The impact of psychological factors on fertility intention among women who have given birth in Shanghai: multiple mediating effects analysis based on the theory of planned behavior Authors: Taiqing Luo, Qi Tang, Jun Lu, Yelin Li, Anni Wang, Fanlei Kong Published: 2025 ABSTRACT BackgroundThe declining fertility rate posed significant risks to social sustainability and development. However, few studies have explored the psychological factors and their potential mediating effects on the fertility intention for another child. Based on the theory of planned behavior (TPB), this study aimed to investigate the relationships among subjective norm (SN), perceived behavioral control (PBC), attitude (AT), and fertility intention (FI) for another child.MethodsA multicenter cross-sectional survey was conducted from November to December 2022, using a sample of 455 women who had given birth from 3 districts in Shanghai, China. Structural equation modeling was employed to validate the proposed model. Multi-group analysis was conducted to explore differences across groups with different number of children.ResultsPBC (β = 0.212) and AT (β = 0.183) had positive direct effects on FI, while SN had a negative direct effect on FI (β=-0.123). PBC (β=-0.057) and AT (β=-0.035) mediated the impact of SN on FI respectively, functioning as sequential mediators (β=-0.028). The mediating effect of SN on FI accounts for 49.38% of the total effect. Among women with one child group, the path coefficient from PBC to FI was larger than that for women with multiple children (CR = -2.012).ConclusionsSN exerted both direct and indirect negative effects on FI through the multiple mediating effects of PBC and AT, which also acted as sequential mediators. PBC had the strongest direct effect, compared with SN and AT. The effect from PBC to FI for women with one child was stronger than that for women with multiple children. Policy recommendations were provided to increase fertility intentions based on these results.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12978-025-02252-2.Keywords: Fertility intention, The theory of planned behavior, Subjective norm, Perceived behavioral control, Attitude, Mediation Background Fertility intention refers to the extent to which individuals are willing or planning to make efforts to take action toward achieving a certain reproductive behavior [1]. It is widely recognized as a key predictor of actual fertility behavior and overall fertility levels [1, 2]. The decline in fertility rate has become an inevitable trend all over the world, particularly in developed countries [3]. In China, the total fertility rate fell to 1.3 in 2020 [4], below the international warning line of 1.5 [5]. In 2024, the total fertility rate among Shanghai’s registered population dropped to 0.72 [6], highlighting a severe demographic challenge. This trend is likely to give rise to a host of social issues, including a growing elderly population and a shrinking workforce of young and middle-aged individuals [7]. In response to the challenges of population aging and declining fertility rates, China has gradually adjusted its fertility policies to relax fertility restrictions and enhance supporting measures. The government introduced a series of policies from the “Conditional Two-Child Policy” (allowing couples in which at least one of the partners was a single child to have two children) in 2013 to the “Universal Two-Child Policy” (allowing all couples to have two children) in 2016 and then to the “Three-Child Policy” (allowing all couples to have up to three children) in 2021 [8]. These policies primarily targeted women who already had one or more children, with the goal of encouraging them to have additional children [9, 10]. However, the effectiveness of these measures appeared to be limited, as the birth rate continued to decline, from 8.82‰ in 2020 to 6.77‰ in 2024 [11, 12]. By 2023, the total fertility rate had declined to approximately 1.0 [13]. Given that low fertility rates are closely linked with low fertility intention, increasing attention has been paid to the direct effects of demographic, sociological, and economic factors on fertility intention [14–22]. However, most research failed to provide a comprehensive understanding of the internal mechanisms shaping the fertility intention. In addition, previous studies have highlighted significant differences in fertility intentions between childless women and mothers [23]. Research on fertility intention for additional children is particularly critical for countries experiencing low fertility rate, especially as fertility rate continued to decline in more and more countries worldwide [24]. Therefore, it is essential to clarify the factors influencing fertility intention for subsequent children among parous women and to provide valuable insights into addressing fertility challenges. The theory of planned behavior (TPB) is a popular social and psychological model of human intention and behavior. It was first proposed by Ajzen, building on the theory of reasoned action [25]. According to TPB, intention is primarily influenced by three key variables: attitude, subjective norm, and perceived behavioral control. The strength of intention is positively correlated with the likelihood of it translating into actual behavior. Subjective norm is the perceived social pressure from important individuals or organizations to perform or not perform a certain behavior. Attitude refers to an individual’s subjective evaluation of the positive or negative outcomes that a certain behavior may bring them. Perceived behavioral control refers to the individual’s perception of the ease or difficulty of performing a particular behavior. According to TPB, these three factors-attitude, subjective norm, and perceived behavioral control-directly influence intention [1] (Fig. 1). Building on this foundation, recent studies have focused on the relationships between these three factors and the mechanisms through which they shape intention, thus expanding the theory. Some studies have explored the mediating effects of perceived behavioral control and attitude between subjective norm and intention. A study using an extended planned behavior model to explain the willingness to pay to reduce noise pollution hypothesized and verified subjective norm indirectly influences the intention through attitudes and perceived behavioral control [26]. Anderson demonstrated that personal attitudes and perceived behavioral control mediated or partially mediated the relationships between subjective norms and entrepreneurial intentions [27]. Moreover, a small but growing body of research has focused on the chain mediation model of perceived behavioral control and attitude. A study predicting nurses’ intention to undertake dual practice found that subjective norm could affect intention indirectly through the sequential mediation of perceived behavioral control and attitude [28]. Fawehinmi et al. identified the sequential mediating effects of perceived behavioral control and attitude between green human resource management and employee green behavior [29]. External opinions, support, and pressure could significantly influence an individual’s perception and judgment of their ability to perform a certain behavior and their attitude toward it. When external support and assistance is positive, individuals are more likely to feel confident in their ability to complete the behavior and perceive favorable outcomes from it [26]. The theory of planned behavior applied to fertility decisions Based on the above, the hypotheses of this study were proposed as illustrated in Fig. 2 (as follow). Research hypothesis model Hypothesis 1 Subjective norm, attitude, and perceived behavioral control are positively associated with fertility intention. Hypothesis 2 Perceived behavioral control and attitude play a mediating role in the relationship between subjective norm and fertility intention. Hypothesis 3 Perceived behavioral control and attitude sequentially mediate the relationship between subjective norm and fertility intention. Existing research has found that fertility intention and influencing factors vary based on the number of children women already have [23, 30, 31]. After childbirth, women’s experiences of pregnancy, childbirth, and child-rearing may shape their subjective norm, attitude, and perceived behavioral control, subsequently influencing their fertility intention. This suggests that the mechanism by which the TPB influences fertility intention may differ depending on the number of children a woman has. Therefore, from a social psychological perspective, based on TPB, this study aimed to investigate the relationship among subjective norm, perceived behavioral control, attitude, and fertility intention of women who have given birth. Specifically, it explored the sequential mediating effects of perceived behavioral control and attitude. Furthermore, this study also examined how the mechanisms these three factors influenced fertility intentions may differ across groups with different number of children. Methods Study design and sample This multicenter cross-sectional study was conducted from November to December 2022 in Shanghai, China. Considering the influence of economic development and geographical location, simple random sampling was employed to select one central urban district, one inner suburban district, and one outer suburban district from the 16 districts of Shanghai as the sample regions. Due to the lack of a full sampling list, convenience sampling was then used to distribute questionnaires to eligible participants from maternal and child health care centers in the sample area. The inclusion criteria for participants were as follows: women aged 20–45 years; having at least one living child; residing in Shanghai for more than three years; no history of severe physical or mental diseases; and voluntary to participate in this study. Participants who were pregnant or had undergone permanent sterilization were excluded. Participants were well-informed and had the right to refuse or quit the study at any point. There are different guidelines for calculating the sample size required for structural equation modeling (SEM). Generally, approximately 200 samples are considered appropriate for such analyses. Kline suggests that the sample size should be proportional to the number of free parameters, with each free parameter requiring 5–10 samples [32]. In the model used in this study, there are a total of 36 free parameters, which means the minimum required sample size should be 180. Based on these criteria, the minimum sample size should be 200. The study received ethical approval from the Medical Research Ethics Committee of the School of Public Health, Fudan University, China (Grant Number: IRB#2019-10-0783). Data collection The survey questionnaire consisted of three sections (Supplementary File 1): (1) an introduction to the study purpose, (2) demographic and socioeconomic information (e.g., age, education level, and number of children), and (3) items assessing participants’ perceptions of constructs in the proposed model. The third section was developed based on established TPB-related literature and scales, adapted to the study context. The questionnaire contained 15 items: four measuring subjective norm, six measuring perceived behavioral control, four measuring attitude, and one measuring fertility intention. Each item was rated on a 5-point Likert scale ranging from (1) strongly disagree to (5) strongly agree (see Table 1). Prior to the formal survey, the questionnaire was pilot tested to refine its details further. Table 1Items of constructsConstructItemContentSourceAttitude (AT)AT1Having another child will increase your security that at old age there is someone to care about you.Billari et al.[33]AT2Having another child will increase joy and satisfaction in your life.Billari et al.[33]AT3Having another child will increase certainty in your life.Billari et al.[33]AT4Having another child will not decrease time for your personal interests, for contacts with friends.Billari et al.[33]Subjective norm (SN)SN1People who are important to you think that you should have another child.Park et al.[34]SN2People who are important to you support having another child.Park et al.[34]SN3People who are important to you encourage you to have another child.Park et al.[34]SN4Most people around you believe that they should have multiple children.Park et al.[34]Perceived behavioral control (PBC)PBC1The decision about whether to have another child depends on your economic situation.Mencarini et al.[35]PBC2The decision about whether to have another child depends on your job.Mencarini et al.[35]PBC3The decision about whether to have another child depends on your health.Mencarini et al.[35]PBC4The decision about whether to have another child depends on your ability to care for the children.Mencarini et al.[35]PBC5The decision about whether to have another child depends on the help from your partner.Mencarini et al.[35]PBC6The decision about whether to have another child depends on the help from your non-cohabiting relatives.Mencarini et al.[35]Fertility intention (FI)FIDo you intend to have another child?Dommermuth et al.[36] Items of constructs The investigators, who were familiar with the study background, objectives, target population, and questionnaire content, distributed the survey with assistance from doctors at maternal and child health care centers. The questionnaire was edited using Questionnaire Star, a widely used online survey platform in China, and was accessed by participants via QR code or WeChat link. The survey took approximately 5 min to complete, and the platform automatically flagged any missing responses before submission. To ensure data quality, each IP address and WeChat account was limited to one submission. In total, 500 questionnaires were collected, of which 455 were valid, yielding an effective response rate of 91.0%. Data analysis Data analysis was performed in SPSS 27.0 (SPSS Inc., Chicago, IL, USA) and AMOS 28.0 (IBM Corp., Armonk, NY, USA). Descriptive statistics were employed to summarize the sociodemographic characteristics of the participants. Internal reliability was assessed using Cronbach’s α and composite reliability, employing a criterion of α and composite reliability values > 0.70. Convergent validity was evaluated based on the average variance extracted (AVE) and factor loadings, with an AVE threshold of at least 0.50 and a factor loading threshold of at least 0.70. Discriminant validity was examined using the square root of the AVE and a cross-loading matrix. For satisfactory discriminant validity, the square root of the AVE for each construct should be greater than the correlation coefficients between constructs [32]. The chi-square (χ2), degrees of freedom (df), χ2/df, incremental fit index (IFI), comparative fit index (CFI), goodness-of-fit index (GFI), adjusted goodness of fit index (AGFI), normed fit index (NFI), and root mean square error of approximation (RMSEA) were employed as indicators of model fit [32]. χ2/df < 5, values of IFI, CFI, GFI, and NFI greater than 0.90, and value of AGFI greater than 0.80 are considered to reflect a good model fit. RMSEA value less than 0.05 suggests good fit, while value up to 0.08 indicates reasonable errors of approximation. To make the results more reliable, the bias-corrected bootstrapping with 1,000 subsamples was used to compute the 95% confidence interval (CI) and p value [32]. Multi-group analysis was conducted to examine differences and similarities across subgroups. Considering the sample size, the data were categorized into two subgroups: one group with one child and another with multiple children [37]. The change in RMSEA (ΔRMSEA) and the change in CFI (ΔCFI) were used to assess the measurement invariance for the comparison of the less restricted model with the more constrained model, and ΔCFI < 0.010 and ΔRMSEA < 0.015 imply that measurement invariance has been successfully established [38]. The critical ratios (CR) of differences between parameters were calculated to evaluate group differences in regression coefficients for each path [39, 40]. The significance level for all analyses was set as p < 0.05. Results Demographic characteristics The demographic features of respondents are presented in Table 2. Among the participants, 39.12% (178/455) were aged between 31 and 35 years, 65.49% (298/455) had one child, 54.07% (246/455) had completed university or higher education, 36.92% (168/455) reported a per capita monthly household income ranging from ¥5,000 ($696.68) to ¥7,999 ($1114.55), and 34.73% (158/455) were employed in enterprise. Table 2Participants characteristics (N = 455)Characteristic n %Age (years) ≤ 25306.59 26–3017438.24 31–3517839.12 ≥ 367316.04Number of children 129865.49 213629.89 ≥ 3214.62Educational level Middle school or below5912.97 High school5010.99 Junior college10021.98 University or above24654.07Per capita monthly household income (¥; ¥1=$0.139336) <49998919.56 5000–799916836.92 8000–11,9999420.66 ≥ 12,00010422.86Career Unemployed8518.68 Public institution12828.13 Enterprise15834.73 Self-employed8418.46 Participants characteristics (N = 455) Measurement model testing The loading, AVE, Cronbach α, and composite reliability are presented in Table 3. The calculated Cronbach α values ranged from 0.848 to 0.959 and composite reliability values ranged from 0.860 to 0.960, indicating strong internal reliability. Table 3 also showed that the item loadings ranged from 0.738 to 0.941 and AVE ranged from 0.606 to 0.801, all surpassing the recommended thresholds. Therefore, the conditions for convergent validity were satisfied in this study. Additionally, the square root of AVE for subjective norm was 0.778, for perceived behavioral control was 0.895, and for attitude was 0.869. For each construct, the square root of the AVE exceeded the inter-construct correlations, indicating a satisfactory level of discriminant validity. Table 3Measurement modelConstructs and itemsLoadingAVEComposite reliabilityCronbach αSubjective norm (SN)0.6060.8600.848SN10.738SN20.821SN30.801SN40.750Perceived behavioral control (PBC)0.8010.9600.959PBC10.894PBC20.928PBC30.941PBC40.910PBC50.876PBC60.814Attitude (AT)0.7550.9250.922AT10.760AT20.908AT30.915AT40.883AVE average variance extracted Measurement model AVE average variance extracted Model fit testing Using the maximum likelihood method, the research model exhibited an ideal goodness of fit, with all fitness indices meeting practical criteria or threshold values (χ2 = 307.298, df = 140, χ2/df = 2.195, IFI = 0.972, CFI = 0.972, GFI = 0.936, AGFI = 0.936, NFI = 0.949, RMSEA = 0.051), indicating that the data fitted our conceptual research model very well. Structural model testing and mediation model analysis The effects of each variable on fertility intention are summarized in Table 4. Subjective norm (β=-0.243), perceived behavioral control (β = 0.317), and attitude (β = 0.183) were identified as significant predictors of fertility intention. Specifically, subjective norm had both a direct (β=-0.123) and an indirect effect (β=-0.120), with the indirect pathway via perceived behavioral control accounting for the majority of indirect effect (β=-0.057). Perceived behavioral control had a significant direct (β = 0.212) and an indirect effect (β = 0.105). Perceived behavioral control and attitude acted as sequential mediators between subjective norm and fertility intention (β=-0.028). The mediation effect accounted for 49.38% of the total effect of subjective norm. Table 4The effect of each variable on fertility intention in the modelVariablePathwayβBias corrected 95% CI(lower bound, upper bound)p valueSNtotal effect-0.243(-0.346,-0.144)0.002direct effect-0.123(-0.224,-0.024)0.014indirect effect-0.120(-0.178,-0.071)0.001via PBC-0.057(-0.103,-0.025)0.001via AT-0.035(-0.077,-0.011)0.005via PBC and AT-0.028(-0.058,-0.011)0.002PBCtotal effect0.317(0.232,0.392)0.003direct effect0.212(0.106,0.314)0.002indirect effect0.105(0.040,0.181)0.005ATtotal effect0.183(0.062,0.301)0.007direct effect0.183(0.062,0.301)0.007indirect effect0--CI confidence interval, SN subjective norm, PBC perceived behavioral control, AT attitude The effect of each variable on fertility intention in the model CI confidence interval, SN subjective norm, PBC perceived behavioral control, AT attitude Multi-group analysis Table 5 presents the results of model invariance test. The model fitness indices of one child group and multiple children group were compared to assess whether group comparison was appropriate. The ΔCFI and ΔRMSEA values were calculated to determine whether measurement invariance was achieved. The fitness indices of the one child group (M1) were χ2/df = 1.979, GFI = 0.930, AGFI = 0.900, NFI = 0.959, IFI = 0.979, CFI = 0.979, and RMSEA = 0.057. The fitness indices of multiple children group (M2) were χ2/df = 1.979, GFI = 0.883, AGFI = 0.833, NFI = 0.917, IFI = 0.957, CFI = 0.957, and RMSEA = 0.079. The results implied the measurement invariance between two groups on the other models could be furtherly explored. All ΔCFI values were less than 0.010 and all ΔRMSEA values were less than 0.015, implying that the measurement invariance was established across the models. Therefore, the multi-group restriction models have the property of constancy and can be estimated across group categories. Figures 3 and 4 show the path coefficients of one child group and multiple children group, respectively. Table 6 presents the critical ratios of differences between path coefficients for multi-group comparison. The path coefficient from perceived behavioral control to fertility intention was significantly higher for one child group compared to the multiple children group (CR=-2.012, p = 0.044). Table 5Model invariance test using multi-group analysisModelχ2dfχ2/dfGFIAGFINFIIFICFIRMSEAΔCFIΔRMSEAM1166.240841.9790.9300.9000.9590.9790.9790.057--M2166.204841.9790.8830.8330.9170.9570.9570.079--M3332.6011681.9800.9130.8760.9450.9720.9720.047--M4349.2201821.9190.9100.8810.9420.9710.9710.0450.0010.002M5353.2301851.9090.9080.8810.9410.9710.9710.04500M6353.4451861.9000.9080.8820.9410.9710.9710.04500M7355.2431881.8900.9080.8820.9410.9710.9710.04400.001M1 one child group, M2 multiple children group, M3 unconstrained, M4 measurement weights, M5 structural weights, M6 structural covariances, M7 structural residuals, χ2 chi square, df degrees of freedom, GFI goodness of fit index, AGFI adjusted goodness of fit index, NFI normed fit index, IFI incremental fit index, CFI comparative fitness index, RMSEA root-mean square error of approximation, ∆CFI change in CFI, ∆RMSEA change in RMSEA Model invariance test using multi-group analysis M1 one child group, M2 multiple children group, M3 unconstrained, M4 measurement weights, M5 structural weights, M6 structural covariances, M7 structural residuals, χ2 chi square, df degrees of freedom, GFI goodness of fit index, AGFI adjusted goodness of fit index, NFI normed fit index, IFI incremental fit index, CFI comparative fitness index, RMSEA root-mean square error of approximation, ∆CFI change in CFI, ∆RMSEA change in RMSEA Structural equation modeling analysis of one child group. SN = subjective norm, PBC = perceived behavioral control, AT = attitude, FI = fertility intention Structural equation modeling analysis of multiple children group. SN = subjective norm, PBC = perceived behavioral control, AT = attitude, FI = fertility intention Table 6Path coefficients and critical ratios for multi-group comparisonPathOne childMultiple childrenCRp valueβp valueβp valueSN-PBC-0.218< 0.001-0.364< 0.001-1.8400.066SN-AT-0.204< 0.001-0.1410.0550.4851.372SN-FI-0.1350.022-0.1060.2250.3041.239PBC-AT0.529< 0.0010.655< 0.0010.5511.418PBC-FI0.275< 0.0010.0630.581-2.0120.044AT-FI0.1750.0110.2940.0150.6601.491SN subjective norm, PBC perceived behavioral control, AT attitude, FI fertility intention, CR critical ratios Path coefficients and critical ratios for multi-group comparison SN subjective norm, PBC perceived behavioral control, AT attitude, FI fertility intention, CR critical ratios Discussion The impact of subjective norm on fertility intention Subjective norm had a significant direct negative impact on fertility intention (β=-0.123). This was contrary to our hypothesis and inconsistent with previous research findings [41, 42]. For instance, a study investigating the childbearing intentions among women in Iran found that subjective norm was a significant positive predicting factor for childbearing intention [43]. This discrepancy may be due to the high expectations from family, friends, and society for women to have more children, which could translate into negative pressure [44]. This phenomenon could be seen especially in countries with strong family ties, such as Italy [45]. According to psychological reactance theory, when individuals perceive external expectations as threatening their autonomy, they may develop resistance or oppositional motivation toward the expected behavior, thereby reducing their intention to comply [46]. The impacts of perceived behavioral control and attitude on fertility intention Perceived behavioral control and attitude had a significant direct positive impact on fertility intention, validating our hypothesis. Consequently, hypothesis 1 was partially supported. This finding was consistent with existed research [45, 47, 48]. Interestingly, among the three variables, perceived behavioral control had the strongest impact on fertility intention. This was different from previous research findings, which typically indicated that attitude has the greatest impact [1, 27, 28]. This may be due to women adopting a more rational approach, considering both subjective and objective factors related to childbirth and child-rearing. Particularly in Shanghai, an international metropolis characterized by a fast-paced lifestyle, high living costs, and intense work pressure, women tend to prioritize their independence and personal development [49]. As a result, they focus more on material conditions such as financial stability, health, career development, and available support, rather than solely on external influences or psychological evaluations. This shift reflects the current trend where women view childbirth as a rational decision [50], weighing the actual pros and cons, rather than being driven solely by traditional family values, gender roles, or the desire to continue the family lineage. Similarly, Novelli et al. realized that in conditions of mounting uncertainty, women in Italy appeared to be more careful in their childbearing plans [51]. The mediating effect of perceived behavioral control and attitude on fertility intention Perceived behavioral control and attitude respectively mediated the impact of subjective norm on fertility intention, and they also functioned as sequential mediators. The mediating effect of subjective norm on fertility intention accounted for 49.38% of the total effect, nearly half. The finding validated our hypotheses 2 and 3, aligning with Bai’s study [28]. While previous research primarily focused on the direct impacts of those variables on intentions, this study was one of the few that examined their mediating relationships [26]. Among the three mediation paths, the mediating effect of subjective norm through perceived behavioral control accounted for 52.68% of the total mediating effect, indicating that it played a major role, similar to the results of direct effects. According to stimulus-organism-response theory [52], external influences must be psychologically evaluated before they can concretely impact intentions or behaviors. The external pressure for women to have more children will only translate into corresponding intentions when they weigh the positive and negative outcomes of this behavior and assess whether their subjective and objective conditions support it. The difference in path coefficients between subgroups In the multi-group analysis, it was found that the path coefficient from perceived behavioral control to fertility intention for women with one child was larger than that for women with multiple children (CR=-2.012). This difference may stem from the fact that mothers with only one child have less parenting experience and are less familiar with the challenges of raising children. As a result, they place more emphasis on evaluating whether they have sufficient resources and support to raise a second child, making perceived behavioral control a more influential factor in their decision-making. In contrast, mothers with multiple children have more parenting experience and a clearer expectation regarding the challenges of raising additional children. Their evaluations of available resources and support are grounded in prior parenting experiences. Consequently, their fertility intention is likely based more on actual experiences rather than perceived behavioral control. Additionally, Zhu et al. found that the ideal number of children among couples in Shanghai with one child was 1.61 ± 0.52, whereas for couples with two kids, it was 2.06 ± 0.34 [14]. This indicates the intention of women with multiple children to have another child may be more stable and thus less influenced by other factors [3]. Implication Based on the findings of this study, several practical suggestions are offered. First, efforts should be made to respect mothers’ reproductive autonomy by fostering equal communication and avoiding external pressure from families and society. Second, both family members and society should promote the equal sharing of childcare responsibilities to alleviate the disproportionate burden on women. Reducing the conflict between family obligations and career development can help ease women’s psychological stress and enhance their positive perceptions of childbearing. Third, the government could enhance the effectiveness of fertility policies by expanding financial support, such as including childcare expenses in the scope of special additional deductions for individual income tax and providing housing subsidies for families with children. This is particularly important in Shanghai, where the per capita expenditure on education and culture ranks first in China [3]. Employers should be encouraged to provide childcare-friendly work environments by providing nursing rooms, flexible work schedules, and protection against discrimination during pregnancy and career development [53]. Moreover, differentiated support strategies can be tailored for various groups: for women with one child, psychological support and enhancing perceived control over childbearing may be prioritized; for women with multiple children, the emphasis may be placed on reinforcing their existing parenting experiences and addressing the practical challenges encountered in raising additional children. Limitations and future research directions This study had some limitations. First, the factors influencing women’s fertility intention are multifaceted and their underlying mechanisms are complex. This study explored the relationships among TPB variables from a social psychological perspective. Future research could incorporate factors from various angles, such as policy, society, family, and individual, to comprehensively analyze their impact on fertility intention. Second, the sample in this study was drawn from Shanghai, which is at the forefront of economic and social development in China. Therefore, the findings may not be representative of the entire country. Future research could expand the sample to include multiple cities with varying economic and social profiles. Third, although the study districts were selected through simple random sampling, convenience sampling was used at the participant level, which may introduce selection bias and limit the generalizability of the findings. The results should be interpreted with caution. Future research could consider multi-stage probability sampling to further improve representativeness. Conclusions Subjective norm negatively impacted fertility intention, while perceived behavioral control and attitude had positive effects, with perceived behavioral control having the strongest direct effect. Perceived behavioral control and attitude sequentially mediated the effect of subjective norm on fertility intention. The effect from perceived behavioral control to fertility intention for women with one child was stronger than that for women with multiple children. The findings suggest that the government and society should avoid placing additional external pressure on women and provide substantial fertility support to help alleviate the burdens of childbirth and child-rearing. This was essential for genuinely increasing their fertility intention and achieving the goals of fertility policies, particularly in cities like Shanghai, where high living costs and societal pressures create additional challenges for women. Supplementary Information Supplementary Material 1. Acknowledgements The authors would like to thank the study participants for their cooperation and contribution. Abbreviations Theory of planned behavior Subjective norm Perceived behavioral control Attitude Fertility intention Structural equation modeling Average variance extracted Incremental fit index Comparative fit index Goodness-of-fit index Adjusted goodness of fit index Normed fit index Root mean square error of approximation Confidence interval The change in root mean square error of approximation The change in comparative fit index Critical ratios Authors’ contributions TQ L and Q T contributed to the design, acquisition, interpretation of data and drafted the work. J L and FL K contributed to the conception, acquisition of data and approved the submitted version. YL L and AN W contributed to the acquisition and interpretation of data. All authors contributed to substantial revision of the manuscript. Funding This study was supported and funded by the National Social Science Fund of China (No. 17ZDA078). Data availability The datasets used and analysed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate The study was approved by the Ethics Committee of the School of Public Health, Fudan University, China (Approval Number: 2019100783). This study was conducted in accordance with the Declaration of Helsinki. All respondents were well-informed and participated voluntarily. Consent for publication Not Applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Taiqing Luo and Qi Tang authors share equal first authorship. Contributor Information Jun Lu, Email: lujun@shmu.edu.cn. Fanlei Kong, Email: kongfanlei@sdu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets used and analysed during the current study are available from the corresponding author on reasonable request.