Parental Burnout Assessment (PBA) in Different Hispanic Countries: An Exploratory Structural Equation Modeling Approach

fpsyg-13-827014 April 6, 2022 Time: 10:53 # 1 ORIGINAL RESEARCH published: 07 April 2022 doi: 10.3389/fpsyg.2022.827014 Edited by: João P. Marôco, University Institute of Psychological, Social and Life Sciences (ISPA), Portugal Reviewed by: Wanderson Silva, Federal University of Alfenas, Brazil Katja Upadyaya, University of Helsinki, Finland *Correspondence: Denisse Manrique-Millones [email protected] Specialty section: This article was submitted to Quantitative Psychology and Measurement, a section of the journal Frontiers in Psychology Received: 02 December 2021 Accepted: 11 March 2022 Published: 07 April 2022 Citation: Manrique-Millones D, Vasin GM, Dominguez-Lara S, Millones-Rivalles R, Ricci RT, Abregu Rey M, Escobar MJ, Oyarce D, Pérez-Díaz P, Santelices MP, Pineda-Marín C, Tapia J, Artavia M, Valdés Pacheco M, Miranda MI, Sánchez Rodríguez R, Morgades-Bamba CI, Peña-Sarrionandia A, Salinas-Quiroz F, Silva Cabrera P, Mikolajczak M and Roskam I (2022) Parental Burnout Assessment (PBA) in Different Hispanic Countries: An Exploratory Structural Equation Modeling Approach. Front. Psychol. 13:827014. doi: 10.3389/fpsyg.2022.827014 Parental Burnout Assessment (PBA) in Different Hispanic Countries: An Exploratory Structural Equation Modeling Approach Denisse Manrique-Millones1*, Georgy M. Vasin2, Sergio Dominguez-Lara3, Rosa Millones-Rivalles4, Ricardo T. Ricci5, Milagros Abregu Rey5, María Josefina Escobar6, Daniela Oyarce7, Pablo Pérez-Díaz8, María Pía Santelices9, Claudia Pineda-Marín10, Javier Tapia11, Mariana Artavia11, Maday Valdés Pacheco12, María Isabel Miranda13, Raquel Sánchez Rodríguez14,15, Clara Isabel Morgades-Bamba16, Ainize Peña-Sarrionandia17, Fernando Salinas-Quiroz18, Paola Silva Cabrera19, Moïra Mikolajczak20 and Isabelle Roskam20 1Carrera de Psicología, Facultad de Ciencias de la Salud, Universidad Científica del Sur, Lima, Peru, 2Centre for Multidisciplinary Research in Education, Russian Academy of Education, Moscow, Russia, 3Escuela de Psicología, Instituto de Investigación de Psicología, Universidad de San Martín de Porres, Lima, Peru, 4Facultad de Educación, Universidad Marcelino Champagnat, Lima, Peru, 5Facultad de Medicina, Universidad Nacional de Tucumán, San Miguel de Tucumán, Argentina, 6Center for Social and Cognitive Neuroscience, School of Psychology, Universidad Adolfo Ibáñez, Santiago, Chile, 7Carrera de Psicología, Facultad de Ciencias Sociales y Comunicación, Universidad Santo Tomás, Talca, Chile, 8Instituto de Psicología, Universidad Austral de Chile, Puerto Montt, Chile, 9School of Psychology, Millenium Institute for Research in Depression and Personality, Pontificia Universidad Católica de Chile, Santiago, Chile, 10 School of Psychology, Konrad Lorenz University, Bogotá, Colombia, 11 Facultad de Ciencias Sociales, Instituto de Investigaciones Psicológicas, Universidad de Costa Rica, San José, Costa Rica, 12 Facultad de Psicología, Universidad de la Habana, La Habana, Cuba, 13 Facultad de Psicología, Pontificia Universidad Católica de Ecuador, Quito, Ecuador, 14 Céres Research Unit, TR3, Free Faculty of Arts and Humanities, Catholic Institute of Toulouse, Toulouse, France, 15 Center for Studies and Research in Psychopathology and Health Psychology (CERPPS), Faculty of Psychology, University of Toulouse Jean Jaurès, Toulouse, France, 16 Departamento Metodología de Ciencias del Comportamiento, Universidad Nacional de Educación a Distancia, Madrid, Spain, 17 Departamento “Psicología Clínica y de la Salud y Metodología de Investigación”, Facultad de Psicología, University of the Basque Country, Donostia-San Sebastián, Spain, 18 “Abby & Anna” SOGIE Lab, Eliot-Pearson Department of Child Study and Human Development, Tufts University, Medford, MA, United States, 19 Departamento de Psicología, Instituto de Psicología, Educación y Desarrollo Humano, Universidad de la República, Montevideo, Uruguay, 20 Department of Psychology, UCLouvain, Louvain-la-Neuve, Belgium Parental burnout is a unique and context-specific syndrome resulting from a chronic imbalance of risks over resources in the parenting domain. The current research aims to evaluate the psychometric properties of the Spanish version of the Parental Burnout Assessment (PBA) across Spanish-speaking countries with two consecutive studies. In Study 1, we analyzed the data through a bifactor model within an Exploratory Structural Equation Modeling (ESEM) on the pooled sample of participants (N= 1,979) obtaining good fit indices. We then attained measurement invariance across both gender and countries in a set of nested models with gradually increasing parameter constraints. Latent means comparisons across countries showed that among the participants’ countries, Chile had the highest parental burnout score, likewise, comparisons across gender evidenced that mothers displayed higher scores than fathers, as shown in previous studies. Reliability coefficients were high. In Study 2 (N= 1,171), we tested Frontiers in Psychology | www.frontiersin.org 1April 2022 | Volume 13 | Article 827014 fpsyg-13-827014 April 6, 2022 Time: 10:53 # 2 Manrique-Millones et al. Psychometric Properties of Spanish PBA the relations between parental burnout and three specific consequences, i.e., escape and suicidal ideations, parental neglect, and parental violence toward one’s children. The medium to large associations found provided support for the PBA’s predictive validity. Overall, we concluded that the Spanish version of the PBA has good psychometric properties. The results support its relevance for the assessment of parental burnout among Spanish-speaking parents, offering new opportunities for cross-cultural research in the parenting domain. Keywords: equivalence, Spanish-speaking, exhaustion, parenting, ESEM INTRODUCTION Parenthood is associated with positive emotions like joy, accomplishment, and pride in many cultures. However, evidence shows that parenting is also a complex and strained activity (e.g., Abidin, 1990;Crnic and Low, 2002;Deater-deckard, 2014). Exposure to too many stressors over long periods can lead to burnout, as has been demonstrated, when there are not enough psychological resources to deal with them in job settings (Bakker and Demerouti, 2007). The same is true in the parenting context: when the balance between parental risks and resources chronically leans to the negative side, parents are at high risk of parental burnout (Roskam et al., 2017; Mikolajczak and Roskam, 2018). Parental Burnout Burnout is a disorder that is associated with chronic stress. However, the problem is not limited to factors increasing stress but to lack of resources to deal with it. This is the sustained imbalance between stress and resources that characterizes the burnout syndrome. According to this view, parental burnout is considered a singular and contextualized syndrome resulting from a chronical imbalance of risks (e.g., poor childrearing, numerous parental duties, and perfectionism) over resources (e.g., positive coparenting, external support, high emotional intelligence, etc.) in the field of parenting (Mikolajczak et al., 2018, 2019). The demand or risk factors would intensify parental burnout. One example of such a stressor is parental perfectionism, which has been proven to be detrimental to parental burnout. A study conducted by Sorkkila and Aunola (2020) involving more than 1,500 Finnish parents determined that socially prescribed perfectionism, demanded by significant others in the individual environment through imposition of very highperformance expectations and the tendency to harshly judge the results (Flett and Hewitt, 2002), constitutes a risk factor. Similar results were obtained by Kawamoto et al. (2018) in a study settled in Japan with 1,200 participants, in which parental perfectionism was a strong stressor contributing to parental burnout. Resources, on the other hand, would lower parental burnout. For example, emotional competences, as the ability to process information about one’s own emotions and those of others, enabling people to pay attention, use, understand, and manage their emotions (Mayer and Salovey, 1997) plays a protective role for parental burnout. The study of Lin et al. (2021) with two independent samples (i.e., Poland and Belgium) shed light on the key role played by emotional competences buffering the effect of parental perfectionism on parental burnout. Mikolajczak and Roskam (2018) proposed a Balance between Risk and Resources (BR2) theory in which “parental burnout develops when parental resources are insufficient to meet the demands (whatever they are)” (p. 2). While resources and risks play opposite roles with regard to parental burnout, the absence of one of them does not equate to the presence of the other. For example, the absence of denigration by the partner does not mean that the parent has the support of the partner. And a risk does not necessarily have to be offset by the corresponding resource but can be offset by one or more other resources. For example, the lack of partner support may be compensated for by social support or by the children’s autonomy. Parental burnout is thus the result of an imbalance between several factors and of risk not compensated by sufficient resources to cope with it. In a twowave longitudinal study conducted over 900 parents, results highlighted that the BR2framework predicted parental burnout (Mikolajczak and Roskam, 2018). Parental exhaustion is the core symptom of parental burnout. Parents experience tiredness at every aspect of their daily routine of attending to their children. The emotional drain is such that they often see themselves reaching their limits. Another symptom is the emotional distancing from their children. Parents also progressively reduce their involvement in parenting, while their relationship with their children suffers; interactions decrease and are limited mostly to functional/instrumental aspects at the detriment of emotions. Another symptom is the loss of feeling of accomplishment as a parent: being fed up with parenting, parents can no longer stand their role, and time with their children is no longer appreciated. All these symptoms and states represent a contrasting view of the parental self, i.e., previous feelings toward parenting (Roskam et al., 2018). Parental burnout should not be regarded as typical parental stress (Roskam et al., 2017;Kawamoto et al., 2018;LebertCharron et al., 2018;Van Bakel et al., 2018;Brianda et al., 2020). It results from tremendous parental stress, as a long-lasting response (Mikolajczak and Roskam, 2018). In a cross-cultural study gathering 42 countries across the world, the prevalence of parental burnout varied from 0 to more than 8% (Roskam et al., 2021), even after controlling for sample size and sex imbalance. Poland (7%), Belgium (7.9%), and the United States (8.4%) were the countries with the highest levels of parental burnout. It is Frontiers in Psychology | www.frontiersin.org 2April 2022 | Volume 13 | Article 827014 fpsyg-13-827014 April 6, 2022 Time: 10:53 # 3 Manrique-Millones et al. Psychometric Properties of Spanish PBA also not comparable to other forms of burnout, such as job burnout, given that these two constructs have low to moderate correlations (Roskam et al., 2017;Kawamoto et al., 2018;Van Bakel et al., 2018). Finally, although parental burnout can lead to depression, it is not the same as depression (Kawamoto et al., 2018;Van Bakel et al., 2018;Sánchez-Rodríguez et al., 2019). Parental burnout is a particular syndrome, which is contextspecific (Mikolajczak et al., 2020). The burnout syndrome matters because is associated with numerous negative outcomes, e.g., psychopathology symptoms (Mesman and Koot, 2001). In the parenthood field, the burnout syndrome has far-reaching consequences. Recent studies suggest that it has harmful effects for the parent, in particular suicidal thoughts. Suicidal ideations are associated with profound feelings of hopelessness. Individuals do not believe that their problems could be solved, and consequently, there is no way out from suffering. Because the parental role can never be left behind and parents can never resign, medium to large associations between parental burnout, and escape and suicidal ideations have been found (Mikolajczak et al., 2018). Parental burnout also has deleterious consequences on children (Mikolajczak et al., 2018, 2019). Medium to large associations between parental neglect and parental violence toward children and parental burnout were reported (Mikolajczak et al., 2018). Burned-out parents are so exhausted that they no longer take care of their children, despite understanding their needs. With emotional distance and loss of accomplishment in parenting, verbal or physical violence toward children also increases (Blanchard et al., 2020;Hansotte et al., 2021). The deleterious consequences of burnout highlight the need for reliable measures in each specific context, either job or parenting, as appropriate. Assessment of Parental Burnout Pelsma et al. (1989) introduced parental burnout measurement by advocating that the Maslach Burnout Inventory (MBI, Maslach et al., 1986) might be a good foundation for constructing a measure of parental burnout. His intuition was partially supported, and he did not pursue the investigation. It took around 20 years until his work gained recognition. In 2007, Norberg, who was then working at a children’s hospital, noticed that parents of severely ill children presented burnout symptoms. Norberg’s team then explored this issue over the next 7 years using the Shirom–Melamed Burnout Questionnaire (SMBQ, Melamed et al., 1992). The SMBQ is a 14-item questionnaire consisting of ten context-free items and four job-related items. Although preliminary support of the notion of parental burnout could be found in these studies, skeptics could argue that having an ill child made parents at risk to develop job burnout. Roskam et al. (2017) were the first authors to validate a specific instrument for assessing parental burnout. They were the first to adapt Maslach Burnout Inventory© (MBI, Maslach et al., 1986) so that all chosen items referred specifically to the parenting domain. The “depersonalization” subscale did not work in the parental context and the authors replaced it with an emotional distancing subscale. Their development and validation studies resulted in the Parental Burnout Inventory (PBI, Roskam et al., 2017), a 22-item measure of parental burnout encompassing three factors: parental exhaustion, emotional distancing from the children, and reduced parental effectiveness, alongside lack of fulfillment. The approach used to develop the Parental Burnout Inventory was deductive since it was based on the MBI. This resulted in unanswered questions about the three dimensions they proposed to depict parental burnout. Roskam et al. (2018) used the inductive method to deepen the conceptualization and measurement of parental exhaustion. For this purpose, colleagues who were oblivious of the aforementioned three-dimensional structure analyzed testimonies from a large sample of Anglo-Saxon and francophone exhausted parents, based on the Interpretative Phenomenological Analysis (IPA) (Hubert and Aujoulat, 2018). From the IPA, 50 items transpired which were reduced to 23 items after factor analysis, thus resulting in the Parental Burnout Assessment (PBA). This questionnaire covers four dimensions, two of which are identical to those of the PBI, i.e., Emotional Exhaustion in the parental role and Emotional Distancing from one’s children, and the other two distinct, i.e., Feelings of Being Fed Up, and Contrast in parental self. Because of its theoretical background, psychometric properties, and open access, the PBA is considered a quality instrument for assessing Parental Burnout. Measurement Equivalence Across Countries One of the main purposes of cross-cultural research is to compare psychological constructs between distinct groups. A measure that has been deemed psychometrically sound in one cultural setting is often assumed to be translatable for use in another cultural context, but this assumption may be misleading and needs to be formally tested. Currently, the only evidence for the cross-cultural equivalence of the PBA construct has been found between Frenchand English-speaking samples in Western countries (Roskam et al., 2018). This does not exclude the possibility that parental burnout may vary, both structurally and expressively, across countries or cultures. An important challenge in the large-scale assessments commonly used in cross-cultural research is to verify that the collected self-reported data measure the object of interest in the same way, irrespective of the macrocontext. Due to differences in language and cultural response biases, participants interpret certain words in different ways. As a result, their understanding of the intended meaning of a question changes, and respondents may attribute different meanings to an instrument, undermining the comparability of constructs across cultures (Markus and Kitayama, 1991;Rutkowski and Svetina, 2014). The Current Research The current research aims to evaluate the psychometric properties of the Spanish version of the PBA. In order to test the validity of its internal structure, we first conducted an exploratory analysis using an Exploratory Structural Equation Modeling (ESEM) framework (Marsh et al., 2010). ESEM is an interesting alternative approach to classical Confirmatory Factorial Analysis (CFA) commonly used in previous studies (e.g., Aunola et al., 2020;Sodi et al., 2020;Szczygieł et al., 2020; Frontiers in Psychology | www.frontiersin.org 3April 2022 | Volume 13 | Article 827014 fpsyg-13-827014 April 6, 2022 Time: 10:53 # 4 Manrique-Millones et al. Psychometric Properties of Spanish PBA Roskam et al., 2021). In CFA, the influence exerted on a given item by dimensions other than the one(s) to which this item belongs can be hidden (Marsh et al., 2010). In previous PBA validation studies, the presence of high inter factorial correlations (>0.90) was not explicitly addressed, which might represent a threat to multicollinearity (Brown, 2015). When overlaps occur, interpreting factors as independent entities may be no longer appropriate. Another reason for implementing ESEM is that, from a conceptual point of view, the good fit indices found for a culturally-invariant model in different countries do not guarantee that this model is the best representation of parental burnout in each country. For these reasons, we examined whether a bifactor model of the PBA could be applied to the sample of Spanish-speaking parents. Bifactor modeling assumes that the correlations between factors are explained by the presence of a general factor (Reise, 2012). The analysis simultaneously models the influence of specific (orthogonal) factors and a general factor on a set of items. This technique provides even more solid evidence regarding the relevance of a general factor (Parental Burnout in the current study) compared to a second-order hierarchical model. Hence, bifactor modeling in ESEM tests the direct influence of the general factor on the items, whereas second-order hierarchical modeling tests the influence of the general factor on the items indirectly through the first-order factors (Canivez, 2016). Figure 1 depicts the conceptual model of the general measure of parental burnout. We then tested measurement invariance across genders and countries (Argentina, Chile, Costa Rica, Cuba, Ecuador, Peru, Spain, and Uruguay), and we examined mean differences. Based on previous research (Pérez-Díaz and Oyarce Cádiz, 2020; Roskam and Mikolajczak, 2020), parental burnout was expected to be higher among mothers than fathers. In an exploratory way, we tested the mean differences across countries regarding the total score of parental burnout. Finally, to obtain evidence of validity for the relationship of the PBA with other theoretically related variables, predictive validity was estimated by testing associations between parental burnout and three specific effects of this construct, i.e., escape and suicidal ideations, parental neglect, and parental violence toward one’s children, in two countries displaying different mean levels of parental burnout, i.e., Chile and Ecuador. GENERAL METHOD Overview The two studies were part of the International Investigation of Parental Burnout (IIPB), a 42-country study of parental burnout around the globe (Roskam et al., 2021). An institutional review board provided ethical clearance for the research in each country [i.e., Spain (Bioethics Committee from The National Distance Education University); Perú (The Research Committee of the San Martín de Porres University); Chile (Ethics Committee from the Autonomous University of Chile, 71–18); Uruguay (Ethics Committee of the Faculty of Psychology of the University of the Republic); Ecuador (Institutional Review Commission of the Catholic University of Ecuador); Cuba (Research Ethics Committee of the Faculty of Psychology of the University of Havana, Dossier N-3); and Costa Rica (Scientific Ethics Committee of the University of Costa Rica, N93 VI-10712018)]. Note that in Argentina, no ethical clearance was required for the research. Both studies complied with APA ethical standards regarding the treatment of human subjects. The inclusion criterium for the IIPB study was having at least one child still living at home. In the current study, we introduced two additional criteria for the inclusion of the participants. Parents were required to have been born and resided in the same country where the data collection took place. Parents signed an informed consent allowing withdrawal at any moment. We also assured participants that the data would remain anonymous. In Study 1, a common survey, made up of sociodemographic characteristics and the measure of parental burnout (i.e., PBA), was used in the eight included countries. The Spanish version of the PBA was produced using translation/backtranslation procedures by the Hispanic members of the IIPB consortium. Minor country-specific language modifications were implemented. Participants completed the survey online (in Argentina, Chile, and Ecuador), on paper (in Cuba and Uruguay), or both (in Costa Rica, Peru, and Spain). The online version was administered using the Qualtrics platform. Information about the study was disseminated through social media (Facebook, LinkedIn, and Instagram), by word of mouth or through an official email invitation sent by the researchers. For the paperand-pencil version, participants were recruited through parents’ associations, as well as private and public schools. The data were mostly collected in urban zones of the capital in each country. Study 2 was based on a second data collection in Chile and Ecuador. The survey comprised measures of parental burnout, escape, suicidal ideations, parental neglect, and parental violence toward one’s children. All participants completed the survey online using the Qualtrics platform. We implemented a snowball sampling technique. As in Study 1, the data were primarily collected in urban zones of the capital in each country. In the section “General Method” as well as in the “Method” sections of Study 1 and Study 2, we reported how we made sure that the statistical analyses were appropriate for the sample sizes in each participating country, all data inclusion criteria (all being established prior to data analysis), the measures used in the two studies, and all the statistical analyses we performed. We reported exact pvalues, effect sizes, and 95% confidence intervals for inferential analyses. Data Analyses In Study 1, we used a bifactor ESEM model to evaluate how adjusted the parental burnout measurement model is. The assumptions of the bifactor model are that the specific factors are orthogonal to each other and that the presence of a general factor explains better the correlations between them. We preliminarily tested the bifactor model structure of the PBA in each country sample independently. Following that, we performed ESEM to inspect the bifactor model on the pooled sample. Frontiers in Psychology | www.frontiersin.org 4April 2022 | Volume 13 | Article 827014 fpsyg-13-827014 April 6, 2022 Time: 10:53 # 5 Manrique-Millones et al. Psychometric Properties of Spanish PBA FIGURE 1 | The conceptual bifactor model of parental burnout. Several items showed deviation from normality, as their skewness and kurtosis proved. This is expected, as mental health indicators usually feature an asymmetric distribution. A positively skewed distribution for these parental burnout items is therefore conceptually sound. We used maximum likelihood estimation with robust (Huber-White) standard errors and a scaled test statistic (MLR) as the estimation method while relying on standard fit indices to ascertain the adequacy of the Bifactor ESEM model: the SatorraBentler scaled chi-square statistics (S-Bχ2; Satorra and Bentler, 1991), the Tucker-Lewis index (TLI), the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). Should the CFI and TLI values reach 0.90 or more, they are considered adequate, and values of RMSEA and SRMR ≤0.8 are preferred (Hu and Bentler, 1999). These Frontiers in Psychology | www.frontiersin.org 5April 2022 | Volume 13 | Article 827014 fpsyg-13-827014 April 6, 2022 Time: 10:53 # 6 Manrique-Millones et al. Psychometric Properties of Spanish PBA analyses were executed with the software Mplus version 7 (Muthén and Muthén, 1998-2015). The fit indices do not report on the strength of the general factor. Thus, complementary indicators were implemented to evaluate the strength of the general factor, i.e., the general hierarchical omega (ωh;Zinbarg et al., 2006), and the explained common variance (ECV; Sijtsma, 2009). Values greater than 0.70 and 0.60, were respectively expected (Reise et al., 2013), while values greater than 0.30 for the hierarchical omega per dimension (ωhs) meant the interpretability of the factors (Smits et al., 2015). When testing for measurement invariance, we used a set of models where constraints were incrementally introduced through a stepwise multiple group analysis (MG-ESEM). First, the parental burnout model was assessed for configural invariance. Second, item factor loadings were tested in a metric invariance model. Third, scalar invariance was verified with the intercepts set as equal across groups. Finally, error variances were constrained to be equal across groups. A CFI decrease within the established limits (1CFI ≤-0.01), RMSEA and SRMR indices within certain thresholds (1RMSEA ≤0.015; 1SRMR ≤0.030) confirms measurement invariance (Cheung and Rensvold, 2002; Chen, 2007). Note that measurement invariance across genders was estimated on the pooled sample, rather than in each of the countries because of the small sample size of fathers, and to a lesser extent, of mothers in some countries. To evaluate gender and country-related PBA mean differences, the latent mean of the reference groups (Chile for countries and fathers for gender) was set to zero within the framework of the scalar invariance model. In this sense, the statistical significance of the mean difference of each sample compared to the reference group was assessed in each analysis, either across countries or genders. Therefore, a latent mean difference (LMdiff) greater than zero indicated that the comparison group had a higher mean than that of the reference group (Brown, 2015). To calculate whether the latent means of the comparison group differed significantly from the baseline, we calculated critical ratios (CRs) by dividing the estimate by its standard error. When the CR was above 1.96, the difference was statistically significant. To calculate whether the latent mean of the comparison group differed significantly from the baseline, we calculated critical ratios (CRs) by dividing the estimate by its standard error. When the CR was above 1.96, the difference was statistically significant (Byrne, 2013). We supplemented the comparisons with Cohen’s d measure of effect size with confidence intervals, where values less than 0.20 indicated an insignificant effect size, between 0.20 and 0.50 a small magnitude, between 0.50 and 0.80, moderate; and greater than 0.80, large (Ferguson, 2009). In Study 2, the predictive validity of the PBA was estimated using bivariate correlations between the general factor of the PBA on the one hand, and escape and suicidal ideations, parental neglect, and parental violence toward one’s children on the other hand. Confidence intervals for correlation coefficients were calculated. We estimated the reliability of the scales used in the current research by computing the Cronbach’s alpha (α) and McDonald’s omega coefficient (ω). A hierarchical omega (ωh) was also calculated within the framework of the bifactor model (Rodriguez et al., 2016). While Cronbach’s alpha is based on the correlations between test items, model-based coefficients like omega are rather based on the latent structure of the test. Cronbach’s alpha can be overestimated because all the shared variance between items counts toward the coefficient. In contrast, omega bypasses this caveat, due to estimating reliability on the factorial loads, thus for the construct in the analysis. STUDY 1 Participants The initial sample (N= 2,502) encompassed participants from Argentina, Chile, Costa Rica, Cuba, Ecuador, Peru, Spain, and Uruguay. Of these, 1,869 (74.7%) were women and 633 (25.3%) were men. Their age ranged from 17 to 78 (M= 38.03, SD = 8.49). However, since one of the aims of the study is a multigroup analysis and due to the absence of convergence, three samples were excluded (Argentina, Cuba, and Uruguay after initial processing). Similarly, Ecuador’s data from Study 1 were replaced by Study 2 due to the larger sample size. With this final sample (N= 1,979) the majority of participants were women (1,588, 80.2%). Their age ranged from 17 to 78 years (M= 38.43, SD = 8.09). The number of children ranged from 1 to 9 (M= 1.79, SD = 0.88). Table 1 displays demographic information for participants in each country. Measures Parental Burnout Parental burnout was assessed by the Parental Burnout Assessment (PBA, Roskam et al., 2018), a 23-item self-report. The instrument encompasses four dimensions: Emotional Exhaustion [nine items; e.g., “I have zero energy for looking after my child(ren)”], Contrast (six items; e.g., “I’m ashamed of the parent that I’ve become”), Feelings of Being Fed Up (five items; e.g., “I feel like I can’t cope as a parent”), and Emotional Distancing [three items; e.g., “I’m no longer able to show my child(ren) how much I love them”]. Items are rated on 7-point Likert scales: never (0), a few times a year or less (1), once a month or less (2), a few times a month (3), once a week (4), a few times a week (5), every day (6). Cronbach’s alphas reported in the original PBA version were high, ranging from 0.77 to 0.94. Spanish PBA scale can be found in Supplementary Material. Results Parental Burnout Assessment Item Description and Factor Analysis The mean of the items was Mmean = 1.09 (range: 0.38–2.64) and the standard deviation was SDmean = 1.44 (range: 1.04–1.99). Several items were positively skewed [MSk = 1.84, range (0.22– 3.42)] and showed high kurtosis [Mk= 3.55, range (−1.27 to 12)]. Methods of estimation which do not require normality were applied, as mentioned in the “Data Analyses” section. Due to the complexity of the ESEM approach and the number of parameters to be estimated, the model could not converge in three samples, so they were removed from the analyses. Frontiers in Psychology | www.frontiersin.org 6April 2022 | Volume 13 | Article 827014 fpsyg-13-827014 April 6, 2022 Time: 10:53 # 7 Manrique-Millones et al. Psychometric Properties of Spanish PBA TABLE 1 | Demographic characteristics of the samples in Study 1. N nWomen nMen Mean age (SD) Mean ageWomen (SD) Mean ageMen (SD) Educational level (years of education) Number of children 12 years (compulsory education) 17 years (university degree) >18 years (postgraduate degree) Mean [Range] Argentina 176 118 58 39.95 (9.88) 40.43 (9.23) 38.97 (11.04) 32 (18.1%) 79 (44.6%) 65 (37.3%) 2.35 [1–8] Chile* 718 641 77 36.89 (6.01) 36.49 (5.70) 40.25 (7.38) 25 (5.2%) 273 (41.9%) 420 (52.9%) 1.79 [1–9] Costa Rica* 234 14 94 37.68 (8.15) 35.66 (6.98) 40.69 (8.85) 29 (16.6%) 102 (41.5%) 103 (41.9%) 1.59 [1–5] Cuba 240 137 103 40.04 (10.24) 39.96 (10.32) 40.16 (10.19) 143(59.6%) 71 (29.6) 26 (10.8%) 1.70 [1–4] Ecuador** 146 102 44 32.45 (7.51) 32.27 (7.34) 32.86 (7.96) 9 (6.2%) 69 (47.3%) 68 (46.6%) 1.64 [1–5] Peru* 368 265 103 40.22 (10.84) 38.97 (9.89) 43.815 (12.34) 99 (30.6%) 144 (57.8%) 125 (11.6%) 1.93 [1–5] Spain* 333 287 46 41.10 (7.86) 40.80 (7.80) 42.98 (8.09) 104(32.4%) 133 (43.1%) 96 (24.5%) 1.74 [1–6] Uruguay 287 179 108 35.02 (6.42) 34.23 (5.80) 36.36 (7.16) 150(50.7%) 90 (33.1%) 47 (16.2%) 1.60 [1–6] *Denotes data sets involved in study 1 and **For the case of Ecuador we used dataset of study 2. TABLE 2 | Bifactor ESEM model and reliability per country and gender. GF EX CO SA ED S-Bχ2[148] CFI TLI RMSEA SRMR ECV ωhωhs ωhs ωhs ωhs ω α [90% IC] [95% IC] Countries Chile 410.519** 0.964 0.939 0.050 0.017 0.823 0.930 0.103 0.181 0.061 0.176 0.900 0.961 [0.044–0.055] [0.0957–0.964] Ecuador 317.956** 0.916 0.857 0.068 0.025 0.837 0.944 0.016 0.203 0.040 0.147 0.894 0.957 [0.060–0.077] [0.952–0.963] Costa Rica 257.012** 0.965 0.940 0.056 0.020 0.821 0.926 0.207 0.040 0.107 0.087 0.905 0.963 [0.044–0.067] [0.956–0.968] Perú 302.643** 0.927 0.874 0.053 0.025 0.803 0.913 0.046 0.269 0.065 0.161 0.869 0.942 [0.045–0.062] [0.934–0.949] Spain 256.985** 0.967 0.944 0.047 0.019 0.811 0.932 0.012 0.320 0.060 0.128 0.895 0.957 [0.037–0.057] [0.952–0.963] Total Sample 542.044** 0.976 0.958 0.037 0.014 0.829 0.929 0.070 0.231 0.065 0.122 0.958 0.958 [0.033–0.040] [0.955–0.960] Gender Fathers 249.703** 0.949 0.913 0.042 0.025 0.750 0.882 0.109 0.332 −0.147 0.109 0.862 0.929 [0.033–0.051] [0.920–0.937] Mothers 509.523** 0.974 0.956 0.039 0.015 0.830 0.931 0.065 0.229 0.055 0.134 0.894 0.959 [0.036–0.043] [0.956–0.961] S-B χ2, Satorra Bentler chi-squared; CFI, Comparative Fit Index; TLI, Tucker–Lewis index; RMSEA, Root Mean Square Error of Approximation; SRMR, Standardized Root Mean Square Residual; ECV, Explained Common Variance; GF, General Factor; EX, Exhaustion; CO, Contrast; SA, Saturation; DE, Emotional Distancing; ωh, General Omega hierarchical; ωhs, Omega hierarchical per dimension; α, Cronbach’s alpha. **p <0.01. The bifactor ESEM model was estimated in the five remaining samples and the pooled sample (N= 1,979). Fit indices were good, as displayed in Table 2. Moreover, the magnitude of the descriptive indicators associated with the bifactor analysis (ECV, ωh, and ωhs) demonstrated the strength of the general factor across countries and genders (see Table 3). In addition, as shown in Table 3, the standardized factor loadings ranged from 0.57 to 0.81 for the general factor of Parental Burnout in the pooled sample (N= 1,979). Frontiers in Psychology | www.frontiersin.org 7April 2022 | Volume 13 | Article 827014 fpsyg-13-827014 April 6, 2022 Time: 10:53 # 8 Manrique-Millones et al. Psychometric Properties of Spanish PBA TABLE 3 | Factor loadings of the bifactor ESEM model in the pooled sample. GF EX CO SA ED EX1 0.574 0.453 −0.053 −0.160 −0.134 EX2 0.736 0.435 −0.007 −0.025 −0.143 EX3 0.696 0.381 0.025 −0.003 0.008 EX4 0.698 0.235 −0.038 −0.057 0.075 EX5 0.767 0.097 0.031 0.347 −0.083 EX6 0.772 0.235 −0.074 −0.062 −0.010 EX7 0.780 0.096 −0.105 −0.036 0.010 EX8 0.802 0.012 −0.138 −0.122 0.030 EX9 0.813 −0.047 0.005 −0.035 0.122 CO1 0.616 0.283 0.244 0.067 0.027 CO2 0.652 0.234 0.400 0.090 0.229 CO3 0.700 0.075 0.399 0.067 0.167 CO4 0.713 −0.214 0.492 0.040 −0.058 CO5 0.663 −0.194 0.466 −0.037 −0.127 CO6 0.726 −0.133 0.374 0.001 0.153 SA1 0.712 0.037 0.080 0.520 −0.017 SA2 0.743 0.344 −0.110 -0.082 −0.102 SA3 0.619 −0.079 0.035 0.125 −0.009 SA4 0.810 0.172 −0.098 0.022 −0.073 SA5 0.732 −0.152 0.066 0.437 −0.071 ED1 0.606 −0.104 0.084 −0.006 0.249 ED2 0.581 −0.219 0.284 0.022 0.241 ED3 0.665 −0.109 0.129 −0.109 0.347 Factor loadings in bold belong to the theoretical factor. GF, General Factor; EX, Exhaustion; CO, Contrast; SA, Saturation; ED, Emotional Distancing. TABLE 4 | Goodness-of-fit indices of measurement invariance of the PBA bifactor ESEM model across countries and gender. S-Bχ2df RMSEA 1RMSEA CFI 1CFI SRMR 1SRMR Across Countries Configural 1.602.712 740 0.054 0.953 0.021 Metric 1.767.875 1100 0.039 −0.014 0.963 −0.01 0.040 0.019 Scalar 1.980.609 1172 0.042 0.003 0.956 −0.007 0.042 0.002 Error 2.168.826 1264 0.043 0.001 0.95 −0.006 0.053 0.011 Across Gender Configural 769.355 296 0.040 0.969 0.017 Metric 846.208 386 0.035 −0.005 0.970 0.001 0.028 0.011 Scalar 890.556 404 0.035 0.000 0.968 −0.002 0.029 0.001 Error NC NC NC NC NC NC NC NC RMSEA, Root Mean Square Error of Approximation; CFI, Comparative Fit Index; SRMR, Standardized Root Mean Square Residual; S-Bχ2, Satorra Bentler chi-squared; df, Degrees of freedom; 1RMSEA, Root Mean Square Error of Approximation Difference Test;1CFI, Comparative Fit Index Difference Test; 1SRMR, Standardized Root Mean Square Residual Difference Test; NC, No Convergence. Measurement Invariance of Parental Burnout Bifactor Exploratory Structural Equation Modeling Model Across Countries We used the Chilean sample as the reference group. As shown in Table 4, the fit indices support configural invariance. Under TABLE 5 | Comparisons across reference and comparison countries. LMdiff SE CR d (95% CI) Costa Rica −0.191 0.090 −2.122* 0.28 [0.13–0.43] Ecuador −0.315 0.097 −3.247* 0.32 [0.19–0.45] Peru −0.506 0.102 −4.961* 0.45 [0.32–0.57] Spain −0.142 0.077 −1.844 0.18 [0.05–0.31] LM, Latent means; LMdiff , difference relative to the mean of this referent group; * denotes Critical Ratio value larger than 1.96 indicating statistically significant differences in the LM; d, Cohen’s d; CI, confidence interval. the constraint of equal factor loadings, the model fit did not significantly worsen, providing evidence for metric invariance. In the next step, when intercepts were constrained, scalar invariance was evidenced. Finally, error invariance was attained. Measurement Invariance of Parental Burnout Bifactor Exploratory Structural Equation Modeling Model Across Genders We used fathers as the reference group. According to the variation observed in the CFI, RMSEA, and SRMR fit indices, an acceptable degree of factorial invariance was found up to the scalar level (see Table 4). Mean Differences Across Countries Latent means, standard errors, effect sizes, and latent means comparisons across countries are presented in Table 5. Significant differences in the level of Parental Burnout were found between the five Spanish-speaking countries. In particular, Chile had the highest parental burnout score, whereas Peru had the lowest. Mean Difference Across Genders Parental burnout general latent mean for fathers was set to be 0, and as expected, we found gender-related differences in parental burnout. Mothers displayed higher score than fathers on the Parental Burnout general factor (LMdiff = 0.545, SE = 0.046), with a significant Critical Ratio (CR = 11.85) but with a small effect size (d= 0.32). Reliability Indexes The coefficients found for the scales used in the current study were high (see Table 2). STUDY 2 Participants The total sample (N= 1,171) encompassed participants from Chile (N= 845) and Ecuador (N= 326). Of these, 1,044 (89.2%) were women and 125 (10.8%) were men. Their age ranged from 19 to 66 (M= 37.99, SD = 7.08). The number of children living at home ranged from 1 to 8 (M= 1.81, SD = 0.87); the mean age of the oldest child was 9.51 (SD = 7.40), and the mean age of the youngest child was 6.03 (SD = 6.30). Table 6 shows the demographic information for participants in each country. Frontiers in Psychology | www.frontiersin.org 8April 2022 | Volume 13 | Article 827014 fpsyg-13-827014 April 6, 2022 Time: 10:53 # 9 Manrique-Millones et al. Psychometric Properties of Spanish PBA TABLE 6 | Demographic characteristics of the samples in study 2. N nWomen nMen Mean age (SD) Mean ageWomen (SD) Mean ageMen (SD) Educational level (years of education) Number of children 12 years (compulsory education) 17 years (university degree) >18 years (postgraduate degree) Mean [Range] Chile 845 789 54 38.08 (6.70) 37.94 (6.62) 40.04 (7.60) 29 (3.4%) 321 (38.08%) 493 (58.52%) 1.78 [1–5] Ecuador 326 255 71 37.79 (7.99) 37.09 (7.46) 40.32 (9.27) 41 (11.9%) 128 (40%) 157 (48.1%) 1.87 [1–8] TABLE 7 | Means, standard deviations, and Pearson correlations with confidence intervals between PBA general factor and measures of escape and suicidal ideations, parental neglect, and parental violence toward one’s children. M SD Range 1 2 3 4 1. Escape, suicidal ideations 9.79 4.48 7–40 −0.33** 0.34** 0.56** [0.28, 0.38] [0.29, 0.39] [0.52, 0.60] 2. Parental neglect 28.54 12.74 17–136 −0.59** 0.45** [0.55, 0.63] [0.40, 0.49] 3. Parental violence 21.86 8.64 15–120 −0.33** [0.28, 0.38] 4. PBA general factor 28.67 25.60 0–133 − M, Means; SD, Standard Deviations. Values in brackets indicate 95% confidence interval for each correlation, **p <0.001. Measures Parental Burnout Parental burnout was measured using the Spanish version of the PBA, as previously presented and described in Study 1. Escape and Suicidal Ideations Escape and suicidal ideations were evaluated by a seven-item questionnaire (e.g., “I have thoughts about leaving my family”) rated on a 7-point scale of frequency, i.e., never (1), less than once a month (2), about once a month (3), two or three times a month (4), once a week (5), several times a week (6), every day (7) (Mikolajczak et al., 2018). We obtained the escape and suicidal ideations score by adding the seven item scores. In the current study, the Cronbach’s alpha was 0.83 [0.81, 0.84], and Omega’s coefficient (ω) was 0.86 [0.80, 0.90]. Parental Neglect Neglectful behavior toward one’s children was evaluated with the Parental Neglect Scale (Mikolajczak et al., 2018), a 17-item questionnaire (e.g., “I don’t care about my child’s schooling and future”) rated on an 8-point scale of frequency, i.e., never (1), less than once a month (2), about once a month (3), a few times a month (4), once a week (5), several times a week (6), every day (7), and several times a day (8). The parental neglect score was obtained by adding the 17 item scores. In the current study, Cronbach’s alpha was 0.88 [0.87, 0.89], and Omega’s coefficient (ω) was 0.88 [0.86, 0.91]. Parental Violence Parental Violence was evaluated with the Parental Violence Scale (Mikolajczak et al., 2018), a fifteen-item questionnaire targeting verbal violence [e.g., “I say things to my children that I then regret (insults, ridiculous nicknames, etc.)”], physical violence (e.g., “I sometimes spank or slap my child”), and psychological violence (e.g., “I tell my children that I will abandon them if they are not good”). Items are rated on an 8-point scale of frequency, i.e., never (1), less than once a month (2), about once a month (3), a few times a month (4), once a week (5), several times a week (6), every day (7), and several times a day (8). The parental violence score was obtained by adding the 15 item scores. In the current study, Cronbach’s alpha was 0.86 [0.85, 0.88], and Omega’s coefficient (ω) was 0.86 [0.80, 0.90]. Results Table 7 shows the correlations between the total score of the PBA on the one hand, and escape and suicidal ideations, parental neglect, and parental violence toward one’s children on the other hand. We found evidence for the predictive validity of the PBA – indeed, we perfectly replicated the medium to large associations previously reported with both French and English-speaking samples of parents (Mikolajczak et al., 2018). In particular, we found large correlations between parental burnout and escape and suicidal ideations. We also found medium correlations between parental burnout and both parental neglect and parental violence. GENERAL DISCUSSION This research aimed to evaluate the psychometric properties of the Spanish version of the PBA. We collected data in eight Spanish-speaking countries (i.e., Argentina, Chile, Costa Rica, Cuba, Ecuador, Peru, Spain, and Uruguay). In Study 1, we analyzed the internal structure of the PBA using the ESEM approach considered as the combination of the best components of CFA and conventional EFA. Previous research Frontiers in Psychology | www.frontiersin.org 9April 2022 | Volume 13 | Article 827014