Method

Results

In the analysis aimed at testing the differences in the means of the first-order factors, it is evident that all the estimated intercepts of the first-order factors for Peru were significant (PEN: 0.170, p = .000; TE: 0.440, p = .000; SCN: 0.084, p = .032; SNC: 0.306, p = .000 and VI: 0.493, p = .000) which means that there are differences between the groups in the intercepts of the first-order factors or between the means of the first-order factors, corresponding to the zero value of the second-order factor. Consistent with the results, the scores for the five specific positive motivations for having children in Peru were significantly higher than those in Brazil. The first-order factors IV (0.493) and TE (0.440) showed the greatest differences between the countries.

In the analysis the aim was to test the difference between the means of the second-order factor; the model results show that the estimated mean MPG factor in Peru is 0.285 (p < .001), which describes the estimated difference in the mean of the second-order MPG factor between the two populations. In other words, the MPG score in the Peruvian sample was 0.285 points higher on average than in the Brazilian sample.

To determine whether the effect of the predictor variables: sex, having or not having children, education, having or not having a partner, labor market insertion, family income, religiosity, and positivity in the MPG remain invariant between the two samples, an MGSEM was performed. The model initially tested is shown in Figure 1.

Figure 1: 

It began by establishing a benchmark structural equation modeling (SEM) for Brazil and Peru that fit the data well, as can be seen in Table 1. In the initial analysis performed on the Brazilian sample, the variables with or without children (-0.289; p < .001) and having or not having a relationship (-0.2078; p = .002) have a significant negative effect on PCM and the variables religiosity (0.374; p < .001) and family income (0.094; p = .001) have a significant positive effect. The variables positivity, gender, schooling and labor market insertion have no effects on PCM'.

In the initial analysis performed in the Peruvian sample, as in the Brazilian sample, the variable having or not having children (-0.171; p < .001) showed a significant negative effect, but unlike what was found in Brazil, the variables level of education (-0.079; p = .016) and income (-0.093; p = .001) also showed this effect on the PCM. As in the Brazilian sample, the variable religiosity (0.330; p < .001) showed a significant positive effect, but unlike what was found in the Brazilian sample, positivity showed a significant positive effect in the Peruvian sample (0.107: p < .001). The variables sex, having or not having a partner and labor market insertion did not present effects on the PCM. It is noteworthy that the variables: sex and insertion in the labor market, did not show effects on the PCM in any sample, therefore, both variables were eliminated from the subsequent analyses.

With these preliminary analyses it was shown that both the benchmark SEM models for Brazil and Peru fit the data, but the estimated path coefficients differ between the two models, we went on to assess whether the variance in the structural path coefficients would imply that relevance in the population moderates the causal relationships in them. To do this, a Multi-group SEM Model of Configuration was estimated with the variables having or not having children, education, having or not having a partner, family income, religiosity, and positivity; where all model coefficients were released in both groups simultaneously. This Model Setting served as the base model for subsequent model comparisons. The fit indices of this base model were adequate (Table 1).

Table 1: Confirmatory factor analysis of PCM 

^Note: PCM = Positive Motivation to Breed; χ ² = chi-square; df = degrees of freedom; CFI = Comparative Fit Index; TLI = Tucker-Lewis Index; RMSEA = Mean Squared Error of Approximation; CI = Confidence Interval.

As a second step, a more restrictive model was analyzed that would be compared with the base model. In this model, equality restrictions were imposed on the effects on all the predictor variables tested previously. The results of the χ² difference test were: χ²= 745.664; df= 20, p < .001, indicating that the effect of the variables tested together on PCM does not remain invariant between the Brazilian and Peruvian samples. Statistically speaking, population membership significantly moderates the effect of these ensemble variables on PCM.

Subsequently, to get a more specific view of the variance of the effect of each variable on the PCM, a separate analysis of each was performed (Table 2).

Table 2: χ² test of difference in individually tested variables 

^Note:χ ² difference results using the Difftest option of Mplus. χ ² = chi-square; df= degrees of freedom; p = statistical significance.

As can be seen in Table 2, the positive and significant effect of the schooling variable on PCM in the Peruvian sample (-0.079; p = .016) and the non-significance of this variable on PCM in the Brazilian sample, is not moderated by belonging to one group or the other. On the other hand, the differences between countries regarding the negative impact of the variable having or not having children (Brazil = -0.289, p < .001; Peru = -0.171; p < .001) and the positive impact of the variable religiosity (Brazil = 0.374, p< .001; Peru: 0.330, p < .001) if influenced by the population they come from. The same occurs with the differences found according to countries in the variables having or not having a partner (Brazil = -0.2078, p = .002; Peru: not significant), family income (Brazil = 0.094, p = .001; Peru = - 0.093, p = .001), and positivity (Brazil: not significant; Peru = 0.107, p < .001).

Discussion

The results show that the Peruvian sample has higher levels of positive motivations for parenthood, both general and specific, than the Brazilian sample. Considering that Peru is the country with the highest birth and fertility rates compared to Brazil, it seems important to highlight that several studies show that positive motivations for parenting are related to behaviors aimed at achieving conception (^{Miller, 2021}; ^{Mynarska & Raybould, 2020}), care aimed at a favorable termination of pregnancy (^{Miller et al., 2004}), and a higher number of children (^{Irani & Khadivzadeh, 2018}).

This difference found can be analyzed considering two points: the socioeconomic situation of the two countries and the socio-demographic differences of the samples of this study. On the first point, Peru is a country that has been showing a steady socioeconomic improvement in recent years (^{Organisation for Economic Co-operation and Development, 2015}) that can generate in citizens a perception of improvement in their economic-family situation (^{Ipsos, 2020}), including a favorable view of what it means to have children. On the other hand, Brazil is a country that has been going through a deep social crisis (^{Cueto & Lopes, 2020}), which is reflected in less favorable social and economic indices than Peru (^{Datos Marco.com., n. d.-a}) that we can relate to a less optimistic evaluation of what it takes to have children. Regarding the second point, although both samples are predominantly female, the Brazilian sample presents higher levels of family income, greater insertion in the labor market and higher education. Profiles of women with these characteristics are usually related to problems in reconciling work and family life (^{Rios-Neto et al., 2018}) and greater reflection on the impact of having children on other personal goals (^{Alves & Cavenaghi, 2019}), which can be seen in the lower levels of positive motivations, since these describe desired aspects of having children.

In relation to the variables tested as predictors of the motivations for parenthood, it can be seen that the impact on the PCM was mediated by the population from which they come from; with the exception of the educational level variable in which the difference found is not moderated by belonging to one country or another. This result corroborates what was proposed by ^{Miller (1992}, ²⁰²¹) that the motivations for parenthood are influenced by the characteristics of the context in which the subject is inserted.

According to the results, religiosity has a greater positive impact on PCM in Brazil than in Peru, and this difference is influenced by nationality. Religion is a variable that transmits family-centered values (^{Miller et al., 2016}) and its relationship with motivations for parenting is something previously found in other studies around the world (^{Ghazanfarpour et al., 2018}; ^{Kubicka et al., 1995}; ^{Miller, 1992}; ^{Miller et al., 2016}).

Latin America has been characterized as a highly religious region (^{Camargo, 2019}; ^{Cipriani, 2015}) and both Brazil and Peru do not escape this characteristic. The difference found can be analyzed considering their religious discrepancies. Brazil has more than 30 religious beliefs (^{Sistema IBGE de Recuperação Automática, n. d.}) and although its population remains predominantly Catholic (61 %), the percentage of Protestants has been systematically growing (26 %). On the other hand, Peru has a smaller number of religious beliefs (11) (Datos Marco.com, n. d.-b) and shows itself as a predominantly Catholic country (76 %) with a relatively low percentage of Protestants (17 %). Both Catholics and Protestants exhibit normative beliefs about family and reproductive aspects (^{Sahgal & Bell, 2014}). Despite this, the study by ^{Sahgal and Bell (2014)} shows that Protestants are more empowered and accepting of the ideas that come with being part of a religious congregation and follow the precepts more faithfully than Catholics.

The results show that the PCM is higher in people who have a partner than in those who do not in the Brazilian sample, not the same with the Peruvian, and this difference was also influenced by the population. In the case of the Brazilian sample, it is evident that having a partner favors the desire to have children, similar to what has been found in studies conducted in other countries (^{Miller & Pasta, 1996}; ^{Miller et al., 2004}; ^{Mitchell & Gray, 2007}). On the other hand, in Peru this factor is not evidenced as relevant. These results confirm the hypotheses of ^{Heuveline and Timberlake (2004}) and ^{Mendes and Pereira (2019}), who point out that the relationship between not having a stable partner and reproductive behavior differs between countries, as each context has a different evaluation of the importance of the partner in achieving reproductive goals.

^{Khadivzadeh et al. (2014}) indicate the need for a satisfactory marital relationship for increased positive motivation to have children, while other studies mention that this increase may also be due to an attempt to improve a deteriorated marital relationship (^{Testa, 2012}; ^{Testa et al., 2012}). In the present study, the variable satisfaction/ dissatisfaction with partner was not measured in order to use it as a control variable, an aspect that could have given us a broader view of the reasons for the difference found between the countries.

Regarding the positivity variable, the results indicate that higher levels of positivity impact on higher levels of motivation for positive parenting in the Brazilian sample, which shows the need for a positive overall view of life for a positive view of parenting. On the other hand, in the Peruvian population this relationship is not found, showing that although the general evaluation of life is not positive, this is not related to a higher or lower positive motivation to have children, which may be a point to consider in the analysis that the Peruvian population shows higher levels of motivation for positive parenting, despite the fact that, like other countries in the region, it is going through economic and social crises. The relationship between both variables has not been previously studied in Brazilian and Peruvian samples (^{Varas & Borsa, 2019}), so this paper presents the first evidence of the possible relationship between the two.

Family income was also presented as a variable whose impact on PCM is influenced by the sample it comes from. This variable presented opposite behaviors in each country, in the Brazilian sample it had a positive impact while in the Peruvian sample it had a negative impact. When analyzing this point, we consider it important to observe the cities of residence of the participants from both countries. In the case of Brazil, the cities with the highest number of respondents are Rio de Janeiro (34.6 %) and São Paulo (21.5 %), two important cities in Brazil with high cost of living, on the other hand, in the Peruvian sample, if well, 28.9 % of respondents are residents of Lima, considered a city with high cost of living (^{Mercer, 2020}), most of the participants come from provinces such as Cajamarca (31.3 %) and Loreto (15.3 %), where the cost of living is considerably lower than in Lima, Rio de Janeiro and São Paulo.

Given the above, a more positive view of having children in high socioeconomic levels, which characterizes the Brazilian sample, may be related to the high cost that having children entails in these cities. Thus, it becomes understandable that the economic factor is an incentive to consider parenthood in a positive way when one has a favorable socioeconomic level or has a more pessimistic view of parenthood when the economic factor is a problem.

Other aspects that we can consider when reflecting on this difference are the economic characteristics of both samples, where Brazilians present a considerable percentage of people with high socioeconomic levels (26 % of the sample received more than ten minimum wages) while in Peru the percentage of people with this income is minimal (5 %), being mostly composed of people of low socioeconomic level (54 % reported receiving less than two minimum wages), which may also have impacted the results.

Final considerations

This work is the first study comparing parental motivations and their predictive factors in two Latin American countries using structural equation modeling, but its main contribution is to present evidence that the influence of contextual and personal variables on parental motivations is subject to the context in which the subject is inserted. These findings pave the way for reflections on how identifying the impact of these factors can contribute to the development of interventions adjusted to each reality to achieve people's reproductive goals.

As for limitations, we report that the samples were of convenience and not random, as well as predominantly female and not homogeneous in their socio-demographic characteristics. Both limitations may prevent the generalization of the results. Likewise, the lack of studies on the subject in Latin America made it difficult to discuss the data in light of the literature.

New studies are recommended in Peru that consider relationship satisfaction as a control variable to corroborate (or reject) the lack of relationship between the variables having or not having a partner and PCM in this population. Studies with samples with greater income diversity in both countries are also recommended to corroborate whether the impact of the family income variable is opposite in Brazil and Peru. Finally, it is important to note that the data collection for this study was conducted before the pandemic of COVID-19, so future studies could test whether health and social constraints, as well as the economic consequences derived from the pandemic, had any interference on motivations for parenting.