Our data were derived from a large-scale cross-sectional survey, the Psychology and Behavior Investigation of Chinese Residents in 2021 (PBICR 2021) [https://www.x-mol.com/groups/pbicr]. This survey was conducted from July 10 to September 15, 2021. The survey employed a multistage sampling method across 31 provinces/autonomous regions/municipalities in mainland China. Using the random number table method, 120 cities were selected, including the capital and two to six prefectural cities in each province/ autonomous region. Based on data from the Seventh National Population Census of China (2021), residents from these 120 cities were selected using quota sampling (quota attributes: sex, age, and urban–rural distribution), ensuring the distributions of these variables in the final sample aligned with population characteristics. Goodness-of-fit tests revealed no significant differences between the sample and the national population across key demographic variables. Given the close alignment of the sample distribution with the national demographic structure after quota adjustment, sampling weights were not applied in the subsequent analysis. We administered 11,709 questionnaires and collected 11,031 valid responses, yielding a response rate of 94.2%.

All participants provided informed consent prior to data collection, and all data were handled with strict confidentiality. This study received ethics approval from the Jinan University Ethics Committee (Approval No. JNUKY-2021-018) and was performed in full adherence to the guidelines and regulations of the Declaration of Helsinki.

The inclusion criteria were as follows: (1) participants held Chinese nationality; (2) participants were women aged 20–49 years who were not currently pregnant. Although China's family planning regulations define women's reproductive age as 15–49 years, the legal age of marriage for women in China is 20 years. Therefore, this study selected data for women aged 20–49 years (21). (3) participants volunteered to participate in the research and completed a consent form; (4) were able to understand the content of each questionnaire item. The exclusion criteria included (1) participants with unconsciousness or severe mental disorders and (2) those unwilling to cooperate or be involved in similar projects. Ultimately, according to the inclusion and exclusion criteria, we obtained a valid sample of 2,973 for this study after deleting data with missing values and outliers.

Latent Profile Analysis (LPA) was conducted using Mplus 8.3 to identify unobserved subgroups of fertility intention among reproductive-aged women. The number of latent classes was not predetermined. The modeling process started with a single-class model and sequentially increased the number of classes. Model estimation and evaluation were performed iteratively until the optimal solution was achieved. The following fit indices were used for model selection: the Akaike Information Criterion (AIC), the Bayesian Information Criterion (BIC), and the sample size-adjusted BIC (aBIC), with lower values indicating better fit. Classification accuracy was assessed by entropy, ranging from 0 to 1, with values ≥ 0.8 suggesting adequate classification. The Lo-Mendell-Rubin Likelihood Ratio Test (LMRT) and the Bootstrapped Likelihood Ratio Test (BLRT) were used to compare the fit between the k-class and (k-1)-class models, where a significant result (P < 0.05) supported the k-class solution. Additionally, the classification probability matrix was examined; diagonal values greater than 0.7 indicated good classification reliability.

Following the identification of the optimal latent profile model, subsequent analyses were conducted using SPSS 27.0. Continuous variables that followed a normal distribution were presented as mean ± standard deviation (SD), while categorical variables were presented as frequency (percentage) [n %)]. Group comparisons were performed using the chi-square test (or Fisher's exact test when expected cell counts were <5) for categorical variables and one-way analysis of variance (ANOVA) for normally distributed continuous variables. Variable selection for multivariable analysis was based on both bivariate associations and theoretical relevance within the Socio-ecological Model (SEM) framework. Therefore, all variables that were either statistically significant in bivariate analyses (p < 0.05) or deemed theoretically important for fertility intention were retained as candidates for the final model. A multinomial logistic regression model was then constructed with the latent fertility intention profiles as the dependent variable to identify factors independently associated with profile membership. A two-tailed p-value < 0.05 was considered statistically significant in all analyses. In the present study, latent class models ranging from 1 to 4 classes were sequentially fitted to the fertility intention scores of reproductive-aged women. The model fit indices for each profile are presented in Table 1.

This study enrolled 2,973 women of childbearing age, of whom 2,262 (76.1%) were urban residents and 711 (23.9%) were rural residents. The 20–30 age group constituted a substantial proportion of the sample, while the majority of families had a per capita monthly income ranging between 3,001 and 6,000 yuan. Additional demographic details are presented in Table 2. Table 2 examined associations between demographic, socioeconomic, psychological factors and latent fertility intention classes via chi-square test, Fisher's exact test, and ANOVA. Results showed no significant differences in educational level (χ² = 8.302, p = 0.405) and monthly household per capita income (χ² = 14.448, p = 0.273), while other factors (age, marital status, family type, etc.) differed significantly across the three classes (all p < 0.001).

In this study, we sequentially constructed latent class models with 1–4 classes to fit the fertility intention scores of reproductive-aged women, and the corresponding model fit indices are summarized in Table 1. As the number of classes increased, the AIC, BIC, and aBIC values consistently decreased, reaching their minimum in the 4-class model. For both the 2-class and 3-class models, entropy values exceeded 0.8, with negligible differences between them. The LMRT and the BLRT results for these two models were statistically significant (both p < 0.05). In contrast, neither the LMRT nor the BLRT reached statistical significance for the 4-class model (both p < 0.05).

Further model selection was guided by interpretability and clinical relevance. In the 4-class model, certain subgroups comprised a relatively small proportion of participants. Excessive partitioning may lead to fragmented classification with limited practical applicability, and therefore this model was not retained. Although the 2-class model demonstrated statistically significant LMRT and BLRT results, its AIC, BIC, and aBIC values remained higher than those of models with more classes, and the two resulting subgroups showed a markedly uneven distribution of participants, which diminished the utility of the classification for targeted intervention. In contrast, the 3-class model exhibited improved model fit indices compared to the 1-class and 2-class models, along with reasonable subgroup sizes and clearer interpretability. Thus, based on a comprehensive evaluation of statistical fit, classification accuracy, and practical relevance, the 3-class model was selected as the optimal solution for capturing latent profiles of fertility intention among reproductive-aged women.

Based on the optimal three-class model, the latent profile distributions of the three fertility intention items among reproductive-aged women are illustrated in Figure 2. Class 1 (n = 746, 25.1%) was characterized by low overall scores and a descending trajectory across items, reflecting a gradually decreasing fertility intention; this group was labeled the “Fertility Intention Decline Group.” Class 2 (n = 1,526, 51.3%) displayed moderate mean scores relative to the other two classes and was designated the “Low Fertility Intention Group.” Class 3 (n = 702, 23.6%) consistently demonstrated the highest mean scores across all items and was therefore identified as the “High Fertility Intention Group.”

Univariate analysis results indicated that there were statistically significant differences in factors influencing fertility intention fertility intention among reproductive-aged women across different latent classes (Table 2). Except for educational level and monthly household income (all p > 0.05), all other factors showed statistically significant differences between the classes (all p < 0.05).

A multinomial logistic regression was performed to identify factors associated with latent fertility intention subgroups. The dependent variable was the three-class fertility intention profile (1 = Fertility Intention Decline Group, 2 = Low Fertility Intention Group, 3 = High Fertility Intention Group). Variables that exhibited statistically significant differences in the univariate analyses (see Table 3 for definitions) were entered as independent variables, with the Fertility Intention Decline Group serving as the reference category.

The results revealed that, compared to the Fertility Intention Decline Group, reproductive-aged women who were part of spousal or nuclear families, aged 20–40 years, resided in urban areas, had two children, or were retired showed a significantly higher likelihood of belonging to the Low Fertility Intention Group (all p < 0.05). Conversely, with no children, or with higher depression scores were significantly more likely to be classified into the High Fertility Intention Group relative to the reference group [(all p < 0.05); see Table 4].

Three distinct fertility intention profiles were identified among reproductive-aged women: a Fertility Intention Decline Group (25.1%), a Low Fertility Intention Group (51.3%), and a High Fertility Intention Group (23.6%). These findings confirm substantial heterogeneity in fertility intention, addressing a key limitation of conventional variable-centered approaches that often overlook within-population differences. Accordingly, this study offers a person-centered perspective that advances the understanding of the complex structure of fertility intention in this population.

The Fertility Intention Decline Group (25.1%) was characterized by a sharp decrease in scores across fertility intention items. Although this group reported a relatively high initial overall fertility intention, their willingness to have a second or third child declined markedly. This discrepancy may be attributable to the combined physical and psychological transitions associated with childbirth, as well as the multifactorial nature of fertility decisions, which often encompass economic pressures, perceived constraints on personal autonomy, and persistent work-family conflict (25). When these cumulative stressors exceed an individual's coping capacity, the positive affective rewards of parenting may be overshadowed, heightening emotional distress and further dampening the intention for subsequent children (26). Consequently, women within this profile may engage in a reassessment of their childbearing plans (27).

The Low Fertility Intention Group comprised 51.3% of the sample, showing no significant within-group differences across the three fertility intention items, yet its scores were consistently and significantly lower than those of the other two groups. As this group constitutes over half of the total participants, it reflects a prevalent pattern of low fertility intention. This finding is consistent with those of Zhao et al. (28), who also reported low overall fertility intention among reproductive-aged women in China. Potential reasons include the following: With the rapid development of the internet, women of reproductive age raised in a multicultural context are exposed to diverse fertility values. They no longer adhere to the traditional marital concept of their parents that “marriage and having children are a matter of course”; instead, they are more focused on self-development and personal autonomy (29). Furthermore, child-rearing entails substantial investment of time, energy, and financial resources, often at the cost of sleep, leisure, social engagement, and individual growth after childbirth.

The High Fertility Intention Group (23.6%) exhibited the strongest overall fertility intention, yet also displayed the highest level of depression. This contradictory finding differs from some previous studies (14), which typically associated poorer mental health with lower fertility intention. An examination of underlying mechanisms suggests that the depressive symptoms observed in this group may not stem directly from childbearing per se, but rather from a perceived “expectation-support mismatch”: women in this profile often endorse strong ideals of family completeness while simultaneously recognizing the substantial investments required for raising multiple children, including time demands, economic pressures, and limited social support. Furthermore, societal expectations surrounding the role of “mothers of multiple children”—such as the pressure to “balance work and family”—likely compound their psychological burden. These findings highlight the importance of attending to the mental well-being of women who intend to have more than one child.

In traditional Chinese society, procreation was culturally mandated to ensure lineage continuity, encapsulated in the saying “having no descendants is the greatest unfilial act.” This belief was central to social identity and closely tied to Confucian filial piety, where bearing children was a key moral duty (45). Since the 1980s, however, family-planning policies and rising costs of child-rearing have weakened traditional norms such as “more children, more blessings,” contributing to today's low fertility. Modern cultural shifts further suppress fertility intention. For many women, childbearing is now a matter of personal choice and gender equality rather than familial obligation. Concepts like “DINK” (double-income, no kids) and “fewer but better-raised children” have reduced desired family size. Finally, emerging social norms also play a role. “Parenting involution”—intense competition over children's education and development—has turned parenting into a source of economic and psychological strain. The anxiety of “not raising children well” is a major concern for those with low or declining fertility intention (44). Although direct evidence is still limited, studies suggest that perceived involution can lower fertility intention. From a resource-scarcity perspective, when individuals perceive limited social and economic resources, they tend to make cautious, often restrictive decisions about having children (46).