual’s current CSA participation, web and word-of-mouth sources significantly influence the probability that a consumer will join a CSA in the future. This highlights the importance of attracting new CSA members using the traditional word-of-mouth approach as well as using newer advertising media such as websites.
Overall, our findings indicate that, although CSA consumers appear to be a homogeneous group, there is a difference between the factors that determine current and future CSA membership. Furthermore, the results indicate that producers should question the effectiveness of traditional information outlets (e.g., news and road signs) on the probability of increased membership. To maintain CSA growth, strategies should concentrate on word-of-mouth information and attractive websites. A limitation of this study should be acknowledged. Specifically, the absence of time series data regarding CSA membership may have inhibited the estimation of individual-specific parameters. Further analysis should focus on the factors that increase loyalty among CSA members, similar to what has been investigated in [57]. Additional research can also be conducted to further identify the barriers for future CSA participation. Finally, it has been demonstrated that the structure of, and the reasons for, consumers to shop at the farmers markets differ significantly between the Unites states and Italy [58]. Therefore, it necessary to extend our study by conducting national wide studies in other regions such as EU countries.

Acknowledgment

The authors would like to thank the editors and the anonymous reviewers for helpful comments. This work was supported by the USDA National Institute of Food and Agriculture, (Hatch project 1001888 to Z.G.) Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the National Institute of Food and Agriculture (NIFA) or the United States Department of Agriculture(USDA).

Author Contributions

Michael Vassalos, as the lead author of the paper, co-conceptualized the idea with the second author and was responsible for the model estimation, results interpretation, and writing of the paper. Zhifeng Gao, as the correspondence/contact author of the paper, contributed to the development of the survey, results interpretation, and writing of the paper. Lisha Zhang was responsible for the survey development and survey data cleaning.

Conflicts of Interest

We have no conflict of interest to declare.

References

  1. Shi, R.; Hodges, A.W. Shopping at farmers’ markets: Does ease of access really matter? Renew. Agric. Food Syst. 2015, 31, 441–451. [Google Scholar] [CrossRef]
  2. Nie, C.; Zepeda, L. Lifestyle segmentation of US food shoppers to examine organic and local food consumption. Appetite 2011, 57, 28–37. [Google Scholar] [CrossRef] [PubMed]
  3. Meas, T.; Hu, W.; Batte, M.T.; Woods, T.A.; Ernst, S. Substitutes or complements? Consumer preference for local and organic food attributes. Am. J. Agric. Econ. 2014, 97, 1044–1071. [Google Scholar] [CrossRef]
  4. Bond, J.K.; Thilmany, D.; Bond, C.A. Direct marketing of fresh produce: Understanding consumer purchasing decisions. Choices 2006, 21, 229–234. [Google Scholar]
  5. Uribe, A.L.; Winham, D.M.; Wharton, C.M. Community supported agriculture membership in Arizona. An explanatory study of food and sustainable behaviors. Appetite 2012, 59, 431–436. [Google Scholar] [CrossRef] [PubMed]
  6. Low, S.A.; Adalja, A.; Beaulieu, E.; Key, N.; Martinez, S.; Melton, A.; Perez, A.; Ralston, K.; Stewart, H.; Suttles, S.; et al. Trends in U.S. Local and Regional Food Systems (AP-068); United States Department of Agriculture, Economic Research Service: Washington, DC, USA, 2015.
  7. Woods, T.; Tropp, D. CSAs and the battle for the local food dollar. Food Distrib. Res. Soc. 2015, 46, 17–29. [Google Scholar]
  8. Snell, S.M. Food miles, local eating and community supported agriculture: Putting local food in its place. Agric. Hum. Values 2013, 30, 615–628. [Google Scholar] [CrossRef]
  9. Kolodinsky, J.M.; Pelch, L.L. Factors influencing the decision to join a community supported agriculture (CSA) farm. J. Sustain. Agric. 1997, 10, 129–141. [Google Scholar] [CrossRef]
  10. Harmon, H.A. Community supported agriculture: A conceptual model of health implications. Austin J. Nutr. Food Sci. 2014, 2, 1024. [Google Scholar]
  11. United States Department of Agriculture, National Agricultural Statistics Service (USDA/NASS). 2012 Census of Agriculture—Summary and Stated Data; USDA/NASS: Washington, DC, USA, 2014.
  12. Pole, A.; Gray, M. Farming alone? What’s up with the “C” in community supported agriculture. Agric. Hum. Values 2013, 30, 85–100. [Google Scholar] [CrossRef]
  13. Sproul, T.W.; Kropp, J.D. A general equilibrium theory of contracts in community supported agriculture. Am. J. Agric. Econ. 2015, 97, 1345–1359. [Google Scholar] [CrossRef]
  14. Brown, C.; Miller, S. The impacts of local markets: A review of research on farmers markets and community supported agriculture (CSA). Am. J. Agric. Econ. 2008, 90, 1296–1320. [Google Scholar] [CrossRef]
  15. Connolly, C.; Klaiber, H.A. Does organic command a premium when the food is already local? Am. J. Agric. Econ. 2014, 96, 1102–1116. [Google Scholar] [CrossRef]
  16. Bruch, M.L.; Ernst, M.D. A Farmer’s Guide to Marketing through Community Supported Agriculture (CSAs); University of Tennessee: Knoxville, TN, USA, 2010. [Google Scholar]
  17. Leffew, M.; Ernst, M.D. A Farmer’s Guide to a Pick-Your-Own Operation; University of Tennessee: Knoxville, TN, USA, 2014. [Google Scholar]
  18. Lea, E.; Phillips, J.; Ward, M.; Worsley, A. Farmers’ and consumers’ beliefs about community-supported agriculture in Australia: A qualitative study. Ecol. Food Nutr. 2006, 45, 61–86. [Google Scholar] [CrossRef]
  19. Zepeda, L.; Li, J. Who buys local food? J. Food Distrib. Res. 2006, 37, 1–11. [Google Scholar]
  20. Curtis, K.R.; Allen, K.; Ward, R.A. Food consumption, attitude, and behavioral change among CSA members: A northern Utah case study. Food Distrib. Res. 2015, 46, 3–16. [Google Scholar]
  21. Peterson, H.H.; Taylor, M.R.; Baudouin, Q. Preferences of locavores favoring community supported agriculture in the United States and France. Ecol. Econ. 2015, 119, 64–73. [Google Scholar] [CrossRef]
  22. Pole, A.; Kumar, K. Segmenting CSA members by motivation: Anything but two peas in a pod. Br. Food J. 2015, 117, 1488–1505. [Google Scholar] [CrossRef]
  23. Wilkins, J.L.; Farrell, T.J.; Rangarajan, A. Linking vegetable preferences, health, and local food systems through community supported agriculture. Public Health Nutr. 2015, 18, 2391–2401. [Google Scholar] [CrossRef] [PubMed]
  24. Schmidt, M.C.; Kolodinsky, J.M.; DeSisto, T.P.; Conte, F.C. Increasing farm income and local food access: A case study of collaborative aggregation, marketing, and distribution strategy that links farmers to markets. J. Agric. Food Syst. Community Dev. 2011, 1, 157–175. [Google Scholar] [CrossRef]
  25. Allen, J.E., IV; Rossi, J.; Woods, T.; Davis, A.F. Do community supported agriculture programmes encourage change to food lifestyle behaviours and outcomes? New evidence from shareholders. Int. J. Agric. Sustain. 2017, 15, 70–82. [Google Scholar] [CrossRef]
  26. Lang, K.B. Expanding our understanding of community supported agriculture (CSA): An examination of member satisfaction. J. Sustain. Agric. 2015, 26, 61–79. [Google Scholar] [CrossRef]
  27. Cox, R.; Holloway, L.; Venn, L.; Dowler, L.; Hein, J.R.; Kneafsey, M.; Tuomainen, H. Common ground? Motivations for participation in a community-supported agriculture scheme. Local Environ. 2008, 13, 203–218. [Google Scholar] [CrossRef]
  28. Sharp, J.; Imerman, E.; Peters, G. Community supported agriculture (CSA): Building community among farmers and non-farmers. J. Ext. 2002, 40, 1–9. [Google Scholar]
  29. Vasquez, A.; Sherwood, N.E.; Larson, N.; Story, M. Community supported agriculture as a dietary and health improvement strategy: A narrative. J. Acad. Nutr. Diet. 2017, 117, 83–94. [Google Scholar] [CrossRef] [PubMed]
  30. Onianwa, O.; Wheelock, G.; Mojica, M. An analysis of the determinants of farmer-to-consumer direct market shoppers. J. Food Distrib. Res. 2005, 36, 130–134. [Google Scholar]
  31. Akter, S.; Bennett, J. Preference uncertainty in stated preference studies: Facts and artefacts. Appl. Econ. 2013, 45, 2107–2115. [Google Scholar] [CrossRef]
  32. Kehlbacher, A.; Bennet, R.; Balcombe, K. Measuring the consumer benefits of improving animal welfare to inform welfare labelling. Food Policy 2012, 37, 627–633. [Google Scholar] [CrossRef]
  33. Hanley, N.; Kristorm, B.; Shorgen, J.F. Coherent arbitrariness: On value uncertainty for environmental goods. Land Econ. 2009, 85, 41–50. [Google Scholar] [CrossRef]
  34. Curits, K.R. Are all direct market consumers created equal? J. Food Distrib. Res. 2011, 42, 26–33. [Google Scholar]
  35. Cameron, C.A.; Trivedi, P.K. Microeconometrics Methods and Applications; Cambridge University Press: New York, NY, USA, 2005. [Google Scholar]
  36. Reagan, A.C.; Golob, T.F. Trucking industry demand for urban shared use freight terminals. Transportation 2005, 32, 23–36. [Google Scholar] [CrossRef]
  37. Sengupta, A.; Wang, H.-M. The roles of mass media and personal information sources on adoption of pandemic vaccines. In Ideas in Marketing: Finding the New and Polishing the Old; Kubacki, K., Ed.; Springer: New York, NY, USA, 2014; pp. 239–242. [Google Scholar]
  38. Stoutenborough, J.W.; Sturgess, S.G.; Vedlitz, A. Knowledge, risk, and policy support: Public perceptions of nuclear power. Energy Policy 2013, 62, 176–184. [Google Scholar] [CrossRef]
  39. Long, S.J.; Freese, J. Regression Models for Categorical Dependent Variables Using Stata; Stata Press: College Station, TX, USA, 2014. [Google Scholar]
  40. Wolfe, R.; Gould, W. An approximate likelihood-ratio test for ordinal response models. Stata Tech. Bull. 1998, 42, 24–47. [Google Scholar]
  41. Ari, E.; Yildiz, Z. Parallel lines assumption in ordinal logistic regression and analysis approaches. Int. Interdiscip. J. Sci. Res. 2014, 1, 8–23. [Google Scholar]
  42. Greene, W.H.; Hensher, D.A. Modeling Ordered Choices: A Primer and Recent Developments; New York University: New York, NY, USA, 2008. [Google Scholar]
  43. United States Census Bureau (USCB). 2010 United States Census; USCB: Washington, DC, USA, 2010.
  44. Kane, D.J.; Lohr, L. Maximizing Shareholder Retention in Southeastern CSAs: A Step toward Long-Term Stability; Organic Farming Research Foundation: Santa Cruz, CA, USA, 1997. [Google Scholar]
  45. Henderson, E.; Gussow, J.D.; Van En, R. Sharing the Harvest: A Guide to Community Supported Agriculture; Chelsea Green: White River Junction, VT, USA, 1999. [Google Scholar]
  46. Perez, J.; Allen, P.; Brown, M. Community Supported Agriculture on the Central Coast: The CSA Member Experience (CASFS Research Brief #1); University of California: Santa Cruz, CA, USA, 2003. [Google Scholar]
  47. Bougherara, D.; Grolleau, G.; Mzoughi, N. Buy local, pollute less: What drives households to join a community supported farm? Ecol. Econ. 2009, 68, 1488–1495. [Google Scholar] [CrossRef]
  48. Lang, K.B. The changing face of community-supported agriculture. Cult. Agric. 2010, 32, 17–26. [Google Scholar] [CrossRef]
  49. Chen, W. Perceived value of a community supported agriculture (CSA) working share. The construct and its dimensions. Appetite 2013, 62, 37–49. [Google Scholar] [CrossRef] [PubMed]
  50. Russell, W.S.; Zepeda, L. The adaptive consumer: Shifting attitudes, behavior change, and CSA membership renewal. Renew. Agric. Food Syst. 2008, 23, 136–148. [Google Scholar] [CrossRef]
  51. O’Hara, S.U.; Stagl, S. Global food markets and their local alternatives: A socio-ecological economic perspective. Popul. Environ. 2001, 22, 533–554. [Google Scholar] [CrossRef]
  52. Lusk, J.; Murray, S. Food Demand Survey; Oklahoma State University: Stillwater, OK, USA, 2016. [Google Scholar]
  53. Bond, J.K.; Thilmany, D.; Bond, C.A. What influences consumer choice of fresh produce purchase location. J. Agric. Appl. Econ. 2009, 41, 61–74. [Google Scholar] [CrossRef]
  54. Zepeda, L. Which little piggy goes to market? Characteristics of farmers’ market shoppers. Int. J. Consum. Stud. 2009, 33, 250–257. [Google Scholar] [CrossRef]
  55. Gumirakiza, J.D.; Curtis, K.R.; Bosworth, R. Who attends farmers’ markets and why? Understanding consumers and their motivations. Int. Food Agribus. Manag. Rev. 2014, 17, 65–81. [Google Scholar]
  56. Jekanowski, M.D.; Williams, D.R., II; Schiek, W.A. Consumers’ willingness to purchase locally produced agricultural products: An analysis of an Indiana survey. Agric. Resour. Econ. Rev. 2000, 29, 43–53. [Google Scholar] [CrossRef]
  57. Gao, Z.; Swisher, M.; Zhao, X. A new look at farmers’ markets: Consumer knowledge and loyalty. HortScience 2012, 47, 1102–1107. [Google Scholar]
  58. Vecchio, R. European and United States farmers’ markets: Similarities, differences and potential developments. In Proceedings of the 13th EAAE Seminar a Resilient European Food Industry and Food Chain in a Challenging World, Crete, Greece, 3–6 September 2009.
Figure 1. Most important reasons for Community Supported Agriculture (CSA) participation among CSA members and responders who plan to join CSA in the future.
Figure 1. Most important reasons for Community Supported Agriculture (CSA) participation among CSA members and responders who plan to join CSA in the future.
Sustainability 09 00478 g001
Figure 2. Most important reasons for not participating in a CSA arrangement.
Figure 2. Most important reasons for not participating in a CSA arrangement.
Sustainability 09 00478 g002
Table 1. Description of the variables and summary statistics.
Table 1. Description of the variables and summary statistics.
VariableDescriptionMeanStd. Dev.
Demographic Variables
AgeAge of the respondent43.0818.2
FemaleDummy variable; female = 10.530.49
CaucasianDummy variable; Caucasian = 10.770.41
IncomeAverage annual household income (US$1000)61.4541.8
High SchoolDummy variable; high school graduate = 10.230.42
Some CollegeDummy variable; Attended some college = 10.380.48
GraduateDummy variable; bachelor’s degree or higher = 10.390.48
No. ChildrenNumber of children < 18 years old in the household1.791.11
Lifestyle Variables
VegetarianDummy variable; Vegetarian or vegan = 10.070.25
GymDummy variable; Responder has gym membership = 10.280.45
CookHours spent cooking in an average week5.282.83
ExpenditureGrocery shopping expenditure (US$100)1.410.93
HealthOrganic food has more health benefits: 0 = strongly disagree/disagree; 1 = neutral; 2 = agree/strongly agree1.220.62
IngredientsHow important is the list of ingredients when you purchase food: 0 = not at all/somewhat unimportant; 1 = neither important/nor unimportant; 2 = somewhat important/very important1.620.67
EnvironmentProduction of food from CSA program has lower environmental impact: 0 = strongly disagree/disagree; 1 = neutral; 2 = agree/strongly agree1.450.62
Chain StoreMoney spent at major chain grocery store for grocery shopping139.27139.03
RecycleHow often do you recycle glass, newspaper, cans, and plastic (0 = never, 4 = always)3.141.25
Past MemberDummy variable; 1 if respondent was a CSA member in the past0.060.25
Information Sources
WebImportance of website information in joining a CSA; 0 = not at all/somewhat unimportant; 1 = neither important/nor unimportant; 2 = somewhat important/very important1.1310.78
Word of MouthImportance of friends/family members in decision to join a CSA: 0 = not at all/somewhat unimportant; 1 = neither important/nor unimportant; 2 = somewhat important/very important1.220.78
NewsImportance of newspaper information in decision to join a CSA: 0 = not at all/somewhat unimportant; 1 = neither important/nor unimportant; 2 = somewhat important/very important1.020.78
Road SignsImportance of road sign information in decision to join a CSA: 0 = not at all/somewhat unimportant; 1 = neither important/nor unimportant; 2 = somewhat important/very important1.090.77
Region of Residence
NortheastDummy variable; 1 for responders who reside in the Northeastern states0.200.39
MidwestDummy variable; 1 for responders who reside in the Midwestern states0.210.41
SouthDummy variable; 1 for responders who reside in the Southern states0.360.48
WestDummy variable; 1 for responders who reside in the Western states0.230.41
Dependent Variables
Current CSA memberDummy variable; 1 = respondent CSA member during survey0.120.32
Plan to join a CSAAre you planning to join a CSA in the future?; 0 = No, 1 = Not sure, 2 = Yes0.670.7
Table 2. Test of means, CSA members, and non-members.
Table 2. Test of means, CSA members, and non-members.
VariableCSA Member (n = 46)Non-Member (n = 722)
MeanStd. ErrorMeanStd. ErrorDifferencet-test a
Demographic Variables
Age32.9510.9543.7318.4710.786.38 **
Female0.320.470.540.490.223.04 **
Caucasian0.730.440.770.410.040.56
Income76.5647.3060.4841.27−16.8−2.25 *
High School0.150.360.230.420.081.47
Some College0.260.440.380.480.121.89
Graduate0.580.490.370.48−0.21−2.78 **
Number of Children2.51.111.751.09−0.75−4.46 **
Lifestyle Variables
Vegetarian0.260.440.0540.22−0.207−3.13 **
Gym0.730.440.250.43−0.48−7.20 ***
Cook6.232.75.222.83−1.011−2.45 *
Expenditures2.301.181.350.89−0.951−5.34 ***
Health1.670.521.190.61−0.48−6.05 ***
Ingredients1.470.801.630.660.161.27
Environment1.670.631.430.62−0.24−2.49 *
Past CSA member0.650.480.020.16−0.623−8.74 ***
Chain Store222.08156.03134.00136.30−88.08−3.74 ***
Recycle3.540.783.111.27−0.433−3.48 ***
Regional Variables
Northeast0.340.480.180.39−0.159−2.2 *
Midwest0.150.360.210.410.0670.06
South0.300.160.360.480.0590.03
West0.190.400.220.420.0320.07
a Significance level in this column refers to the difference between non-members and CSA members. *, **, and *** represent significant at the 10%, 5%, and 1% significance levels, respectively.
Table 3. Summary statistics by responder’s intention to join a CSA in the future.
Table 3. Summary statistics by responder’s intention to join a CSA in the future.
VariableY = No (n = 367)Y = Not Sure (n = 311)Y = Yes (n = 90)
MeanSDMeanSDMeanSD
Demographic Variables
Age4618.7842.6317.9832.7412.45
Female0.460.490.610.480.510.50
Caucasian0.810.390.770.420.630.48
Income65.4042.0055.8739.7464.6046.01
High School0.200.400.250.430.240.43
Some College0.350.480.420.490.340.48
Graduate0.440.490.320.460.410.49
Number of Children1.671.041.781.112.31.21
Lifestyle Variables
Vegetarian0.030.180.070.260.180.38
Gym0.260.440.220.410.550.49
Cook4.842.815.652.805.832.84
Expenditures1.320.871.370.891.921.15
Health1.110.621.220.581.630.53
Ingredients1.570.701.670.641.670.67
Environment1.280.641.560.571.730.55
Past CSA member0.040.200.0190.130.320.47
Chain Store131.13129.98135.09143.17186.95151.92
Recycle3.081.273.091.303.50.87
Information Sources
Web0.830.771.260.721.710.56
Word of mouth0.930.791.390.701.790.48
News0.770.741.170.731.580.70
Road Signs0.860.791.200.681.60.59
Regional Variables
Northeast0.190.390.200.400.200.40
South0.340.470.360.480.460.50
West0.210.410.240.430.200.40
Midwest0.250.430.190.390.140.35
Table 4. Probit estimation results and marginal effects.
Table 4. Probit estimation results and marginal effects.
Dependent Variable = CSA MemberEstimation ResultsMarginal Effects
CoefficientStd. ErrorCoefficientStd. Error
Constant−4.625 ***0.539
Demographic Variables
Age−0.0100.010−0.00050.0006
Female−0.1680.244−0.00930.0124
Caucasian0.1300.2730.00680.0153
Income−0.0020.003−0.00010.0001
Some College a0.1910.3490.01060.0209
Graduate a0.3720.3470.02140.0231
Number of Children0.1460.0980.00790.0054
Lifestyle Variables
Vegetarian0.692 **0.3260.04980.0304
Gym0.670 ***0.2440.0393 **0.0182
Cook0.0640.0470.00340.0025
Expenditures0.289 **0.1170.0157 **0.0063
Health0.1610.2140.00870.0117
Ingredients−0.2580.169−0.01410.0094
Environment0.0200.1990.00100.0108
Past CSA member1.791 ***0.2690.2271 ***0.0614
Chain Store0.0000.0010.0000.0000
Recycle0.258 *0.1440.0141 *0.0081
Information Sources
Web0.0380.2000.00200.0109
Word of mouth0.2330.2040.01270.0113
News0.2610.2290.01430.0127
Road−0.0940.234−0.00510.0127
Regional Variables b
Northeast0.1830.3590.01040.0220
South−0.2470.351−0.01300.0166
West−0.1810.397−0.00940.0191
Adjusted McFadden R20.37
Percent Correctly Classified96%
a High school graduate is the base category. b Midwest is the base category. *, **, and *** represent significant at the 10%, 5%, and 1% significance levels, respectively.
Table 5. Ordered probit estimation results of consumers’ intention to join a CSA in the future.
Table 5. Ordered probit estimation results of consumers’ intention to join a CSA in the future.
Dependent Variable = Plan to Join a CSACoefficientStd. Error
Demographic Variables
Age−0.0010.003
Female0.1350.094
Caucasian−0.1260.111
Income−0.002 *0.001
Some College a−0.0810.117
Graduate a−0.245 *0.129
No. of Children0.0100.045
Lifestyle Variables
Vegetarian0.356 **0.178
Gym−0.0060.106
Cook0.0260.016
Expenditures0.0700.053
Health0.0710.080
Ingredients0.0010.072
Environment 0.302 ***0.078
Chain Store0.0000.000
Recycle0.062 *0.038
Current CSA Member1.135 ***0.209
Information Sources
Web0.186 **0.081
Word of Mouth0.349 ***0.082
News0.175 **0.084
Road Signs−0.0530.089
Regional Variables b
Northeast0.1450.143
South0.1600.125
West0.1380.138
Threshold Parameter
A11.518 ***0.296
A23.098 ***0.309
Adjusted McFadden R20.144
a High school graduate is the base category; b Midwest is the base category; *, **, and *** represent significant at the 10%, 5%, and 1% significance levels, respectively.
Table 6. Ordered probit marginal effects for the probability of joining CSA in the future.
Table 6. Ordered probit marginal effects for the probability of joining CSA in the future.
VariableY = NoY = Not SureY = Yes
CoefficientStd. ErrorCoefficientStd. ErrorCoefficientStd. Error
Demographic Variables
Age0.0000.001−0.0000.001−0.0000.000
Female−0.0530.0370.0360.0250.0170.012
Caucasian0.0500.044−0.0330.028−0.0170.016
Income0.001 *0.000−0.001 *0.000−0.000 *0.000
Some College a0.0320.047−0.0220.032−0.0100.015
Graduate a0.097 *0.051−0.067 *0.036−0.031 *0.016
No. of Children−0.0040.0180.0030.0120.0010.006
Lifestyle Variables
Vegetarian−0.137 **0.0660.080 **0.0310.0580.035
Gym0.0020.042−0.0020.028−0.0010.014
Cook−0.0100.0060.0070.0040.0030.002
Expenditures−0.0280.0210.0190.0140.0090.007
Health−0.0280.0320.0190.0210.0090.010
Ingredients−0.0000.0290.0000.0190.0000.009
Environment −0.120 ***0.0310.081 ***0.0220.039 ***0.011
Chain Store−0.0000.0000.0000.0000.0000.000
Recycle−0.025 *0.0150.017 *0.0100.008 *0.005
Current CSA Member−0.369 ***0.0460.095 ***0.0350.274 ***0.074
Information Sources
Web −0.074 **0.0320.050 **0.0220.024 **0.011
Word of Mouth−0.139 ***0.0330.094 ***0.0230.045 ***0.011
News−0.070 **0.0330.047**0.0230.023 **0.011
Road Signs0.0210.035−0.0140.024−0.0070.012
Regional Variables b
Northeast−0.0570.0560.0370.0350.0200.021
South−0.0630.0490.0420.0320.0210.018
West−0.0550.0540.0360.0340.0190.020
a High school graduate is the base category; b Midwest is the base category; *, **, and *** represent significant at the 10%, 5%, and 1% significance levels, respectively.
Table 7. Multinomial probit results for responders’ intention to join a CSA in the future.
Table 7. Multinomial probit results for responders’ intention to join a CSA in the future.
VariableNot Sure vs. NoYes vs. No
CoefficientStd. ErrorCoefficientStd. Error
Demographic Variables
Age0.0030.005−0.0130.008
Female0.281 *0.1490.1870.224
Caucasian−0.0090.183−0.436 *0.242
Income−0.004 *0.002−0.0020.003
Some College a−0.0230.189−0.1940.279
Graduate a−0.3230.206−0.3990.300
No. of Children−0.0170.0750.1170.097
Lifestyle Variables
Vegetarian0.0030.3160.694 *0.382
Gym0.281 *0.1710.2230.231
Cook−0.0090.0260.0220.038
Expenditures−0.004 *0.0870.1060.118
Health−0.0230.1290.322 *0.193
Ingredients−0.3230.1160.0600.180
Environment−0.0170.1230.465 **0.190
Chain Store0.0030.001−0.0000.001
Recycle0.281 *0.0590.263 **0.105
Current CSA Member−0.0090.4481.439 ***0.373
Information Sources
Web 0.2020.1310.513 ***0.190
Word of Mouth0.482 ***0.1310.669 ***0.196
News0.368 ***0.1370.1460.182
Road Signs−0.2000.1450.0620.197
Regional Variables b
Northeast0.381 *0.2280.1540.359
South0.0840.1990.4240.305
West0.3080.2160.1540.352
Constant−2.022 ***0.462−5.062 ***0.799
Adjusted R20.314
a High school graduate is the base category. b Midwest is the base category. *, **, and *** represent significant at the 10%, 5%, and 1% significance levels, respectively.
Sustainability EISSN 2071-1050 Published by MDPI AG, Basel, Switzerland RSS E-Mail Table of Contents Alert
Back to Top