Introduction

Food safety is a major public issue worldwide (Krom, 2009). Green et al. (2003) and Bailey & Garforth (2014) suggested that the traditional governance model has been difficult to effectively meet the social demand for food safety. Thus, it is the common responsibility of all stakeholders to ensure food safety (FAO/WHO, 2003). Since the 1960s and 1970s, as a new governance model, social co-governance has emerged and become the most basic mode of food safety risk management in Western countries (Rouviere & Caswell, 2012). Despite having different national conditions from those in Western countries, the Chinese government has begun to reform its governance model since 2013 (Wu et al., 2021). However, the development of a food safety risk management system in China is currently facing challenges, and one major challenge is the low willingness of citizens to participate.

PsyCap was considered difficult to measure due to its instability, variability, and tendency to undergo dynamic changes (Allen & Potkay, 1981). Only after Goldsmith et al. (1997) found a correlation between an individual’s PsyCap and income did PsyCap begin to be measurable, developable, and manageable. PsyCap is a positive psychological state, and it serves as an important resource in the growth and development of citizens. It provides a new theoretical perspective for investigating and improving citizens’ willingness to participate to promote the establishment of a food safety social co-governance system (Meng et al., 2021). Studies carried out in management and other fields have repeatedly verified that PsyCap has an important impact on citizens’ perceptions, attitudes, and behaviors (Rego et al., 2019). However, to date, no research has examined citizens’ PsyCap in the field of food safety. Does citizens’ PsyCap affect their willingness to participate? If so, what is the path of influence? Answering these questions will not only help fill some of the gap in existing research and promote the application of the PsyCap theory in the field of public affairs, but also help promote the establishment of a food safety social co-governance system in practice.

Citizens with independent behavior are not mere food consumers and should not only bear the responsibility of self-protection (Houghton et al., 2008), but can also be the best regulators of food safety. As such, they hold potential to serve as an important force in social co-governance, playing an irreplaceable role (Qiu & Zhao, 2019). The fact that citizens’ PsyCap affects their attitudes and behaviors, has reached basic consensus in psychological theory (Donaldson et al., 2020). Indeed, there are great differences in cognitive models among individual citizens. For instance, individuals with high PsyCap have abundant psychological resources and are usually able to analyze their surrounding environment. They are more motivated to explore actively when encountering difficulties (Luthans et al., 2007; Xiao, 2021). Jang (2016) suggested that citizens with high PsyCap are more motivated to engage in public affairs. PsyCap also has the function of igniting individual positive emotions and encouraging individuals to actively participate in social affairs (Chen & Wu, 2019). At present, it is generally agreed that self-efficacy, resilience, optimism and hope are the four core elements that constitute PsyCap. These elements are developable psychological states that can promote positive perceptions and action tendencies (Luthans & Youssef-Morgan, 2017).

Self-efficacy is an individual’s convictions (or confidence) about his or her personal abilities to mobilize the motivation, cognitive resources, and courses of action needed to successfully execute a specific task within a given context (Stadjkovic et al., 1998). Individuals with high self-efficacy can better employ positive emotion-regulation strategies and are more confident in their response to challenges, making them more likely to help others and serve society (Eisenberg et al., 1989). Citizens with higher self-efficacy have higher expectations for public utilities, are often confident in social and public services, and are willing to exert greater effort for achievements (Paglis & Green, 2002).

Resilience is the ability of individuals, families, or groups to quickly recover from adversity, setbacks, and failures and actively change their mentality (Masten, 2001). Individuals with high resilience usually find a positive perspective on things and strive to maintain a cognitive and emotional state of optimism, perseverance, and active coping in the face of negative events or difficulties, which helps to stimulate their motivation and behavior of volunteering to serve others (Jiang et al., 2016).

Optimism is a positive attitude towards current and future success expectations (Luthans, 2002). Optimism motivates and guides individuals in distress to take positive actions and find the right response, and increases the possibility of their influence on policies and practices (Christopher & Alistair, 2018). Optimistic individuals have a tendency for positive, problem-oriented coping when faced with stress (Billingsley et al., 2016). Citizens with high optimism are also more likely to bear realistic pressures and challenges and take direct action to overcome adversity and achieve their goals (Crane & Crane, 2007).

Hope not only provides lasting beliefs and positive expectations for individuals in terms of their potential to achieve their goals, but also motivates their sustained efforts (Lopez et al., 2003). Hope is also a powerful resource for social change. When citizens share a common vision for social change and take concerted action to achieve common goals, hope can become the driving force for social change (Bar-Tal, 2001). Hope is also crucial for motivating citizens to actively participate in environmental governance (Lueck, 2007). Ojala (2012) argued that hope influenced citizens’ behavior in environmental governance, and citizens with higher hope had a higher tendency to act in favor of environmental protection.

PsyCap, as an important variable that affects citizens’ attitudes and behaviors, has been verified by many researchers and been widely used in many fields, but it has not yet been universally verified in citizen participation in public affairs (Youssef-Morgan & Dahms, 2016). PsyCap seems to be a core construct in which the whole is greater than the sum of its parts. The participants experience outcomes that are greater than the sum of the four parts (Luthans et al., 2006). Luthans et al. (2007) argued that self-efficacy, resilience, optimism and hope may have a common core that was labeled as PsyCap that can be measured and related to performance and satisfaction and the overall Psycap level may be a better predictor of performance and satisfaction than the four dimensions.

To this end, drawing on existing literature, this study constructed an analytical framework (Fig. 1) based on the general consensus in the academic community that self-efficacy, resilience, optimism, and hope are the four dimensions of PsyCap. PsyCap composed of self-efficacy, resilience, optimism and hope shows different types due to the differences of citizens’ characteristics. Therefore, this paper attempts to classify citizens into different latent groups by latent profile analysis, and further uses analysis of variance to examine the differences of PsyCap and its four dimensions among different latent groups. In addition, usefulness analysis can be used to assess the degree of influence of PsyCap and its four dimensions on citizens’ willingness to participate in food safety social co-governance and the difference in the degree of influence.

Fig. 1: Framework for analyzing the relationship between citizens’ PsyCap and their willingness to participate in food safety social co-governance.
figure 1

The research framework. Self-efficacy, resilience, optimism and hope are the four core elements that constitute PsyCap. Latent profile analysis applied to classify citizens into different latent groups and usefulness analysis can be used to assess the degree of influence of PsyCap and its four dimensions on citizens’ willingness to participate.

Full size image

Methods

Survey implementation

The data for this study were collected using a questionnaire survey. Wuxi in Jiangsu Province is one of the leading prefecture-level cities in China in terms of economic and social development. In 2020, the per capita GDP of Wuxi ranked first among all Chinese cities. Ordinary people enjoy relatively balanced living conditions and their satisfaction with food safety is generally relatively high. Therefore, taking the citizens of Wuxi as the research sample provides a good foundation for this study. Sample data were collected from a field questionnaire survey in the five administrative districts of Wuxi, including Liangxi, Xishan, Huishan, Binhu, and Xinwu districts. The sample size was directly proportional to the resident population of each administrative district.

The survey was conducted in farmers’ markets and supermarket chains with high traffic in each district, and the respondents were citizens over 18 years old (hereinafter referred to as respondents). Investigators were instructed to select the third person coming into view as a respondent to ensure the randomness of sampling as much as possible (Wu et al., 2012). Questionnaires were filled out anonymously by respondents at the survey site and collected upon completion. The survey occurred during December 5–10, 2020. In total, 752 valid samples were obtained.

Sample characteristics

The individual characteristics of respondents are given in Table 1. As demonstrated in Table 1, there were more females than males in the sample, which is consistent with the fact that most household food purchasers in China are female. Most respondents were aged 26–45 (66.09%), had a bachelor’s degree (40.96%), and had a personal annual income of more than 100,000 yuan (26.46%). It should be noted that males and females accounted for 51.96% and 48.04% of the Wuxi urban population in 2020, respectively, and individuals aged 25 years and below, between 26–40 years, and over 41 years accounted for 12.17%, 53.89%, and 33.94%, respectively (Wuxi Bureau of Statistics, 2021). These are generally consistent with the overall demographics of China. The demographics of the sample in this study are not exactly the same as the overall demographics of Wuxi. However, it cannot be completely judged that the survey samples in this study are not representative, because the food in the general family is mostly purchased by one or several family members, and most of them are female.

Table 1 Individual characteristics of respondents.
Full size table

Main variables

Dependent variable

The dependent variable in this study is citizens’ willingness to participate in food safety social co-governance. Governance is here defined as the totality of the many ways that individual citizens and institutions, both public and private, participate in the management of public affairs. It is seen as an ongoing process through which diverging interests may be accommodated and cooperative action may be taken (Commission on Global Governance, 1995). Willingness is essentially individual citizens’ volitional modality towards a certain thing or psychological activity of an agent (Coates, 1983). Therefore, willingness only exists in citizens’ subjective consciousness and is a virtual state that does have a physical existence (Langacker, 1991). Existing research on citizens’ willingness to participate in the governance of public affairs has so far mainly focused on environmental governance and other public affairs (Bisung et al., 2014; Qu et al., 2020), whereas little research has dealt with citizens’ willingness to participate in food safety social co-governance. Based on the existing literature, this study defines citizens’ willingness to participate in food safety social co-governance as “the behavioral tendency of citizens to actively exercise their legal rights to protect their own food safety and fulfill their responsibilities for managing food safety risks in accordance with the law while taking safeguarding the public interest of food safety as their basic duty.” Based on this definition, we specifically included the question “Are you actively participating in food safety social co-governance?” in the questionnaire. Values of 1, 2, 3, 4, and 5 were assigned to responses of “very unwilling,” “unwilling,” “moderate,” “willing,” and “very willing,” respectively.

Independent variable

The independent variable in this study is PsyCap. With reference to Luthans et al. (2007b), we developed a 16-item scale to measure citizens’ PsyCap for participating in food safety social co-governance based on the four-dimensional composition of PsyCap generally recognized by the academic community. We also revised the scale based on a pre-survey. In general, a Cronbach’s alpha of >0.8, >0.7, and >0.6 indicates very good, good, and acceptable reliability of the scale, respectively. However, a value lower than 0.6 indicates that the scale needs to be revised (Nunnally, 1978). As indicated in Table 2, the Cronbach’s alpha of PsyCap scale for this study is 0.750, indicating that the developed scale is valid.

Table 2 Reliability of the PsyCap scale.
Full size table

Control variables

As indicated by previous studies, variables such as gender, age, education, and personal annual income may all impact citizens’ willingness to participate in food safety social co-governance (Yan, 2012; Wang & Jiang, 2020). However, because these variables are not the focus of this study, they were included as control variables to avoid interference.

The descriptive statistics and relevant calculation data of the main variables involved in this study are presented in Table 3. It can be found that there was a significant pairwise correlation between the main variables.

Table 3 Descriptive statistics and correlation coefficients of main variables.
Full size table

Results and discussion

Common method bias test

The questionnaire used in this study was designed by the authors based on literature research and PsyCap components. All questions in the questionnaire were answered by respondents after independent consideration. Therefore, a common method bias might be present in the measurement of the questionnaire items. The common method bias is a systematic error (Podsakoff et al., 2003). Therefore, it is necessary to perform process control and Harman’s single-factor test (Podsakoff et al., 2000) based on data analysis. In other words, it is necessary to perform unrotated principal component analysis on all variables at the same time. If multiple common factors are obtained and the variance explained by the first common factor does not exceed 40%, the common method bias can be ignored (Ashford & Tsui, 1991). The results indicated that four common factors with eigenvalues greater than 1 were extracted for the PsyCap scale, and the variance explained by the first common factor was only 21.66%, which is less than the critical value of 40%. Therefore, there is no substantial common method bias among the items of the PsyCap questionnaire.

Latent profile analysis of citizen PsyCap

LPA is a statistical method that analyzes the differences between individuals and thereby classifies them into different groups (Qiu, 2008). Bouckenooghe et al. (2019) and Wu et al. (2021) suggested that LPA can be used to classify individuals by PsyCap. Moreover, it performs better than traditional clustering methods. Therefore, this study used LPA to investigate the differences in citizens’ willingness to participate in food safety social co-governance and classified them into groups accordingly. To this end, using self-efficacy, resilience, optimism, and hope, the four dimensions that constitute PsyCap as indicators, fit indices of 1–4 classes were extracted for model fitting analysis of LPA.

According to Peugh & Fan (2013), there are three types of fit indices in LPA: information criteria, classification criteria, and likelihood ratio test derivatives. The commonly used information criteria are log likelihood (LL), Akaike information criterion (AIC), Bayesian information criterion (BIC), and sample-size adjusted BIC (SSA-BIC). The smaller the absolute value of the criterion, the better the model fit (Muthén & Muthén, 2010). The commonly used classification criterion is entropy. The closer it is to 1, the more accurate the individual classification (Carragher et al., 2009). The commonly used likelihood ratio test derivatives are Lo-Mendell-Rubin (LMR) and Bootstrap likelihood ratio test (BLRT). If they reach a significant level, the model fits well (Muthén & Muthén, 2010). The LPA fit indices for different numbers of classes of citizen PsyCap are given in Table 4. It can be seen that the four information criteria, LL, AIC, BIC, and SSA-BIC, had the largest absolute values in the 1-class model. In the 2-class model, LL, AIC, BIC, and SSA-BIC had relatively large absolute values, and the smallest entropy was obtained. In the 4-class model, LL, AIC, BIC, and SSA-BIC had the smallest absolute values, but LMR did not reach the level of significance. In contrast, in the 3-class model, LL, AIC, BIC, and SSA-BIC had relatively small absolute values, the entropy value was relatively large, and both LMR and BLRT reached the level of significance. Therefore, the 3-class model is found to be the best model to use to classify individuals by PsyCap.

Table 4 LPA fit statistics for citizen PsyCap.
Full size table

According to the LPA results in Table 4, citizens can be divided into three different latent groups by PsyCap. The mean values of the four dimensions of PsyCap for the first group (group I) were between 2.39 and 2.83, which are lower than those of other groups. Hence, group I can be called the “low PsyCap group.” In this study, 19 respondents fell in group I, accounting for 2.53%. The mean values of the four dimensions of PsyCap for the second group (group II) were between 3.30 and 3.58, which are in the middle range among the three groups. A total of 460 respondents were in group II, accounting for 61.17%. This can be called the “medium PsyCap group.” The mean values of the four dimensions of PsyCap for the third group (group III) were between 3.82 and 4.26, which are higher than those of groups I and III. A total of 273 respondents were found to fall into group II, accounting for 36.30%. This can be called the “high PsyCap group.”

As indicated in Table 5, there were significant differences in the mean values of the four dimensions of PsyCap among the three groups of citizen groups composed of different individuals (p = 0.00). The post hoc test showed that the mean values of each dimension of PsyCap were significantly higher in group III than in group II, and those in group II were significantly higher those in group I. Figure 2 was plotted based on the above results. It can be seen that there is no intersection between the dimensions of PsyCap of groups I, II, and III, and they show similar morphological trends.

Table 5 Analysis of variance of each dimension of PsyCap of the three groups.
Full size table
Fig. 2: The response probability of each dimension of PsyCap of the three groups.
figure 2

The each dimension of PsyCap were significantly higher in group III than in group II, and those in group II were significantly higher those in group I. Three groups of citizen groups composed of different individuals show similar morphological trends.

Full size image

Based on the difference in the mean values of the four dimensions of PsyCap, citizens were divided into three groups as indicated in Table 5. Do differences exist in the individual characteristics of citizens among the three groups? Or did citizens in the same group have similar individual characteristics? This information is valuable for identifying the individual characteristics of citizens who are more willing to participate in social co-governance. To this end, we constructed Eq. (1) for the multiple logistic regression model on the basis of LPA.

Multiple logistic regression was performed with the PsyCap groups as the dependent variable, and with gender (female as the reference group), age (>55 years of age as the reference group), education (master’s degree or higher as the reference group), and personal annual income (>100,000 yuan as the reference group) as independent variables, respectively. Groups II and III were compared with group I, respectively, to examine the differences in individual characteristics of citizens with different levels of PsyCap. The odds ratios (OR) from multiple logistic regression in Table 6 indicate that education had an impact on the PsyCap of all groups. However, gender, age, and personal annual income did not have a significant impact. Compared with group I, the probability of having a bachelor’s degree is 4.05 and 5.22 times that of having a master’s degree or higher education in groups II and III, respectively. Therefore, citizens with a bachelor’s degree may have higher PsyCap and are more likely to be in groups II and III. In contrast, gender, age, and annual income had no significant influence on an individual’s PsyCap classification.

$${\mathrm{Logit}}\left[ {P\left( {Y = j} \right)} \right] = In\left\{ {\frac{{P = \left( {Y = j} \right)}}{{P = \left( {Y = i} \right)}}} \right\} = \beta _0 + \beta _{1j}X_j + \cdots + \beta _{nj}X_n + \varepsilon _j;j \,\ne\, i$$
(1)

where \({\mathrm{Logit}}\left[ {P\left( {Y = j} \right)} \right]\) is the probability of Y = j; X is the independent variables, including gender, age, education, and personal annual income; β0 is the constant term; βij is the coefficient to be estimated (i = 1, 2, ⋯, n); and εj is the random error term.

Table 6 Multiple logistic regression of individual characteristics on different PsyCap groups.
Full size table

Usefulness analysis of citizens’ PsyCap and willingness to participate

Usefulness analysis can be used to assess the degree of influence of different independent variables on the dependent variable and the difference in the degree of influence, especially for estimating the significant contribution that may be made to variance changes by adding a new variable (Judge & Bono, 2001). Accordingly, in this study, with PsyCap and its various dimensions as independent variables and citizens’ willingness to participate in food safety governance as the dependent variable, we assessed the differences in the influence of PsyCap and its various dimensions on the willingness to participate in co-governance exhibited by different groups of citizens by usefulness analysis based on the LPA results. To this end, the following regression model is established:

$$Y = \beta _0 + \beta _1X_1 + \beta _2X_2 + \beta _3X_3 + \beta _4X_4 + \varepsilon$$
(2)

where Y is citizens’ willingness to participate in co-governance; Xi (i = 1, 2,…, 4) is gender, age, education, and personal annual income in order; βi (i = 1, 2,…, 4) is the corresponding coefficient; β0 is the constant term; and \(\varepsilon \sim N\left( {0,\sigma ^2} \right)\) is the random error.

In the usefulness analysis, gender, age, education, and personal annual income are control variables. We selected one of the four independent variables, i.e., self-efficacy, resilience, optimism, and hope, and entered it into the regression equation first to assess the multiple correlation (R) with the willingness to participate in co-governance. We then introduced PsyCap into the regression equation on this basis to measure the change in the multiple correlation (△R). Next, we switched the order of entering the variables into the regression equation, that is, PsyCap was first entered into the regression equation to calculate the R value of the model, and then the independent variable that was selected in the first step was entered into the equation to obtain the value of △R. Finally, we compared the △R values from the two regressions. The larger the △R value, the greater the influence of the selected independent variable on the dependent variable (Lepak & Snell, 1999). Therefore, four variable orders, A, B, C, and D, were established for the four independent variables, self-efficacy, resilience, optimism, and hope, being selected, and combined with the PsyCap variable, respectively, to perform usefulness analysis according to the aforementioned method.

The results are presented in Table 7. In the low PsyCap group using variable order A, we first entered self-efficacy into the regression equation (this phase is represented by 1). At this point, R was 0.53, but it was not significant at the 0.05 level, indicating that self-efficacy did not affect willingness to participate. Next, we introduced PsyCap into the equation (this phase is represented by 2). At this point, R was 0.84. Hence, △R = 0.31, which is significant at the 0.05 level, indicating that the model can explain the difference in the variance after introducing PsyCap.

Table 7 Usefulness analysis results of different PsyCap groups.
Full size table

To compare differences in the influence of PsyCap and self-efficacy on willingness to participate in co-governance, we reversed the above calculation order, that is, PsyCap was first entered into the regression equation (also denoted by 1). At this point, R was 0.82, which was significant at the 0.01 level, indicating that PsyCap significantly affected willingness to participate. Subsequently, we included self-efficacy in the equation (also denoted by 2). At this point, R = 0.25 and △R = 0.02. As this is significant at the 0.05 level, it indicates that self-efficacy can explain the difference in variance. We then compared the two △R values obtained by entering PsyCap and self-efficacy into the regression model in different orders. It is obvious that the former (△R = 0.31) is greater than the latter (△R = 0.02), indicating that PsyCap had more influence than self-efficacy on willingness to participate. Similarly, we entered the three other independent variables, namely resilience, optimism, and hope, into the regression model with PsyCap in different orders, respectively. The results of this step indicated that the influence of PsyCap on willingness to participate was greater than the influence of resilience, optimism, and hope.

Moreover, as indicated by the regression results in Table 8, resilience had the greatest impact on citizens’ willingness to participate in co-governance in groups I (β = 0.547) and II (β = 0.356), and hope had the greatest impact on citizens’ willingness to participate in co-governance in group III (β = 0.404). Therefore, to improve willingness to participate in social co-governance, it is necessary to strengthen the qualities of resilience and hope for different individual citizens, as well as make up for the deficiencies of self-efficacy and optimism.

Table 8 Regression of each dimension of PsyCap on willingness to participate in co-governance in different groups.
Full size table

Conclusion

Through the analysis reported in the preceding sections, the following main conclusions are drawn: First, citizens had significantly different levels of PsyCap for participating in food safety social co-governance. These levels could be classified into three latent groups, i.e., low, medium, and high PsyCap groups. Based on the sample data of the survey, it can be considered that ~60% of citizens are in the medium PsyCap group. In addition, citizens with a bachelor’s degree are more likely to be in the medium or high PsyCap groups. Second, PsyCap significantly affected citizens’ willingness to participate. Compared with citizens in the low PsyCap group, those in the medium and high PsyCap groups had stronger willingness to participate in co-governance. Third, PsyCap and each of its dimensions all had an impact on citizens’ willingness to participate, but PsyCap had a greater impact than any single dimension. It is worth noting that resilience had the greatest impact on citizens’ willingness to participate in the low and medium PsyCap groups, whereas hope had the greatest impact in the high PsyCap group.

The PsyCap of citizens can be increased. Therefore, in the process of promoting citizens’ participation in food safety social co-governance in China, an interactive communication platform among citizens, the government, and the market should be established to provide channels for citizens to actively express their food safety demands and obtain food safety information, create a favorable environment for all of society to participate in social co-governance, and enhance the PsyCap of the entire population. Moreover, considering the different PsyCap levels in different groups, the government should promote education, especially higher education, focus on guiding different groups of citizens to improve their resilience and hope, and make up for the obvious deficiencies of their PsyCap, thereby improving the PsyCap of all citizens for participating in social co-governance.