Impact of Household Registration Status on Income and Attitudes towards Society in China ()
1. Introduction
Social equality, as an indicator of stable and inclusive social development, is based on the extent to which diverse population groups have equitable access to resources, opportunities, and welfare. In China, rural-urban disparities have historically been influenced by hukou system, or a household registration system, which is a fundamental governmental framework in Chinese society (Wu & Treiman, 2007).
Such an institutional structure shapes social stratification by legally assigning individuals to rural or urban origins—thereby regulating access to education, employment, healthcare, and related domains. Rural hukou holders, however, frequently encounter systemic barriers that constrain their mobility—and, in turn, diminish their quality of life. Consequently, the interaction between hukou and social equality constitutes a matter of considerable scholarly as well as policy concern.
To our knowledge, dozens of studies have sought to reveal the multifaceted impacts of hukou on social equality—creating valuable insights into its far-reaching consequences. Studies have documented how rural hukou status correlates with heightened risks of precarious employment (Wu & Treiman, 2007); restricted access to higher education (Hung, 2022); and diminished social capital (Lyu et al., 2022). Comparative analyses further reveal that—although class structure now exerts greater influence on workers’ earnings nationwide—hukou stratification remains highly consequential, particularly in inland regions characterized by lower levels of marketization. Moreover, research has explored heterogeneities in hukou’s effects, noting variations across gender, conversion pathways (active vs. passive), and regional development levels (Jia, 2024).
Despite these earlier contributions, two critical gaps persist in the literature—limiting a comprehensive understanding of hukou’s role in shaping social equality. First, much of the existing research relies on data from earlier decades (e.g., Wu & Treiman, 2007)—or focuses on specific time points—thus failing to capture the dynamics of recent hukou reforms and their impacts under rapid urbanization. This temporal lag raises questions about the contemporary relevance of findings, as policy shifts and social changes may have altered the patterns of systematic stratification. The second gap lies in the lack of comprehensive analysis: previous studies have predominantly examined the pairwise relationships between hukou status and other factors; however, they have not simultaneously investigated the interconnections among these factors.
Addressing these gaps requires a renewed investigation of hukou’s contemporary significance. This study, therefore, leverages recent nationally representative survey data (CGSS 2021) and utilizes a multidimensional framework—one that integrates institutional, structural, and individual-level factors. It seeks to assess the persistence of hukou’s economic effects on income as well as test whether its impact on social attitudes remains robust—or whether other determinants—such as political identity, education, and economic resources—now exert greater influence.
2. Literature Review
2.1. The Household Registration System and Urban-Rural
Differentiation
The hukou system, or a household registration system—a foundational institutional framework in China—has evolved from the initial role in population statistics into a mechanism for resource allocation and social control, which has long shaped social stratification, economic behavior, and labor market outcomes in China. To be more specific, the hukou system functions in education, employment and healthcare and social welfare through resource allocation mechanisms.
Studies have documented how rural hukou status correlates with higher risks of precarious employment (Wu & Treiman, 2007), restricted access to higher education (Xiao & Bian, 2018; Hung, 2022), and diminished social welfare such minimum living allowances (Cai, 2011; Shi et al., 2024).
Even after controlling for human capital variables like education and work experience, urban hukou holders earn significantly more than rural counterparts (Wu & Wallace, 2021). Chen et al. (2023) propose that the gap might be explained by cumulative effects of occupational barriers and welfare disadvantages. Based on the above analysis, the following hypothesis is proposed:
H1: the current hukou type significantly affects personal income.
2.2. Hukou System and Attitudes Towards Society
2.2.1. The Impact of Household Registration Types on Social Attitudes
The hukou system shapes systematic differences in social attitudes between urban and rural residents—primarily through differential resource allocation. For example, previous studies have shown that rural hukou holders are more likely to perceive unfairness in social distribution (Hung, 2022; Lyu et al., 2022), whereas urban hukou holders exhibit higher satisfaction with social order (Wu & Treiman, 2007).
From an international comparative perspective, China’s hukou system shares functional similarities with identity-based stratification systems in other countries, all influencing social attitudes through identity segmentation. For instance, the social alienation of European immigrants due to legal status restrictions is similar to the lack of belonging among Chinese rural migrants (Chan, 2019). But the uniqueness of the hukou system lies in its deep integration with the socialist welfare system: urban hukou not only symbolizes social status but links to social resources, making its impact on social attitudes far stronger than general identity markers (Wu, 2019).
Beside Hukou, previous research also finds other factors affecting social attitude in China. Lyu & Chai (2024) show that the internet tends to liberalize sexual views and widen attitudinal gaps across age groups, while television generally preserves generational consensus on issues like premarital sex. However, both media reinforce conservative views on extramarital sex, with excessive television use further deepening divides between young people and older generations. Wu & Wang (2024), based on the 2021 Chinese General Social Survey, identifies five dimensions of well-being among young adults in China—physical and mental health, positive outlook, sense of purpose, personal growth, and self-fulfillment—indicating that health is the foundation of happiness. It further shows that poor health reduces perceptions of fairness and trust, while a positive outlook strengthens them, suggesting that well-being functions both individually and collectively to enhance social cohesion (Yan, 2020).
2.2.2. Hukou Type and Social Trust as Well as Social Distrust
Trust is a core dimension of social capital, and its distribution is closely tied to resource access opportunities. Lyu et al. (2022) note that rural hukou holders, confined to homogeneous social networks and lacking cross-class interaction opportunities, exhibit significantly lower generalized trust than urban counterparts. Such disparity is a product of unequal social capital—urban hukou facilitates trust accumulation by expanding social networks and enhancing social participation.
As an institutionalized stratification mechanism, the hukou system reinforces categorical gaps between urban and rural groups through legalized rights differences. This leads to urban hukou holders displaying lower trust towards rural groups and vice versa, with this inter-group trust breakdown stemming from institutionally shaped class divergence (Wu & Wallace, 2021).
The huge gap in urban-rural resource allocation generates strong relative deprivation among rural hukou groups, which further translates into social distrust. Fu et al. (2018) found a positive correlation between depression levels and distrust among rural residents and migrant workers. Even with income growth, non-urban hukou status reinforces the psychological perception of exclusion, undermining the foundation of trust.
Hukou status directly influences individuals’ trust in institutions: urban hukou holders, benefiting more from policy dividends (e.g. education, employment, social security), exhibit higher trust in government and legal systems; while rural hukou groups, experiencing institutional unfairness, are more skeptic towards public institutions (Shi et al., 2024). This disparity in individual level further diffuses into overall social trust differentiation.
Based on the above findings, hukou type influences the formation of social trust and distrust through multiple mechanisms. Thus, we propose:
H2A: After controlling other major objective variables (e.g., gender, age), current hukou type significantly affects social trust variables, with urban hukou holders exhibiting significantly higher social trust levels than rural hukou holders.
H2B: After controlling for other major objective variables (e.g., gender, age), current hukou type significantly affects social distrust variables, with rural hukou holders exhibiting significantly higher social distrust levels than urban hukou holders.
2.2.3. Household Registration Types and Perceptions of Social Fairness
The sense of social fairness, as individuals’ subjective judgment on the rationality of social resource distribution, is profoundly influenced by the hukou system. As mentioned above, the hukou system is a characteristic mechanism of instructional social stratification through differential resource allocation
Hukou differences trigger feelings of unfairness: rural hukou holders, in daily comparisons with urban counterparts, are more likely to perceive opportunity deprivation (e.g., restrictions on education and career advancement) and welfare gaps (e.g., inadequate social security). Lyu et al. (2022), leading to negative evaluations of social fairness.
Hukou-based identity differences strengthen group-based perceptions of inequality. When rural hukou holders recognize that their disadvantages stem from institutional identity rather than personal ability, they are more likely to negate social fairness (Wu & Wallace, 2021). Migrant workers denied equal housing benefits due to hukou restrictions tend to attribute this disparity to institutional injustice rather than insufficient effort (Shi et al., 2024).
Rural hukou holders are more likely to categorize themselves as “lower class”, and this self-positioning amplifies negative fairness perceptions—considering their class status as a result of exclusion rather than fair competition (Fu et al., 2018). Conversely, urban hukou holders regard themselves as middle or upper classes, attributing resource access to social fairness, thereby strengthening positive fairness perceptions.
Based on the above theories and anallysis, hukou type shapes perceived social fairness through structuring resource distribution, triggering relative deprivation, reinforcing group identity differences, and influencing subjective class identification. Thus, we propose:
H2C: After controlling for other major objective variables (e.g., gender, age), current hukou type significantly affects perceived social fairness, with urban hukou holders reporting significantly higher levels of perceived social fairness than rural hukou holders.
2.2.4. Household Registration Types and Perceptions of Social Well-Being
Social well-being, as an overall evaluation of life quality by individuals, is deeply influenced by the resource allocation differences and psychological experiences brought about by the hukou system.
The hukou system directly leads to differences in well-being through institutionalized resource inequality. Urban hukou holders can gain legalized priority access to high-quality education, healthcare, and stable employment resources. These objective conditions lay a material foundation for their well-being. In sharp contrast, rural hukou holders, due to resource scarcity such as low accessibility to medical care and high income volatility, have significantly lower levels of well-being (Wu & Treiman, 2007). For example, the pension level of urban residents is usually 3 to 5 times that of rural residents. This huge gap in social security directly enhances the sense of life security of urban residents and thus strengthens their well-being (Cai, 2011).
The “well-being gap” resulting from urban-rural comparisons greatly undermines the well-being of the rural hukou group. When rural residents or migrant workers see that urban residents’ living conditions far exceed their own, and realize that this gap is caused by hukou rather than personal efforts, they will have a strong sense of relative deprivation, thereby reducing their own well-being (Fu et al., 2018). Research shows that the well-being of migrant workers in the “urban-rural fringe” area is lower than that of rural residents who live stably in rural areas for a long time. This is because frequent cross-group comparisons further magnify their perception of this gap (Shi et al., 2024).
Hukou differences affect well-being by weakening individuals’ “capability set for well-being”. Rural hukou restricts the opportunities for individuals to develop key capabilities. The lack of educational resources limits the improvement of their “knowledge capabilities”, employment discrimination hinders the enhancement of “economic capabilities”, and the absence of social security restricts the development of “risk resistance capabilities” (Sen, 1999). The lack of these capabilities makes it more difficult for rural hukou groups to achieve “valuable life goals”, thus reducing their well-being (Xiao & Bian, 2018).
Differences in social networks also lead to inequality in well-being. The social networks of urban hukou holders are more diverse, covering members from multiple fields such as government and enterprises, enabling them to obtain more emotional support and resource access channels. In contrast, the social networks of rural hukou groups are mostly limited to homogeneous groups, and their support functions are relatively limited (Lyu et al., 2022). This gap in social capital makes it easier for urban residents to get help when facing life pressures, thus maintaining a higher level of well-being.
The institutional exclusion brought about by hukou identity consumes individuals’ psychological resources and thus reduces well-being. Rural hukou holders are often discriminated against and excluded because of their “outsider” identity. These negative experiences can trigger adverse emotions such as anxiety and helplessness, and eventually reduce their well-being (Wu, 2019). For example, migrant workers are unable to send their children to urban public schools due to hukou restrictions. Their children have no choice but to study in those schools specifically established for them, usually equipped with relatively low-quality teachers and inadequate teaching facilities. Such reluctance will make them feel a strong sense of “failure”, thus reducing their overall life satisfaction (Hung, 2022).
Based on the above analysis, hukou type affects social well-being through multiple paths such as resource allocation, relative deprivation, capability restriction, social capital, and psychological resources. After controlling for other major objective variables such as gender and age, this impact might still remain significant. Therefore, we propose the hypothesis:
H2D: After controlling for other major objective variables (e.g. gender, age), the current hukou type has a significant impact on social well-being. The social well-being of urban hukou holders is significantly higher than that of rural hukou holders.
3. Data and Methods
3.1. Data Sources
The Chinese General Social Survey (CGSS), launched in 2003, is China’s first nationwide, comprehensive, and continuous large-scale social survey project. The National Survey Research Center (NSRC) at Renmin University of China is responsible for the long-term implementation, management, and data release of this project. The data from the 2021 survey of the CGSS (CGSS 2021) was officially released to the public on March 31, 2023.
For the 2021 survey, a total of 8148 valid samples were collected across China. The released CGSS 2021 dataset contains 700 variables, and all respondents were asked questions covering both the core modules and thematic modules of the survey.
3.2. Variable Definition and Measurement
3.2.1. Independent Variables
Current Hukou Type
In the CGSS 2021, hukou type is measured by the questionnaire item A18 (“What is your current household registration status?”). This item includes seven categories of household registration status, specifically: Agricultural hukou (Coded 1), Non-agricultural hukou (Coded 2), Residential hukou (previously agricultural hukou) (Coded 3), Residential hukou (previously non-agricultural hukou) (Coded 4), Military hukou (Coded 5), No hukou (Coded 6), Other types (Coded 7)
In this study, considering that the focus of analysis is on the differences between urban and rural hukou identities and their social effects, the hukou type was recoded into a dichotomous variable: 0 = Agricultural hukou, 1 = Non-agricultural/residential hukou. Specifically, “Agricultural hukou” refers to groups that still retain the traditional rural hukou identity, while “Non-agricultural/residential hukou” includes non-agricultural hukou, residential hukou (previously agricultural hukou), and residential hukou (previously non-agricultural hukou). Groups in this category typically have higher urbanization characteristics and better access to public resources. Military hukou, no hukou, and other types were not included in the main analysis and were coded as missing values, due to their limited sample sizes and high heterogeneity.
Among the 8,148 valid samples, 8,055 respondents (accounting for 98.9% of the total samples) had valid responses after recoding. Specifically: 4,842 respondents (60.1% of the valid samples) held agricultural hukou; 3,213 respondents (39.9% of the valid samples) held non-agricultural/residential hukou. In addition, 93 questionnaires (1.1% of the total samples) had missing values for this item:
3.2.2. Dependent Variables
1) Logarithm of Annual Personal Income
In CGSS 2021, personal income is measured by questionnaire item A8a: “What was your total personal income for the entire last year [2020]?” According to the statistics of valid samples, among the 8,148 respondents, 7,333 provided valid data. The 2020 annual personal income of these respondents shows a highly discrete distribution. The minimum value is .00, indicating that some respondents had no income or reported zero income in the survey year; the maximum value is as high as 9,999,990.00, reflecting a small number of individuals with extremely high incomes in the sample. The average income is 51,241.10 yuan. Though this figure is relatively high, it does not accurately reflect the actual income level of most respondents due to the presence of extreme high values. The standard deviation reaches 332,375.66, much higher than the mean, which further highlights the significant income gap between samples and the strong right-skewness of the data distribution.
To address this, this study applies a logarithmic transformation to the variable (referred to as the “logarithm of annual personal income”) to improve its distribution characteristics. Specifically, the following formula is used:
Income_log = log10(Income + 1)
The “+1” is added to avoid calculation errors caused by zero values. Such treatment not only reduces the impact of extreme values but also helps meet the normality assumption of multiple regression and other parametric statistical models. After applying the logarithmic transformation to the 7,333 valid samples, the distribution of the “logarithm of annual personal income” variable (Log10 income (new)) is significantly improved compared to the original income data. Its value range is from .00 to 7.00, with a standard deviation of 1.83604—markedly lower than the standard deviation of 332,375.66 in the original data, indicating that the distribution is effectively controlled.
2) Social Trust
In this study, four social attitude variables from CGSS 2021 describe individuals’ social attitudes and life evaluations from four dimensions—social trust (A33), social distrust (A34), sense of social fairness (A35), and subjective well-being (A36). These variables have a close logical relationship: Social trust and social distrust represent two opposing dimensions of social capital: the former emphasizes positive interdependence between people, while the latter reflects caution and suspicion in social interactions, and they are often negatively correlated. Sense of social fairness is a key antecedent of social trust: if people generally perceive social systems and resource distribution as fair, they are more likely to trust others; otherwise, distrust may arise. Subjective well-being is an outcome variable, typically positively influenced by social trust and a sense of fairness, and weakened by the negative expectations of social distrust. In summary, these four variables jointly form a research framework of “social attitudes—well-being”, where the sense of social fairness affects the formation of social trust and distrust, and trust/distrust ultimately exerts a significant impact on individuals’ subjective well-being.
Social trust (the overall perception of social trust) is measured by questionnaire “Generally speaking, do you agree or disagree that most people in this society are trustworthy?” Respondents answer on a 5-point Likert scale.
The distribution of the social trust variable shows an obvious positive tendency. The total sample size is 8,148, among which 8,079 are valid responses (accounting for 99.2%), and missing values only account for .8%, indicating high overall data integrity.
Among the valid responses, the largest group is those who chose “relatively agree”, with 4,405 people (54.5%), followed by “strongly agree”, with 1,208 people (15.0%). This shows that most respondents hold a positive attitude toward “most people in society being trustworthy”, with approximately 70% (69.5%) demonstrating different degrees of trust.
In contrast, the proportion of respondents with negative attitudes is relatively low: “relatively disagree” accounts for 13.4%, and “strongly disagree” only accounts for 3.3%, totaling 16.7%. In addition, 13.8% of respondents chose “neither agree nor disagree”, indicating that some respondents hold a neutral or uncertain attitude toward social trust.
3) Social Distrust
In this study, social distrust is measured by questionnaire “Generally speaking, do you agree or disagree that in this society, if you are not careful, others will try to take advantage of you?” Respondents answer on a 5-point Likert scale. In contrast to A33 (social trust), A34 is often regarded as a reverse indicator of social distrust or social defensive psychology, which can reveal the public’s negative expectations of potential risks and reciprocal relationships in social interactions. When combined with social trust, this variable can more comprehensively reflect the trust and distrust dimensions in the structure of social capital.
Among the 8,057 valid samples: 774 people (9.6%) chose “strongly disagree”, while 2,698 people (33.5%) chose “relatively disagree”; The two groups together account for 43.1%, indicating that nearly half of the respondents hold a negative attitude toward the idea that “others will take advantage of them.”
On the other hand: 1,526 people (18.9%) held a neutral attitude (“neither agree nor disagree”), showing that about one-fifth of respondents had no clear tendency on this issue; The proportion of respondents who agreed was also relatively high: 2,546 people (31.6%) chose “relatively agree”, and 513 people (6.4%) chose “strongly agree”, totaling approximately 38.0%.
4) Perception of Social Fairness
The perception of social fairness is measured by item A35: “Generally speaking, do you think today’s society is fair or not?” This question aims to capture individuals’ subjective perception and attitude toward the overall fairness of society. A 5-point Likert Scale is used for scoring, and codes 98 and 99 are generally treated as missing values.
In terms of variable meaning, A35 mainly reflects individuals’ evaluation of the fairness of social structure and institutional operation. A higher score indicates that the individual believes the society is more fair, while a lower score indicates that the individual perceives significant unfairness. This variable is not only a key antecedent of social trust (as a sense of fairness is one of the psychological foundations for the formation of trust) but also often used as a core indicator to measure social cohesion and institutional legitimacy.
There are a total of 8,148 samples, among which 8,090 are valid responses (accounting for 99.3%) and 58 are missing values (accounting for .7%). From the distribution: 346 people (4.3%) think it is “completely unfair” (a relatively low proportion); 1,131 people (14.0%) choose “relatively unfair”; 1,735 people (21.4%) express “neither fair nor unfair” (a neutral attitude); The largest group is those who think it is “relatively fair”, with 4,233 people (52.3%), exceeding half of the valid samples; 645 people (8.0%) hold the attitude of “completely fair”.
The cumulative percentage shows that approximately 18.3% of people believe society is unfair (including “completely unfair” and “relatively unfair”), while 60.3% believe society is fair (including “relatively fair” and “completely fair”). The remaining 21.4% of respondents hold a neutral attitude. Overall, the data indicates that most respondents hold a relatively positive evaluation of social fairness, with “relatively fair” being the dominant view; however, a certain proportion of the population (about one-fifth) is skeptical or holds a negative view of social fairness. This distribution shows that the perception of social fairness is differentiated among the group, but the overall tendency is positive.
5) Subjective Well-being
Subjective Well-being is measured through the question “Do you feel your life is happy?” Overall, a total of 5,448 respondents provided valid answers, accounting for 66.9% of the total sample, while 2,700 respondents (33.1%) had missing data, indicating a certain proportion of invalid samples for this variable. Among the valid samples, most respondents held a positive attitude toward their lives. Specifically, the largest group was those who chose “relatively happy”, with 3,113 respondents (57.1%); in addition, 1,319 respondents (24.2%) selected “very happy”. This shows that over three-quarters of the respondents (81.3%) perceived their lives as generally happy, demonstrating a clear positive tendency. In contrast, only 75 respondents (1.4%) reported being “very unhappy”, and 243 respondents (4.5%) chose “relatively unhappy”, together accounting for just 5.8% of the valid samples. This indicates that the proportion of individuals with extremely negative well-being in the group is very low. At the same time, 698 respondents (12.8%) selected “neither happy nor unhappy”, reflecting that some respondents hold ambiguous or neutral attitudes toward their well-being.
Overall, the results show a distinct positive trend: most individuals perceive themselves as happy or relatively happy, and very few feel unhappy. This suggests that the level of subjective well-being among respondents in the overall sample is relatively high. However, attention should also be paid to the relatively high data missing to avoid any impact on the overall representativeness of the results.
3.2.3. Other Variables
1) Gender
In the sample of this study, the gender variable (A2) was recorded by interviewers on-site, covering a total of 8,148 valid respondents. Among them, 4,469 were female, accounting for 54.8% of the total sample, and 3,679 were male, accounting for 45.2%. The gender distribution shows that females are slightly more than males, but the overall gender composition is relatively balanced.
2) Age
In this study, the age variable was calculated based on the respondents’ year of birth (A3), specifically by subtracting the year of birth from the survey year (2021) to obtain the actual age. The results show that a total of 8,148 respondents provided valid age information, with ages ranging from 18 to 99 years old and an average age of 51.64 years old (standard deviation = 17.57). The sample covers groups from young adults to the elderly, reflecting broad representativeness in terms of age structure.
3) Educational Level
In this study, the educational level variable (A7a) was recoded according to the respondents’ “highest level of education attained” to facilitate subsequent statistical analysis. In the original questionnaire, this item included 14 options, ranging from “no education received” to “postgraduate or above”. To simplify the analysis and reflect the progressive nature of educational levels, this study recoded it into five grades: Code 1: “No education received” and “private school/illiteracy elimination class”; Code 2: “Primary school”; Code 3: “Junior high school”; Code 4: “Vocational high school”, “regular senior high school”, “technical secondary school”, and “technical school”; Code 5: “Junior college (adult or formal higher education)”, “undergraduate (adult or formal higher education)”, and “postgraduate or above”. Other codes were treated as missing values.
The distribution of the recoded variable shows that among 8,127 valid samples: 905 respondents (11.1% of the total sample) had an educational level coded as 1; 1,751 respondents (21.5%) as code 2; 2,311 respondents (28.4%) as code 3; 1,489 respondents (18.3%) as code 4; 1,671 respondents (20.6%) as code 5. In addition, 21 questionnaires (.3% of the total sample) had missing data for this item. The distribution indicates that the educational level of respondents in the sample is generally characterized by a medium-to-high structure: the proportion of those with junior high school and primary school education is relatively large, and the proportion of those with higher education background (code 5) reaches one-fifth.
4) Marital Status (Spousal Status)
Variable A69 (marital status) is designed to measure the respondents’ current marital and family relationship status, and it is one of the commonly used basic demographic variables in sociological and demographic research. The item lists different marital status options, allowing respondents to select the one that best matches their situation. In this study, marital status (A69) was initially measured with seven options, specifically: unmarried a); cohabiting b); first marriage with spouse c); remarriage with spouse d); separated but not divorced e); divorced f); widowed g). In statistical analysis, this variable was treated as a nominal variable; however, to enhance model simplicity and statistical efficiency, this study dichotomized it into “with spouse” and “without spouse”. Among them, cohabiting, first marriage with spouse, and remarriage with spouse were uniformly recoded as 1 (with spouse), while the other four categories (unmarried, separated, divorced, widowed) were uniformly coded as 0 (without spouse). The logic behind this classification is that the presence of a spouse or partner often implies stronger emotional support, economic security, and social recognition, whereas the lack of a stable partner relationship may lead to weakened social connections and insufficient resources.
According to the descriptive statistical results, among the total sample of 8,148 respondents, 5,954 (73.1%) had a spouse, and 2,194 (26.9%) did not. It can be seen that most respondents were in a partner relationship at the time of the survey, reflecting the prevalence of marital or partner relationships in Chinese society. In subsequent analyses, this variable will be used as a key indicator of family and social support to explore the role of spousal relationships in social-psychological outcomes such as well-being, social trust, and sense of social fairness.
5) Political Identity
Variable A10 (political identity) is used to measure the respondents’ political identity/political affiliation at the level of social organizations or political parties. It reflects an individual’s formal or informal affiliation status within China’s political system and is usually used to analyze social structure, political participation, social trust, and differences in values.
In this study, political identity (Political status) was treated as a dichotomous variable. Based on the original questionnaire options, the “mass public” a), “Communist Youth League members” b), and “members of democratic parties” c) were uniformly coded as 0 (non-CCP members, NON-CCP), while “Communist Party of China (CCP) members” d) were coded as 1 (CCP members, CCP). Among the total sample of 8,148 respondents, there were 8,135 valid samples. Among them, non-CCP members accounted for the majority, with 7,169 respondents (88.0% of the total sample; 88.1% of valid samples), and CCP members totaled 966 respondents (11.9%). In addition, 13 samples (.2%) were excluded from the analysis due to systematic missing data. This variable clearly reflects the political identity structure of the respondent group—non-party members are dominant, and the proportion of party members is relatively small. This dichotomous classification facilitates comparing differences in social attitude variables (such as sense of social fairness, well-being, and social trust) between groups with different political identities in subsequent analyses, thereby revealing the potential impact of political identity on the formation of social cognition and values.
6) Square of Houses
As a continuous variable, square of houses (unit: square meters) serves as an objective physical indicator reflecting the living conditions of respondents. The value of this variable is the figure reported by respondents, ranging typically from smaller single-person housing to larger multi-bedroom housing, with the specific number representing the size of the housing space.
In this study, the variable “Square of Houses” reflects the internal gross floor area of the respondents’ current housing. The original values range from 7 to 9,997 square meters, with a mean of 121.73 square meters and a large standard deviation (175.79), indicating a significant positive skewness and the presence of extreme values in the data distribution. To improve the distribution shape and reduce the impact of extreme values on statistical analysis, this study applies a natural logarithm (LN) transformation to the Square of House.
For the transformed variable “LN SQUARE OF HOUSE”, among 7,235 valid samples: The minimum value is 1.95 (corresponding to an original area of approximately 7 square meters); the maximum value is 9.21 (corresponding to an original area of approximately 9,977 square meters); the mean is 4.60 (corresponding to an original area of approximately 99 square meters); the standard deviation is .60.
After the logarithmic transformation, the distribution of Square of House is much closer to normal, and its volatility is significantly reduced—thereby improving its applicability and robustness in regression analysis and other statistical models.
4. Results
4.1. Verification of H1
This study employs an Independent Samples t-test to compare differences in the logarithm of annual personal income (Log10 income) between groups with different hukou types (agricultural hukou vs. non-agricultural hukou). The dependent variable “annual personal income” was logarithmically transformed to reduce the skewness of its original distribution and better meet the normality assumption.
First, from the results of descriptive statistics (Group Statistics), the mean value of the logarithmic annual income for the non-agricultural hukou group (N = 2,963) is 4.11 (SD = 1.53), which is significantly higher than that of the agricultural hukou group (N = 4,290) at 3.08 (SD = 1.91). This indicates a significant gap in the average income level between the two groups.
Further results of the Independent Samples t-test show that Levene’s Test for Equality of Variances is significant (F = 587.94, p < .001), suggesting that the variances of the two groups are unequal. Therefore, it is more reliable to adopt the t-test results under the assumption of unequal variances. According to Table 1, the test results reveal that the difference in the mean logarithmic income between the two groups is highly significant (t = 25.37, df ≈ 7,091.50, p < .001). Specifically, the income of the non-agricultural hukou group is significantly higher than that of the agricultural hukou group, with a mean difference of 1.03 and a corresponding 95% confidence interval of [.95, 1.11].
Effect size analysis further supports this result. Cohen’s d is .582 (95% CI: .535 - .630), indicating a moderately large effect size for the difference; the result of Hedges’ g is consistent (g = .582), while Glass’s delta is .538. Overall, this demonstrates that hukou type has substantial explanatory power for differences in personal income.
In conclusion, the results of this study indicate that against the backdrop of Chinese society, the personal income level of the non-agricultural hukou group is significantly higher than that of the agricultural hukou group. This difference is not only statistically significant but also of moderate practical importance, reflecting the crucial impact of the hukou system on the distribution of economic resources and social inequality.
Table 1. Independent samples t-test to compare differences in the logarithm of annual per-sonal income.
Hukou Type |
t Test for Log10 Income by Hukou Type |
N |
M |
SD |
Non-agriculture Hukou |
2963 |
4.11 |
1.53 |
Agriculture Hukou |
4290 |
3.08 |
1.91 |
t (df) |
25.37 (7091.50) |
|
|
p |
<.001 |
|
|
Mean Diff. |
1.03 |
|
|
95% CI [LL, UL] |
[.95, 1.11] |
|
|
Cohen’s d |
.58 |
|
|
Note: M = mean; SD = standard deviation; CI = confidence interval; LL = lower limit; UL = upper limit. Equal variances not assumed.
4.2. Verification of H2A
The dependent variable in this hypothesis is social trust (A33), which was originally measured on a 5-point Likert scale (ranging from “strongly disagree” to “strongly agree”). In regression modeling, most academic literature treats such Likert-scale variables as continuous dependent variables, thus adopting multiple linear regression for analysis. In this study, social trust (A33), measured on a 5-point Likert scale, serves as the dependent variable. A multiple linear regression model is constructed with hukou type as the core independent variable, while controlling for variables including gender, age, educational level, marital status (spousal status), political identity, logarithm of Square of House, and logarithm of income.
The overall fit of the regression model shows that R2 = .025 and adjusted R2 = .024, indicating that the model can explain approximately 2.5% of the variance in social trust. Additionally, the overall regression is statistically significant (F (8, 6484) = 20.895, p < .001).
In terms of specific variables, according to Table 2, gender (B = 0.057, p = .023), age (B = .008, p < .001), educational level (B = .045, p < .001), political identity (B = .144, p < .001), and logarithm of Square of House (B = .086, p < .001) exert a significant positive impact on social trust. This indicates that males, older individuals, those with higher educational attainment, CCP members, and people with better housing conditions are more inclined to agree with the view that “most people in society are trustworthy.”
In contrast, marital status (spousal status) (B = −.026, p = .373), logarithm of income (B = .010, p = .189), and hukou type (B = −.038, p = .182) are not statistically significant. This suggests that marriage, income level, and urban-rural hukou differences do not have a significant effect on social trust.
Overall, the formation of social trust is more closely related to individuals’ demographic characteristics (e.g., gender, age, education) and social identity (e.g., political identity). In contrast, factors traditionally assumed to influence trust—such as economic resources (income) and urban-rural hukou differences—receive no empirical support in this study.
Table 2. Multiple regression Predicting social trust.
Predictor |
B |
SE |
β |
t |
p |
Gender |
.057 |
.025 |
.029 |
2.27 |
.023 |
Age |
.008 |
.001 |
.147 |
9.40 |
<.001 |
Education |
.045 |
.013 |
.058 |
3.36 |
<.001 |
Spouse status |
−.026 |
.029 |
−.011 |
−.89 |
.373 |
Political status |
.144 |
.039 |
.050 |
3.74 |
<.001 |
LN Housing size |
.086 |
.020 |
.053 |
4.23 |
<.001 |
Log10 income |
.010 |
.007 |
.017 |
1.32 |
.189 |
Hukou (0 = rural, 1 = urban) |
−.038 |
.028 |
−.019 |
−1.33 |
.182 |
4.3. Verification of H2B
To test H2B, this study constructs a multiple linear regression model with social distrust (A34) as the dependent variable, hukou type as the core independent variable, and controls for variables including gender, age, educational level, marital status (spousal status), political identity, logarithm of Square of House, and logarithm of income.
The model results show a relatively low overall fit (R2 = .008, adjusted R2 = .007), but the overall regression equation is statistically significant (F (8, 6468) = 6.379, p < .001). This indicates that although the proportion of variance explained by the model is limited (approximately .8%), the independent and control variables have statistical significance for explaining differences in social distrust.
In terms of specific variables, gender and political identity exhibit significant impacts. According to Table 3, the regression coefficient for gender is positive (B = .145, p < .001), which means that male respondents have significantly higher levels of social distrust than female respondents. In contrast, the regression coefficient for political identity is negative (B = −.224, p < .001), indicating that members of the Communist Party of China (CCP) have significantly lower social distrust than non-CCP members.
Other control variables—including age (B = −.001, p = .373), educational level (B = .003, p = .849), marital status (spousal status) (B = .017, p = .601), logarithm of Square of House (B = −.023, p = .331), logarithm of income (B = .007, p = .431), and hukou type (B = .007, p = .828)—all fail to reach statistical significance. This suggests that these socioeconomic variables play a limited role in explaining social distrust.
In summary, social distrust is mainly influenced by gender and political identity, while traditional socioeconomic resources (income, education, housing) and urban-rural hukou differences do not show significant effects. This indicates that social distrust more strongly reflects differences in social-psychological characteristics and political identity, rather than mere economic or institutional differentiation.
Table 3. Multiple regression Predicting social distrust.
Predictor |
B |
SE |
β |
t |
p |
Gender |
.145 |
.029 |
.064 |
5.01 |
<.001 |
Age |
−.001 |
.001 |
−.014 |
−.89 |
.373 |
Education |
.003 |
.016 |
.003 |
.19 |
.849 |
Spouse status |
.017 |
.033 |
.007 |
.52 |
.601 |
Political status |
−.224 |
.045 |
−.068 |
−5.02 |
<.001 |
LN Housing size |
−.023 |
.024 |
−.012 |
−.97 |
.331 |
Log10 income |
.007 |
.009 |
.011 |
.79 |
.431 |
Hukou (0 = rural, 1 = urban) |
.007 |
.033 |
.003 |
.22 |
.828 |
4.4. Verification of H2C
To test H2C, this study constructs a multiple linear regression model with sense of social fairness (A35) as the dependent variable, urban-rural hukou type as the core independent variable, and controls for variables including gender, age, educational level, marital status (spousal status), political identity, logarithm of Square of House, and logarithm of income.
The model results show a limited overall fit (R2 = .015, adjusted R2 = .014), but the regression equation is statistically significant overall (F (8, 6482) = 12.129, p < .001). This means that although the explanatory power of the independent and control variables for the sense of social fairness is not high, the model is statistically valid as a whole.
According to Table 4, Several factors exert a significant impact on the sense of social fairness:
Gender (B = .067, p = .006): Male respondents have a significantly higher evaluation of social fairness than female respondents.
Age (B = .005, p < .001): The older the respondents, the stronger their sense of social fairness, indicating certain generational differences.
Political identity (B = .127, p < .001): Members of the Communist Party of China (CCP) have a significantly higher sense of social fairness than non-CCP members.
Logarithm of Square of House (B = .047, p = .018): People with better housing conditions have a higher perception of social fairness.
Logarithm of income (B = .021, p = .003): Groups with higher income are significantly more inclined to perceive society as fair.
In contrast, educational level (p = .371), marital status (p = .197), and hukou type (p = .313) do not show significant effects. This indicates that urban-rural differences and educational attainment play an insignificant role in explaining the sense of social fairness.
In summary, differences in the sense of social fairness are more strongly influenced by the combined effects of gender, age, political identity, and socioeconomic resources (income, housing conditions), rather than mere urban-rural hukou differences. This implies that in the context of contemporary China, political identity and resource distribution play a more prominent role in shaping the sense of social fairness.
Table 4. Multiple regression predicting perceived social fairness.
Predictor |
B |
SE |
β |
t |
p |
Gender |
.067 |
.025 |
.035 |
2.75 |
.006 |
Age |
.005 |
.001 |
.091 |
5.82 |
<.001 |
Education |
.012 |
.013 |
.015 |
.89 |
.371 |
Spouse status |
−.037 |
.028 |
−.016 |
−1.29 |
.197 |
Political status |
.127 |
.038 |
.045 |
3.34 |
<.001 |
LN Housing size |
.047 |
.020 |
.030 |
2.37 |
.018 |
Log10 income |
.021 |
.007 |
.039 |
2.93 |
.003 |
Hukou (0 = rural, 1 = urban) |
−.028 |
.028 |
−.015 |
−1.01 |
.313 |
4.5. Verification of H2D
Variable A36, which measures respondents’ subjective evaluation of their life happiness, is a typical indicator of Subjective Well-being (SWB). It is designed using a 5-point Likert scale.
This study further employs multiple linear regression to examine the impact of social structural variables on subjective well-being (A36) as the dependent variable. The overall model fit is limited (R2 = .027, adjusted R2 = .025), but the regression equation is statistically significant (F (8, 4348) = 15.066, p < .001), indicating that the selected variables can collectively predict individuals’ subjective well-being in a statistical sense.
According to Table 5, several factors show significant positive impacts on subjective well-being:
Age (B = .004, p < .001): Older individuals report higher subjective well-being, suggesting that life satisfaction tends to increase with age.
Educational level (B = .063, p < .001): Respondents with higher educational attainment exhibit significantly greater subjective well-being than those with lower education.
Marital status (spousal status) (B = .113, p < .001): Married respondents or those with a spouse report higher levels of happiness.
Political identity (B = .151, p < .001): Members of the Communist Party of China (CCP) show significantly higher well-being than non-CCP members.
Logarithm of Square of House (B = .043, p = .033): Improved housing conditions are positively correlated with increased well-being.
Logarithm of income (B = .014, p = .049): Higher income is associated with greater well-being, though the effect is relatively weak.
Meanwhile, gender (p = .086) and urban-rural hukou type (p = .246) are not statistically significant. This indicates that after controlling for other socioeconomic variables, there are no significant differences in well-being between males and females, or between urban and rural residents.
Overall, the results suggest that subjective well-being is shaped more by the combined effects of factors such as education, income, marital status, political identity, and socioeconomic resources, rather than being determined solely by gender or urban-rural differences. This aligns with social stratification theory, which emphasizes the critical impact of “resources and status” on well-being.
Table 5. Multiple regression predicting subjective well-being.
Predictor |
B |
SE |
β |
t |
p |
Gender |
−.043 |
.025 |
−.027 |
−1.72 |
.086 |
Age |
.004 |
.001 |
.090 |
4.65 |
<.001 |
Education |
.063 |
.014 |
.100 |
4.63 |
<.001 |
Spouse status |
.113 |
.029 |
.061 |
3.95 |
<.001 |
Political status |
.151 |
.038 |
.065 |
4.00 |
<.001 |
LN Housing size |
.043 |
.020 |
.033 |
2.13 |
.033 |
Log10 income |
.014 |
.007 |
.032 |
1.97 |
.049 |
Hukou (0 = rural, 1 = urban) |
.033 |
.029 |
.021 |
1.16 |
.246 |
5. Discussions and Conclusions
5.1. Verification Results and Theoretical Implications of the Four
Social Attitude Models
As shown in Table 6, in the overall comparison, all four regression models control for the same social-structural variables, including gender, age, educational attainment, spousal status, political identity, Square of House and income. However, the hukou type fails to reach a significant level in all four models, which indicates that the urban-rural gap has limited explanatory power for social trust, distrust, sense of fairness, and well-being. Meanwhile, the overall fit is generally low (with R2 ranging from .8% to 2.7%), suggesting that these indicators of social attitudes are influenced by more complex factors rather than relying solely on social-structural variables.
Specifically, the results for social trust (H2A) show that it is more dependent on demographic and identity characteristics, such as gender, age, education, and political identity, while in the case of social distrust (H2B), only gender and political identity reach a significant level, and the direction is opposite to that of social trust, indicating that political identity is a key dividing line between trust and distrust. Moreover, the analysis of the sense of social fairness (H2C) suggests that economic resources (e.g. income and housing) start to play a role, meaning that individuals’ perception of fairness is more influenced by their socio-economic status. In contrast, subjective well-being (H2D) is jointly affected by multiple factors such as education, income, marriage, and political identity, and it is the result that most comprehensively reflects the role of socio-economic resources among the four dependent variables.
Table 6. Comparative results of hypothesis verification for the four social attitude models.
Hypothesis No. |
Dependent Variable (DV) |
Core Independent Variable |
Control Variables |
Significant Predictors |
R2 (Explanatory Powe) |
Hukou Effect |
H2A |
Social Trust (A33) |
Hukou Type |
Gender, Age, Education, Spouse, Political Identity, Square of House, Income |
Significant: Gender, Age, Education, Political Identity, Square of House |
H2A |
Social Trust (A33) |
H2B |
Sense of Social Distrust (A34) |
Hukou Type |
Same as above |
Significant: Gender (+), Political Identity (−) |
H2B |
Sense of Social Distrust (A34) |
H2C |
Sense of Social Fairness (A35) |
Hukou Type |
Same as above |
Significant: Gender, Age, Political Identity, Square of House, Income |
H2C |
Sense of Social Fairness (A35) |
H2D |
Sense of subjective well - being (A36) |
Hukou Type |
Same as above |
Significant: Age, Education, Spouse, Political Identity, Square of House, Income |
H2D |
Sense of subjective well - being (A36) |
From a theoretical perspective, the non-significant result of the hukou type reflects that, at the level of subjective evaluations such as social trust, fairness, and well-being, the importance of the urban-rural difference is diminishing; and political identity emerges as a significant factor (either positive or negative) in all four models, demonstrating that institutional identity plays a prominent role in shaping social attitudes. To sum up, the trust-related indicators (H2A/H2B) are more likely to be explained by social psychology and political identity, while the fairness and well-being indicators (H2C/H2D) are more influenced by socio-economic resources.
The results of this study reveal a striking divergence—what may be termed the “significant economic effect but insignificant psychological effect” of hukou. On the one hand, hukou continues to structure material inequalities by shaping income distribution and access to economic resources, thereby reinforcing its role as a key mechanism of structural stratification. On the other hand, its influence on subjective social attitudes—trust, fairness, and well-being—appears muted once factors such as political identity, education, and income are controlled. This divergence underscores a transitional pattern in stratification logic: while hukou still operates as an institutional barrier in the economic sphere, its symbolic power in shaping everyday perceptions and values is waning. This suggests that in transitional societies such as China, stratification may increasingly shift from identity-based exclusion to resource- and institution-based differentiation.
Admittedly, the persistently low R2 values also underscore that a constellation of unmeasured variables may be exerting a decisive influence on the formation of social attitudes. Extending beyond structural and demographic determinants, local community contexts—shaped by the presence of neighborhood networks, sustained by grassroots organizations, and governed through community-level institutions—could profoundly mold individuals’ perceptions of fairness and trust. Likewise, media exposure—ranging from entrenched traditional outlets to rapidly proliferating digital platforms—may be amplifying narratives of inequality or cohesion, thereby recalibrating collective attitudes. Moreover, personal life trajectories, marked by experiences of discrimination, disrupted families, or upward mobility, are likely exerting powerful yet unobserved effects on subjective evaluations of trust, fairness, and happiness. These considerations compel future research to integrate psychological, cultural, and contextual dimensions, thereby more comprehensively illuminating the determinants of social attitudes in contemporary China.
5.2. Comprehensive Analysis of the Impact of Household
Registration
The test results of H1 clearly reveal the significant impact of the household registration system at the economic level. Through the independent-samples t-test, it is found that the personal annual income of the non-agricultural household registration group is significantly higher than that of the agricultural household registration group. The difference is not only highly statistically significant (t = 25.37, p < .001), but also the effect size is moderately large (Cohen’s d = .582). This finding indicates that the type of household registration remains a key social stratification factor in the distribution of economic resources, significantly shaping the income gap between urban and rural residents, and also reflects the structural role of the household registration system in social inequality.
However, in the further analysis of social psychological attitudes, the results of the H2 series (social trust, distrust, sense of fairness, and sense of happiness) show another picture. All four regression models control for the same social structure variables, and the results show that the type of household registration does not reach a significant level in these four models. This indicates that although household registration still has a substantial role in the distribution of economic resources, its explanatory power in individuals’ social psychological evaluations (such as trust, sense of fairness, and sense of happiness) is limited, and the importance of urban-rural differences at these subjective attitude levels is weakening.
This contrast of “significant economic effect-insignificant psychological effect” has important theoretical significance. On one hand, it highlights that the household registration system remains an important structural factor affecting the income gap; on the other hand, it also suggests that at the level of social attitudes and value cognition, urban-rural differences are no longer the main dividing line. In contrast, variables such as political identity, education, and income show stronger explanatory power in the models of social trust, sense of fairness, and sense of happiness, indicating that institutional identity and socio-economic resources are gradually replacing household registration and becoming the key dimensions in shaping social psychological attitudes. This finding further raises an important theoretical question: does the weakening of hukou effects on social attitudes indicate that China’s social stratification is undergoing a fundamental transformation—from an identity-based regime to a resource-based regime?
To sum up, the role of the household registration system shows domain differentiation: at the economic level, it is still a significant stratification mechanism; but at the social psychological level, its explanatory power has relatively weakened. This result indicates that the social stratification in contemporary China is undergoing transformation, and economic resources and political identity can better explain people’s social attitudes and sense of happiness than the traditional urban-rural dual division.
5.3. Comparison of Similarities and Differences between the
Findings of This Study and Previous Studies
This study is consistent with previous studies in terms of conclusions regarding economic effects: urban-rural hukou differences still significantly affect income and employment opportunities. Studies by Wu & Treiman (2007), Wu (2019), and Jia (2024) all illustrate that rural hukou groups are more likely to be in low-income and unstable employment, with restricted social capital; similarly, the T-test results of this study show that the income of non-agricultural hukou groups is significantly higher than that of agricultural hukou groups. Such consistency suggests that the hukou system, as an important institutional factor in the distribution of economic resources, still plays a role today.
In addition, when it comes to the gap in social trust and perceived social fairness, the findings of this study are partially consistent with those of Lyu et al. (2022) and Hung (2022). Existing literature points out that rural hukou groups are more likely to perceive institutional unfairness and exhibit lower levels of social trust. This study also found that urban-rural differences have a potential impact on social attitudes, though the statistical effect was not significant.
Compared with studies by Wu (2019), Hung (2022), etc., this study found different results at the level of social psychological variables. Previous studies emphasized that urban-rural hukou differences significantly affect social trust, fairness perception, and well-being. However, the regression models of this study based on CGSS2021 show that after controlling for variables such as education, income, and political identity, the explanatory power of urban-rural hukou on social attitudes was not significant. In other words, urban-rural segmentation in contemporary China is no longer a core factor in shaping social psychological perceptions.
There are three main reasons for this discrepancy:
1) Temporal differences: Previous studies mostly used data from 2000-2015, while this study uses CGSS2021, reflecting the latest developments in hukou system reform and the promotion of urban-rural integration policies in recent years.
2) Transformation of social stratification logic: Conventional studies emphasized “identity constraints”, but the results of this study indicate that social psychological attitudes are more determined by “resource constraints” and “institutional identities” (such as political identity, educational resources), suggesting that China’s social stratification logic is shifting from “hukou-oriented” to “resource and identity-oriented”.
3) Methodological and variable control: This study controlled for multiple social resource variables such as education, political identity, and housing. Under a more rigorous model, the hukou effect was weakened, highlighting the important role of other socioeconomic resources in explaining social attitudes
5.4. Theoretical and Practical Significance of This Study
Traditional social stratification theory tends to emphasize the urban-rural dual opposition, treating the hukou system as a core structural factor of social inequality. This study finds that the logic of social stratification in contemporary China is undergoing transformation: gradually shifting from “hukou-oriented” to “resource and identity-oriented”. In other words, economic resources (income, housing conditions, educational opportunities) and institutional identities (political identity, social roles) are gradually replacing hukou as more important dimensions in shaping individuals’ social attitudes and psychological perceptions.
Moreover, this study further distinguished between different types of social psychological outcome variables. Specifically, trust-related indicators (such as social trust and social distrust) are more influenced by social psychological and institutional identity factors, among which political identity and group belonging are especially crucial in shaping trust perceptions. In contrast, fairness and well-being indicators mainly depend on socioeconomic resources, including educational level, income level, and housing conditions. This distinction not only reveals the differentiated mechanisms behind different social attitude variables, but also provides a new analytical perspective for understanding how social stratification acts on individual psychological evaluations through multiple paths.
Research results show that hukou still has a significant impact on economic outcomes, especially in terms of uneven income distribution; however, its influence on social psychological attitudes has significantly weakened. This conclusion suggests that policies should further promote the decoupling of hukou identity and economic resource allocation. Specifically, it is reasonable to continue reducing hukou-based discrimination in employment, education, social security, and other fields, thus gradually eliminate resource barriers caused by institutional identity, in order to narrow the urban-rural economic gap and enhancing social fairness and social cohesion.
5.5. Limitations and Future Research
5.5.1. Limitations
First, although the CGSS 2021 cross-sectional data used in this study is nationally representative, its cross-sectional nature limits the inference of causal relationships. We can reveal the correlations between hukou type and income as well as social attitudes, but we cannot fully examine dynamic processes and causal mechanisms. For example, the long-term effects of hukou conversion, migration experiences, or policy reforms cannot be systematically presented.
Second, the explanatory power in the study is limited. The R2 values of the four regression models regarding social attitudes (social trust, distrust, sense of fairness, and sense of well-being) are all low (ranging from .8% to 2.7%), indicating that these psychological attitudes are influenced by more complex cultural, psychological, and situational factors, rather than being solely explainable by social structural variables. This suggests that there are still deficiencies in the model’s variable selection and theoretical integration.
Third, there are limitations in variable measurements. Although this study has log-transformed the income variable and controlled for key factors such as education and political identity, the hukou variable is simplified into an urban-rural binary opposition, failing to fully reflect the heterogeneity between different hukou types (e.g., the conversion between “residential hukou” and “agricultural hukou”). In addition, social trust, sense of fairness, and sense of well-being are all measured using single-item scales, which may lead to measurement errors and make it difficult to capture more detailed psychological dimensions.
Fourth, the issue of potential omitted variables still exists. This study fails to include important factors such as regional economic development level, urban-rural migration experiences, social network structure, and cultural values, which may lead to an underestimation or bias of the hukou effect.
Finally, the analysis of this study is based on self-reported data, which may be affected by social desirability bias and recall bias. Especially in terms of income declaration and well-being evaluation, respondents may adjust their answers due to social norms or psychological hints.
5.5.2. Future Research
Based on the above limitations, the further research could be elaborated in the following aspects:
1) Dynamic Perspective and Longitudinal Data Analysis
It is suggested to use the annual tracking data of the CGSS or other panel data to explore the long-term impact of hukou system reform on social equality and social attitudes. For example, the differences before and after the hukou policy reforms in 2014 and 2021 can be compared to better reveal the causal mechanisms.
2) Analysis of Multi-Dimensional Hukou Identities and Mobility Paths
Subsequent studies should distinguish between different hukou types (agricultural hukou, residential hukou, military hukou, etc.) and combine the active and passive paths of hukou conversion to examine disparities in social attitudes among groups with different identities. Migrant populations and the “semi-urbanized group” deserve focused attention.
3) Introduction of Psychological and Cultural Variables
To improve the explanatory power of the model, variables such as social capital (e.g., the breadth of interpersonal networks), psychological capital (e.g., optimism, resilience), and cultural values (e.g., collectivist and individualist orientations) can be incorporated to explore the mediating mechanism of “social structure-psychological resources-social attitudes”.
4) Regional Comparison and Institutional Context Analysis
Future research can conduct group analysis in eastern, central and western regions to test how regional economic development and institutional environments interact with the hukou effect. Meanwhile, it deserves consideration how the role of institutional support at the urban-rural community level (e.g., accessibility of public services) matters in the construction of social fairness and well-being.
5) Methodological Expansion
In addition to quantitative regression analysis, future research can adopt mixed methods. Specifically, the combination of in-depth interviews, focus groups, and questionnaire data serves to delve into the subjective experiences and narrative logic of urban and rural residents regarding social fairness, trust, and well-being, thereby supplementing the restrictions of quantitative research.
6) Cross-National Comparative Research
The comparison between China’s hukou system with identity stratification systems in other countries (e.g., legal immigrant status, racial/ethnic identity) helps provide a richer international perspective to understand the universality and peculiarity of the impact of identity systems on social attitudes.