This cross-sectional study was approved by the Ethics Committee of Jaume I University (Ref. CD/22/2021; April 15, 2021) and conducted in accordance with the Declaration of Helsinki. Data were collected through an anonymous online survey hosted on Qualtrics between April and June 2022. Participation required approximately 15–20 minutes.

Participants were recruited through targeted social media advertisements (e.g., Facebook, Instagram) and via collaborations with obstetric, midwifery, and maternal health associations in Colombia. Inclusion criteria were as follows: (a) age 18 years or older; (b) Colombian nationality or current residence in Colombia; and (c) being either pregnant or having at least one child aged 0–2 years.

Of the 512 women who accessed the survey, 56 were excluded for the following reasons: missing MAS data (n = 24) or not meeting inclusion criteria for pregnancy or postpartum status (n = 32). The final Colombian sample included 456 women, of whom 296 were pregnant and 160 were postpartum.

To ensure cultural and linguistic appropriateness of the MAS, the original Spanish version (Martín-Sánchez et al., 2022) was reviewed by two Colombian perinatal experts (a midwife and a gynecologist). Minor wording adjustments were made to improve clarity and naturalness for Colombian Spanish while preserving semantic equivalence. No items were added or removed.

For the remaining instruments (see below), the validated Spanish versions were used. These questionnaires were also reviewed by the same Colombian perinatal experts to ensure linguistic suitability. Given their brevity, clarity, and previously established validity in Spanish-speaking populations, no modifications were required.

All participants provided informed consent prior to beginning the survey and were informed of its voluntary and anonymous nature. The Qualtrics platform was programmed to require completion of each questionnaire before proceeding to the next. As a result, there was no item-level missing within any questionnaire (0%), although some participants chose not to complete later instruments. The MAS, administered first, was therefore completed by the full sample (n = 456), while subsequent measures (see below) had smaller valid subsamples, reported in all analyses.

Data analyses followed established guidelines for psychometric evaluation (Appelbaum et al., 2018). All analyses were conducted using IBM SPSS Statistics v27.0 (IBM Corp., Armonk, NY, USA) and R version 4.5.1 for Windows (Version 4.5.1; R Foundation for Statistical Computing, Vienna, Austria), employing the lavaan (Rosseel et al., 2012) and semTools (Jorgensen et al., 2025) packages for CFA and measurement invariance testing.

$\bullet$ Descriptive Statistics: Means (M), standard deviations (SDs), skewness, and kurtosis were calculated for each MAS item to assess distributional properties and univariate normality.

$\bullet$ Structural Validity: CFA was performed to test the hypothesized three-factor structure of the MAS (Doubts, Rejection, and Suppression). Models were estimated using maximum likelihood with robust standard errors (MLR) to account for non-normality. Model fit was evaluated using $\chi$², $\chi$²/df, Comparative fit index (CFI), Tucker-Lewis index (TLI), standardized root mean square residual (SRMR), and root mean square error of approximation (RMSEA) with 90% confidence intervals (CIs).

$\bullet$ Internal Consistency: Cronbach’s $\alpha$, corrected item–total correlations (r_(jx)), and $\alpha$ if item deleted were computed for each MAS subscale to assess reliability.

$\bullet$ Construct Validity: As the Kolmogorov*Smirnov test indicated significant deviations from normality across all scales (p 0.05), Spearman’s rho ($\rho$) correlations were used to examine associations between MAS dimensions (Doubts, Rejection, and Suppression) and psychological adjustment indicators (ODSIS, OASIS, and SWLS). All tests were two-tailed, with p 0.001 considered statistically significant.

$\bullet$ Cross-Cultural Comparisons: Cross-country differences (Colombia vs. Spain) were analyzed using Mann–Whitney U tests due to non-normal distributions and ordinal data. Effect sizes were calculated as r = $|$Z$\mathrm{|}$ / $\surd$N. Comparisons were conducted separately for pregnant and postpartum participants.

$\bullet$ Measurement Invariance: Multi-group CFA was conducted to test configural, metric, and scalar invariance of the MAS across (a) countries (Colombia vs. Spain) and (b) perinatal stages (pregnant vs. postpartum). Invariance was evaluated using $\mathrm{\Delta }$CFI $\le$ 0.010 and $\mathrm{\Delta }$RMSEA $\le$ 0.015, following recommended criteria (Cheung and Rensvold, 2002).

$\bullet$ Assumption Checks and Robust Estimation: Given the ordinal nature of items and observed non-normality, robust MLR estimators were used for CFA, while nonparametric tests (Spearman correlations and Mann-Whitney U tests) were applied for bivariate and group-level analyses. Although the MAS uses a 4-point Likert scale, items were treated as continuous in CFA, because scales with four or more response categories yield comparable estimates under robust maximum likelihood (Park, 2023; Rhemtulla et al., 2012).

$\bullet$ Sensitivity Analyses: Partial correlations controlling for age were conducted to examine potential confounding effects. Adjusting for this covariate did not meaningfully alter the strength or significance of associations between maternal ambivalence, mental health, and life satisfaction.

A total of 456 Colombian women participated in the study (M = 27.5 years, SD = 6.1; range = 18–48 years). Of these, 296 were pregnant (M = 26.2, SD = 5.4; range = 18–42) and 160 were postpartum (M = 29.8, SD = 6.8; range = 18–48). Most participants were in a relationship (87.2%), and nearly half held a university degree or higher (48.9%). Employment status varied: 38.5% employed, 32.9% unemployed, 12.4% homemakers, 5.8% students, and 10.4% on medical leave. 68 participants (16.5%) reported receiving ongoing mental health treatment.

Table 1 displays the descriptive and reliability statistics for the 14 MAS items. All items demonstrated adequate variability and acceptable distributional properties. Item means ranged from M = 1.91 to M = 3.26, reflecting moderate variability in endorsement across items. Skewness values ranged from S = –0.91 to S = 0.60, and kurtosis (K) values ranged from –0.93 to 0.87, which fall within the recommended limits for Likert-type data. Item–total correlations (r_(jx)) ranged from 0.42 to 0.74, and “Cronbach’s alpha if item deleted” ($\alpha$_–x) values indicated that removing any item would not meaningfully improve internal consistency. Standardized factor loadings ($\lambda$) ranged from 0.46 to 0.93, and item-level explained variance (R²) ranged from 0.21 to 0.86, confirming that all items contributed substantially to the underlying latent construct.

CFA results supported the three-factor structure of the MAS (Doubts, Insecurity, and Suppression) in the Colombian sample. The model demonstrated good fit: $\chi$²(74) = 143.71, p 0.001; robust $\chi$² = 143.71; CFI = 0.968; TLI = 0.960; robust RMSEA = 0.050, 90% CI [0.037, 0.062]; SRMR = 0.042. All standardized factor loadings ($\lambda$) exceeded the recommended threshold of 0.32, ranging from 0.418 to 0.829, confirming that each item contributed meaningfully to its latent factor. Item-level explained variance (R²) ranged from 0.175 to 0.687. Covariances between latent factors were positive and significant (Doubts–Rejection = 0.790, Doubts–Suppression = 0.495, Rejection–Suppression = 0.458). These findings replicate the original Spanish validation (Martín-Sánchez et al., 2022) and provide strong evidence for the structural validity of the Colombian version of the MAS.

A series of multi-group confirmatory factor analyses were conducted to examine measurement invariance of the MAS across pregnancy status (pregnant vs. mother; Table 2). The configural model, which allowed factor loadings and intercepts to vary across groups, demonstrated adequate fit: $\chi$²(148) = 284.74, p 0.001, CFI = 0.948, TLI = 0.936, RMSEA = 0.064 (90% CI [0.052, 0.075]), SRMR = 0.055. This indicates that the three-factor structure is comparable across pregnant women and postpartum mothers.

Constraining factor loadings across groups (metric invariance) resulted in minimal changes in fit indices ($\mathrm{\Delta }$CFI = –0.002, $\mathrm{\Delta }$RMSEA = –0.001) and the $\chi$² difference test suggested no significant degradation of model fit, supporting metric invariance. Further constraining item intercepts across groups (scalar invariance) produced minimal changes in fit indices ($\mathrm{\Delta }$CFI = –0.003, $\mathrm{\Delta }$RMSEA = –0.0002) and did not meaningfully worsen model fit, indicating scalar invariance. These results suggest that the MAS measures maternal ambivalence equivalently in pregnant women and postpartum mothers, allowing for meaningful comparisons of latent factor means between groups.

Internal consistency was high across the three subscales: $\alpha$₍Doubts₎ = 0.84, $\alpha$₍Rejection₎ = 0.80, $\alpha$₍Suppression₎ = 0.80.

Bivariate Spearman correlations (Table 3) supported the expected associations between maternal ambivalence dimensions and indicators of psychological adjustment in Colombian perinatal women:

$\bullet$ Doubts were positively associated with depressive symptoms (ODSIS: $\rho$ = 0.50, n = 372, p 0.001) and anxiety (OASIS: $\rho$ = 0.42, n = 358, p 0.001), and negatively associated with life satisfaction (SWLS: $\rho$ = –0.48, n = 416, p 0.001).

$\bullet$ Rejection was positively associated with depression (ODSIS: $\rho$ = 0.40, n = 372, p 0.001) and anxiety (OASIS: $\rho$ = 0.28, n = 358, p 0.001), and negatively associated with life satisfaction (SWLS: $\rho$ = –0.56, n = 416, p 0.001).

$\bullet$ Suppression showed weaker but significant positive correlations with depression (ODSIS: $\rho$ = 0.33, n = 372, p 0.001) and anxiety (OASIS: $\rho$ = 0.25, n = 358, p 0.001), and a moderate negative correlation with life satisfaction (SWLS: $\rho$ = –0.43, n = 416, p 0.001).

These results provide evidence of construct validity, demonstrating that higher maternal ambivalence, particularly in Doubts and Rejection, is associated with greater emotional distress and lower life satisfaction during the perinatal period.

Partial correlations controlling for participants’ age revealed a similar pattern, with only minimal attenuation of effect sizes. For example, Doubts remained strongly associated with Insecurity (r = 0.603) and moderately with Suppression (r = 0.378), and its associations with depressive (r = 0.491) and anxiety (r = 0.448) symptoms remained substantial. Life satisfaction also remained negatively associated with ambivalence and symptoms when controlling for age (e.g., Doubts: r = –0.477; Insecurity: r = –0.556; ODSIS: r = –0.508).

Overall, these results indicate that maternal ambivalence is strongly associated with higher psychological symptoms and lower life satisfaction, and that these associations are largely independent of age.

Group comparisons (Table 4) using Mann-Whitney U tests indicated that pregnant women reported significantly higher Rejection scores (M = 8.19, SD = 2.59; Median [Mdn] = 8) than postpartum mothers (M = 7.27, SD = 2.67; Mdn = 7), U = 18,628.00, Z = –3.792, p 0.001, r = 0.18. Pregnant women also reported significantly lower life satisfaction (SWLS: M = 22.12, SD = 7.18; Mdn = 23) compared with postpartum mothers (M = 23.64, SD = 7.47; Mdn = 25), U = 17,245.00, Z = –2.345, p = 0.019, r = 0.11.

No significant differences were observed between groups for Doubts scores (M = 13.41, SD = 3.66; Mdn = 13 vs. M = 12.79, SD = 4.07; Mdn = 13), U = 21,519.50, Z = –1.614, p = 0.107, r = 0.08; Suppression (M = 9.03, SD = 2.53; Mdn = 9 vs. M = 8.91, SD = 2.90; Mdn = 9), U = 22,867.00, Z = –0.610, p = 0.542, r = 0.03; Depression (ODSIS: M = 10.79, SD = 4.85; Mdn = 10 vs. M = 10.12, SD = 4.84; Mdn = 10), U = 14,548.50, Z = –1.415, p = 0.157, r = 0.07; or Anxiety (OASIS: M = 10.07, SD = 4.54; Mdn = 9 vs. M = 10.01, SD = 4.81; Mdn = 9), U = 14,498.00, Z = –0.344, p = 0.731, r = 0.02.

Significant cross-cultural differences emerged across several MAS subscales (Tables 5,6).

Among pregnant women, Colombian participants (n = 296) reported higher scores than Spanish participants (n = 407) on Doubts (M = 12.56, SD = 3.66; Mdn = 13 vs. M = 11.76, SD = 3.66; Mdn = 12), U = 46,306.50, Z = –5.260, p 0.001, r = 0.20; Rejection (M = 7.59, SD = 2.50; Mdn = 8 vs. M = 6.92, SD = 2.42; Mdn = 7), U = 46,744.00, Z = –5.160, p 0.001, r = 0.20; and Suppression (M = 8.18, SD = 2.60; Mdn = 9 vs. M = 8.17, SD = 2.64; Mdn = 8), U = 39,120.00, Z = –8.020, p 0.001, r = 0.30.

Colombian pregnant women also reported higher depressive symptoms (ODSIS: M = 9.40, SD = 4.63; Mdn = 10 vs. M = 8.70, SD = 4.29; Mdn = 7), U = 30,109.50, Z = –6.320, p 0.001, r = 0.26; and anxiety symptoms (OASIS: M = 9.32, SD = 4.23; Mdn = 9 vs. M = 9.27, SD = 4.18; Mdn = 9), U = 33,659.50, Z = –3.210, p = 0.001, r = 0.13, as well as lower life satisfaction (SWLS: M = 23.44, SD = 6.69; Mdn = 23 vs. M = 23.60, SD = 6.60; Mdn = 25), U = 41,278.00, Z = –3.998, p 0.001, r = 0.17.

Among postpartum women (0–2 years), Colombian participants (n = 160) continued to report higher maternal ambivalence than Spanish participants (n = 834) on Doubts scores (M = 12.79, SD = 4.07; Mdn = 13 vs. M = 11.56, SD = 3.55; Mdn = 12), U = 55,285.00, Z = –3.45, p = 0.001, r = 0.11; Suppression scores (M = 8.91, SD = 2.90; Mdn = 9 vs. M = 8.03, SD = 2.57; Mdn = 8), U = 54,572.50, Z = –3.64, p 0.001, r = 0.13; and Rejection scores (M = 7.27, SD = 2.67; Mdn = 7 vs. M = 6.85, SD = 2.36; Mdn = 7), U = 61,527.00, Z = –1.58, p = 0.115, r = 0.05, although the effect was not statistically significant for Rejection.

Colombian postpartum women also reported higher depressive symptoms (ODSIS: M = 10.12, SD = 4.84; Mdn = 10 vs. M = 8.44, SD = 4.14; Mdn = 8), U = 40,131.00, Z = –3.82, p 0.001, r = 0.13. No significant differences were observed in anxiety (OASIS: M = 10.02, SD = 4.81; Mdn = 10 vs. M = 9.14, SD = 4.05; Mdn = 9), U = 43,498.00, Z = –1.60, p = 0.110, r = 0.05, or life satisfaction (SWLS: M = 23.64, SD = 7.47; Mdn = 23 vs. M = 23.59, SD = 6.42; Mdn = 25), U = 58,636.00, Z = –0.64, p = 0.521, r = 0.02.

These findings indicate consistent cross-cultural differences in maternal ambivalence, with Colombian women generally reporting higher Doubts, Suppression, and Rejection than Spanish women, both during pregnancy and the postpartum period, along with higher depressive symptoms and lower life satisfaction in the pregnant sample.

To evaluate whether the MAS subscales were interpreted equivalently across Colombian and Spanish participants, a sequence of measurement invariance models was tested using CFA (Table 7).

Configural invariance was first established, indicating that the overall factor structure of the MAS was similar across groups. Metric invariance was then tested by constraining factor loadings to be equal across groups. Fit indices showed minimal changes ($\mathrm{\Delta }$CFI = 0.003, $\mathrm{\Delta }$RMSEA = –0.003), and the Satorra-Bentler scaled chi-square difference test was non-significant, supporting metric invariance. These results indicate that the latent constructs were interpreted similarly in both cultural groups, allowing meaningful comparisons of relationships between constructs (e.g., correlations with anxiety or depression).

Scalar invariance, tested by constraining both factor loadings and item intercepts, was not fully supported. Changes in fit indices ($\mathrm{\Delta }$CFI = –0.015, $\mathrm{\Delta }$RMSEA = 0.002) and the chi-square difference test ($\mathrm{\Delta }$$\chi$²(11) = 130, p 0.001) indicated that item intercepts differed between groups. This finding implies that differences in raw scores may reflect both true latent differences and systematic variations in item responses, making direct comparisons of observed tentative means. Partial scalar invariance could be achieved by freeing non-invariant intercepts, allowing for more accurate comparisons of latent means.

Item-level analyses confirmed that all MAS items demonstrated adequate variability, acceptable distributional properties, and strong item–total correlations, indicating that each item contributed meaningfully to its respective latent construct. The three-factor model demonstrated good fit indices (CFI = 0.97, TLI = 0.96, RMSEA = 0.05), with standardized loadings ranging from 0.42 to 0.83 and explained variances (R²) between 0.18 and 0.69. Internal consistency was high across subscales ($\alpha$ = 0.80–0.84), consistent with prior Spanish validations (Martín-Sánchez et al., 2022). These results confirm the structural validity of the MAS in Colombian perinatal women and support its applicability across culturally diverse Spanish-speaking populations.

Multi-group CFA supported full metric and scalar invariance between pregnant and postpartum women, indicating that the MAS measures ambivalence equivalently across perinatal stages. Consequently, pregnant and postpartum women interpret the Doubts, Rejection, and Suppression factors similarly, allowing meaningful comparisons of latent factor means. This invariance underscores the temporal stability of maternal ambivalence, suggesting that ambivalent emotions toward motherhood persist from pregnancy to early motherhood. However, subtle mean-level differences, specifically higher Rejection and lower life satisfaction among pregnant women. suggest that emotional ambivalence may be particularly pronounced during pregnancy, possibly reflecting heightened uncertainty and role adjustment during this stage (Cutler et al., 2018).

Spearman correlations supported the expected associations between maternal ambivalence and psychological adjustment. Doubts and Rejection were moderately to strongly associated with depressive and anxiety symptoms and inversely related to life satisfaction, whereas Suppression showed weaker yet significant correlations with these outcomes. These findings validate the MAS as a multidimensional construct associated with emotional vulnerability during the perinatal period (Pollack, 2024). The persistence of these associations after controlling for age suggests that ambivalence exerts a robust effect on well-being, independent of demographic factors. Importantly, Rejection showed weaker correlation with emotional symptoms than Doubts, suggesting that rejection-related feelings may not necessarily indicate psychopathology but rather reflect the tension between idealized maternal expectations and lived experiences (Newell, 2010). This finding aligns with prior research emphasizing the importance of interpreting ambivalence contextually rather than pathologizing it (Raphael-Leff, 2010; Takševa, 2017).

Consistent and significant differences emerged between Colombian and Spanish women across several dimensions of ambivalence. Colombian participants—both pregnant and postpartum—reported higher levels of Doubts and Suppression, and, to a lesser extent, Rejection. These results suggest that sociocultural factors, including traditional gender norms, ambivalent sexism, and collectivist expectations surrounding motherhood, influence the emotional expression of maternal ambivalence (Alzate et al., 2024; Rodríguez-Burbano et al., 2021).

In Spain, where motherhood is increasingly framed through individualized and egalitarian perspectives (Moreno-Mínguez et al., 2018), ambivalence may be more openly acknowledged and normalized. In contrast, in Colombia, cultural ideals emphasizing self-sacrifice and devotion may discourage emotional disclosure (Urbina-Garcia, 2025), fostering suppression as a culturally sanctioned coping mechanism. Although adaptive for maintaining social harmony, sustained suppression may hinder emotional processing and increase vulnerability to depressive symptoms (Chen, 2025; Tsai et al., 2017)

The persistence of Suppression across both perinatal stages in Colombian women, unlike its decline in Spanish samples, further illustrates the role of cultural values in emotional regulation. Cross-cultural psychology frameworks (Badaan and Choucair, 2023) suggest that maternal ambivalence is not a universal emotional conflict but a culturally mediated experience shaped by gender ideologies, social expectations, and access to emotional validation.

Measurement invariance analyses confirmed configural and metric invariance across the Colombian and Spanish samples, indicating that participants in both countries conceptualize the MAS constructs similarly. However, full scalar invariance was not achieved, as several item intercepts differed between groups. This finding implies that, although relationships among latent constructs (e.g., correlations with depression or anxiety) can be meaningfully compared across countries, comparisons of raw scores should be interpreted with caution. Observed cross-cultural differences in Rejection and Suppression likely reflect both true variations in maternal ambivalence and systematic cultural response patterns.

The significant associations between maternal ambivalence and perinatal distress underscore the need for culturally sensitive prevention and intervention strategies. In Colombia, where expressing doubt or dissatisfaction with motherhood may be stigmatized, psychoeducational programs should normalize ambivalence as a common and valid emotional experience rather than pathologize it. The Suppression dimension suggests that many women inhibit negative maternal feelings, underscoring the importance of therapeutic approaches that promote emotional integration rather than avoidance.

Integrative treatments combining cognitive-behavioral therapy (CBT) techniques (Li et al., 2022) with brief psychodynamic (Thoma and Abbass, 2022) or emotion-focused approaches (Kramer et al., 2026) may be particularly effective. These interventions facilitate the acknowledgment and processing of repressed or conflicting emotions, fostering greater authenticity and self-compassion in the development of maternal identity. Early screening for ambivalence during pregnancy, along with preventive group interventions as recommended in guidelines (Rodríguez-Muñoz et al., 2023), could further help mitigate postpartum distress and enhance well-being.

Despite its contributions, this study has several limitations. Its cross-sectional design precludes causal inferences and limits understanding of how ambivalence evolves and changes across the perinatal timeline. Convenience sampling and reliance on self-report measures may introduce bias and restrict the representativeness of the sample, particularly among women from lower socioeconomic backgrounds. Future research should employ longitudinal and mixed-method approaches to capture the dynamic nature of ambivalence and its interaction with sociocultural and psychological variables. Moreover, examining moderators such as family support, religious beliefs, and gender ideology could clarify how structural factors shape maternal emotional expression across Latin American contexts.

This study constitutes the first validation of the MAS in a Latin American population and demonstrates its psychometric robustness in Colombian perinatal women. It provides empirical evidence of cross-cultural variability in maternal ambivalence and emphasizes that emotional experiences of motherhood are culturally constructed and influenced by social norms and gender expectations. Clinically, these findings highlight the significance of emotional suppression as a culturally mediated coping mechanism and advocate for interventions that facilitate the authentic expression of ambivalence. By integrating brief psychodynamic and emotion-focused components with established CBT frameworks, practitioners may more effectively address the deeper affective dimensions of maternal distress. Overall, the present findings advance cross-cultural psychometric research and contribute to a more nuanced understanding of maternal ambivalence as both a universal and culturally embedded aspect of the perinatal experience.