The participants representing the target population in stages 1, 2, and 3 were students from the 3rd to the 7th grade of elementary school. Inclusion criteria required participants to be between 8 and 13 years old. Exclusion criteria included having repeated a grade more than twice (to avoid cognitive or performance deficits), being illiterate, or not completing at least 90% of the administered instruments. A convenience sampling strategy was used, with the consent of six public schools and one private school in the metropolitan area of Porto Alegre, in the state of Rio Grande do Sul (Brazil).

Stage 1 included 10 participants per age group, totaling 60 students from both public and private schools. Stage 2 was conducted with eight focus groups, comprising 21 students from two public schools in the city of Porto Alegre. The groups were composed to include different age ranges (based on the availability of participants) to observe discussions across a diversity of ages present in each group. For stage 3, 287 students participated, with 165 girls (57.5%) and 122 boys (42.5%). The mean age was 10.2 years (SD = 1.5). Among the sample, 34 (11.8%) were 8 years old, 80 (27.9%) were 9 years old, 61 (21.3%) were 10 years old, 43 (15%) were 11 years old, 39 (13.6%) were 12 years old, and 30 (10.5%) were 13 years old. Only 26 (9.1%) attended a private school.

Regarding responses to the parental styles instrument, 11 participants (3.8%) responded only the instrument regarding the paternal caregiver, 70 (24.4%) responded only the one about the maternal caregiver, and 206 (71.8%) responded both instruments.

It is important to note that the project was developed following all ethical guidelines related to research with human subjects. As a benefit, at the end of the project, the schools received the overall results of the participating students (not individualized). The Research Ethics Committee of the Universidade Federal de Ciências da Saúde de Porto Alegre (UFCSPA, Federal University of Health Sciences of Porto Alegre) approved the project, under CAAE No. 90802718.0.0000.5345.

The Parental Responsiveness and Demandingness Scale, developed by Teixeira et al. (2004), is designed for adolescents and adults aged 14 and older. This scale is a refinement of the first Brazilian version by Costa et al. (2000), translated and revised from the instrument by Lamborn et al. (1991), based on the principles of Baumrind (1966) and MacCoby and Martin (1983). It has a bidimensional structure addressing the constructs of parental demandingness and responsiveness through 24 Likert-scale items (12 assessing demandingness and 12 assessing parental responsiveness). Children indicate the intensity or frequency that best represents their caregivers’ attitudes and behaviors separately, yielding a score for each dimension for each assessed adult. Internal consistency results (Cronbach’s alpha) were 0.78 for demandingness and 0.92 for responsiveness (Teixeira et al., 2004). The EREP scale was utilized in stages 1 and 2 of this study.

The Escala de Responsividade e Exigência Parental Infanto-Juvenil (EREP-inf, Child and Adolescent Parental Responsiveness and Demandingness Scale) was developed in stages 1 and 2 and administered in stage 3.

A Sociodemographic Questionnaire (developed by the authors for this study), in which the responsible caregiver who authorized participation filled out questions related to the child/adolescent (e.g., age range) and themselves (e.g., parental relationship with the participant). This instrument was used in all three stages.

The seven schools participating in the project signed the Research Site Consent Form, and invitations were extended to the families of enrolled students. Interested caregivers were sent the Informed Consent Form (ICF) and the sociodemographic questionnaire. All data collection used printed instruments. For stage 1, individual meetings with students were scheduled during school hours. One researcher was responsible for reading, questioning the comprehension and relevance of each item, and collecting suggestions for vocabulary or content adjustments. A second researcher recorded the respondent’s feedback.

Concurrently, judges were invited via email to fill out an individual document to assess content validity through: (a) theoretical dimension analysis using the Kappa Mean analysis method, suitable for categorical variables and investigating inter-judge agreement, allowing more than two participants. Subsequently, the judges assessed items for (b) content relevance and (c) language coherence, dividing the age range into two groups: 8-11 years and 12-13 years, considering different developmental stages for a more accurate assessment (Macarini et al., 2010). For stage 2, to enhance analysis precision, the following themes were added: (d) theoretical relevance of items and (e) overall scale assessment. The judges were also invited to suggest modifications. Items (b) through (e) were analyzed using the Content Validity Coefficient (CVC) method, known for mitigating deficiencies in other methods by using a Likert scale for individual and group assessments. The analysis criterion uses the mean score for each item, ranging from 0 to 1, with items with a CVC > 0.8 being considered acceptable (Cassepp-Borges et al., 2010).

In terms of the judges’ profiles, six researchers participated in stage 1, all with PhDs and postdoctoral experience in Psychology and Health Sciences, with a mean of 21 years of professional experience. They were from the Southern and Northeastern regions of Brazil. Stage 2 involved eight judges, two of whom also participated in stage 1. These professionals, holding master’s and doctoral degrees in Psychology and Health Sciences, had expertise in psychological assessment, positive psychology, clinical psychology, parenting, and/or child development, with a mean of 19 years of professional experience and residing in the Southern, Southeastern, and Northeastern regions of Brazil.

Stages 1 and 2 took place from June to October 2019 and were completed following the assessment and authorization of the EREP-inf by the original scale’s authors. Stage 3 was conducted from November 2019 to March 2020 in participating schools during school hours, with collective application for up to 20 students per room. The scale completion, supervised by up to three project researchers who read the instrument with the students and remained available for questions, took approximately 20 minutes and was finalized with a review by a researcher to ensure no items were left unanswered. Due to the school year transition between 2019 and 2020, the participants’ age groups were categorized based on the date of their first project participation.

In stage 3, the psychometric properties were assessed using Classical Test Theory analyses. The statistical assumption of normality was verified through Kolmogorov-Smirnov and Shapiro-Wilk normality tests, resulting in a non-homogeneous distribution. Sociodemographic data were analyzed using descriptive statistics and group comparisons with Mann-Whitney and Wilcoxon tests to verify differences between independent and paired groups, respectively. Statistical significance was set at 5%, and analyses were conducted using IBM®SPSS® (version 25).

Internal structure was assessed at two points using the MPlus software, version 7.11. Firstly, through Exploratory Factor Analysis (EFA) to investigate the factor structure, as this was the first application of the scale in this age range. Secondly, a model with fewer items was assessed through Confirmatory Factor Analysis (CFA). Both analyses were conducted independently for paternal and maternal responses. EFA was performed using the Principal Axis Factoring extraction method, which does not assume normality of indicators, with Promax oblique rotation allowing factor correlation (Costello & Osborne, 2005). Factor retention was analyzed using the Kaiser-Guttman method (eigenvalue > 1), suggesting the extraction of factors with eigenvalues greater than 1, through scree plot analysis and theoretical coherence of the found solution. For scale refinement in EFA, items were eliminated if they had: (a) factor loadings below 0.30, (b) cross-loadings, or (c) did not load on their theoretically expected structure in one of the samples, maternal or paternal. CFA was used to assess the bidimensionality of the solution using the WLSMV (Weighted Least Squares Mean-and-Variance adjusted) estimation method, robust for ordinal data (Wang & Wang, 2019). Fit indices analyzed included chi-square (χ²), Comparative Fit Index (CFI > 0.90 for satisfactory fit; > 0.95 for excellent fit), Tucker Lewis Index (TLI, same criteria as CFI), and Root Mean Square Error of Approximation (RMSEA < 0.08 for acceptable fit; < 0.05 for good fit; 90% confidence interval not exceeding 0.10) (Brown, 2015). Besides scale dimensionality, item factor loadings were assessed to investigate item content relevance for factor estimation. Items without significant factor loadings (p > 0.05) were excluded after CFA.

Following the EREP structure definition, internal consistency indices were investigated using Cronbach’s alpha (α), the most widely used reliability index in Classical Test Theory, and McDonald’s Omega (ω), more precise than the former, as it considers factor loadings and item error. Coefficients were interpreted as follows: < 0.60 = inadequate; 0.60 to 0.69 = marginal reliability; 0.70 to 0.79 = acceptable; 0.80 to 0.89 = good; and 0.90 or higher = excellent (Borsa et al., 2022).

In Table 1, the results of the judges’ assessments in stage 1 (regarding the opinions on the scale by Teixeira et al. (2004) for application in the sample between 8 and 13 years old) and stage 2 (regarding the opinions on the preliminary adaptation of the scale for the target audience) are compared using the Mean Kappa index (used to assess the theoretical dimension) and CVC (used in other assessments). Each analyzed object presents the total value of the scale assessment (ranging between 0 and 1) and the items that showed unsatisfactory results.

As shown in Table 1, the theoretical dimension showed satisfactory results in both stages, although in stage 2 the judges disagreed on items 6 and 8. The relevance of the items for the age group between 8 and 11 years showed inadequate results in stage 1 (items 8, 22, and 24) and satisfactory results in stage 2. The relevance for the age group between 12 and 13 years obtained adequate results in both stages, although the value decreased in stage 2 compared to the first stage. Unsatisfactory results were found regarding the coherence of the language for the age group between 8 and 11 years in stage 1 for the scale and items 4, 6, 8, 9, 11, 13, 14, 17, 18, 19, 21, 22, 23, and 24, which were revised for stage 2, with only the results of items 4 and 6 remaining below the acceptable criterion. The language coherence for ages 12 and 13 showed similar results to the younger age group: items 8, 17, 22, and 24 were below expectations in stage 1, and only item 6 remained with an unsatisfactory result in stage 2. The theoretical relevance and overall assessment of the scale for both age groups showed satisfactory results in stage 2.

In the qualitative assessment by the judges, some comments complemented the quantitative analyses described above and were considered for the revision of the scale. In stage 1, a general assessment of the instrument stood out: “it is difficult to know about the vocabulary of eight-year-old children, as it depends on variables such as socioeconomic educational level of the parents.” In stage 2, comments regarding item 2 (demandingness): “it reflects parental care for school activities”; items 3, 4, 5, 7, 8 (demandingness): “it establishes a subjective standard of demandingness”; item 7: “the question is too focused on games when, in fact, the focus is on monitoring the child’s leisure activities. It risks being biased”.

Among the main changes to the scale at the end of stage 2 are the fact that it was divided into two sheets (in which a different pronoun is used for each gender of the responsible caregiver, “he” and “she”) and a blank space was included to insert the name of the responsible person being assessed, to facilitate the understanding of the child-adolescent audience. If there are two caregivers of the same gender, the same pronoun version can be used twice. The header includes a visual explanation with blocks corresponding to the progressive increase of the Likert scale. Additionally, some items were reordered to mix the assessment of demandingness with responsiveness, thereby avoiding response bias. Furthermore, the scale’s graphic representation was enhanced (the font and table size within the sheet were enlarged), and the lines were alternately painted gray and white to facilitate reading. The language of the instrument and all 24 items were adapted according to the target audience and judges’ assessments, ensuring appropriate vocabulary for the age group.

The following are the results of the internal structure and scale accuracy analysis. In the EFA, Bartlett’s tests of sphericity (maternal 1774.6; df = 276; p < 0.001 and paternal 1927.3; df = 276; p < 0.001) and KMO (maternal 0.87 and paternal 0.90) suggested the correlation matrix’s adequacy for factor analysis. Five factors showed eigenvalues > 1, while the scree plots indicated the presence of two factors, which explained 41.4% (paternal) and 35% (maternal) of the total variance, with the first factor representing 31.6% and 25.8%, respectively.

Regarding the factor loadings in the EFA (Table 2), some items did not meet the established criteria and were excluded. These were items 8 and 19, which showed reversed loadings in the maternal sample, justifying their exclusion from both samples; item 14 showed similar values in both dimensions in both samples and had loadings < 0.3 in the paternal sample. Item 2 (“He/She wants to know about the activities I did at school [e.g., looks at my notebook, my diary, wants to know about my exams, etc.]”) had reversed loadings in both samples relative to the theoretically proposed dimension in the adaptation process. Given the similarity of the result in both samples and the relevance of its content (as it represents care and attention in the responsiveness dimension), it was decided to keep it in this dimension, which had the highest loading.

In the CFA, all loadings were significantly different from zero, as were the factor variances, except for item 11, which was not significant in the maternal sample. Consistently, the content analysis also influenced the exclusion of item 11 (“When he/she wants me to do something, he/she tells me to do it regardless of what I think”), as its construct of decision power remained represented by item 6 (“When something needs to be decided about me, what he/she says is more important than what I say”). In this model, the responsiveness dimension remained with 13 items and the demandingness dimension with seven items (Table 3).

Regarding the fit indices, the chi-square results for the maternal sample were χ² = 248.270; df = 169; p < 0.001 and for the paternal sample, χ² = 322.473; df = 169; p = 0.001, indicating that the p-value rejected the model for both samples. Since the chi-square result is sensitive to larger samples, it was important to consider the other results (González-Blanch et al., 2018). The RMSEA, CFI, and TLI indices indicated good fits: maternal sample - RMSEA (90% CI) = 0.041 (0.030 – 0.052); CFI = 0.965, TLI = 0.960; paternal sample - RMSEA (90% CI) = 0.065 (0.054 – 0.075); CFI = 0.940, TLI = 0.933. As for the precision indices, the Cronbach’s Alpha value was 0.87 (responsiveness) and 0.61 (demandingness) in the maternal sample and 0.90 (responsiveness) and 0.76 (demandingness) in the paternal sample. It is important to note that the demandingness Alpha results were median, with marginal reliability for the maternal sample (0.61) and acceptable for the paternal sample (0.76). Compared to Omega, the indices were higher: 0.91 (responsiveness) and 0.78 (demandingness) in the maternal sample and 0.92 (responsiveness) and 0.87 (demandingness) in the paternal sample. The paternal sample showed better values for both Alpha and Omega.

Regarding the analysis of the factor loadings of items that showed statistical significance, it was evident that the responsiveness dimension had better internal structure, as all items had factor loadings above 0.49. In the demandingness dimension, only one item in the maternal scale (i21) had a factor loading below 0.30. Of the others, three maternal demandingness items had loadings below 0.5 (i4: λ=0.49, i6: λ=0.43, and i23: λ=0.49) (Table 3). Specifically, despite the loading for item 21 in the maternal sample being below 0.30, it was retained due to meeting the stipulated criteria, demonstrating significance with p < 0.05, and being deemed content-relevant in representing the demandingness dimension by expert judges consulted during the content validity stage. This item assesses the child’s involvement in household chores (“He/She says I have to help with household chores [e.g., tidy my room, fold my clothes, wash the dishes, etc.]”), a content not covered by other scale items.

In terms of classifying the dimensions as high or low (as a criterion for defining PS), stratification was done based on medians, as per the previous version of the scale. Comparing the values of parental dimensions in the sample that responded about maternal and paternal caregivers (n = 206), the results showed statistical differences, with maternal responsiveness: mean = 41.0 (SD = 9.2) and median = 43.0; paternal responsiveness: mean = 39.4 (SD = 10.5) and median = 41.0; p = 0.025. Maternal demandingness: mean = 22.4 (SD = 4.2) and median = 23.0; paternal demandingness: mean = 22.7 (SD = 5.0) and median = 23.0; p = 0.039. Thus, the female sample showed higher mean scores and medians in both dimensions. No significant differences were found in the assessment of maternal and paternal demandingness and responsiveness concerning the child-adolescent gender variable (p > 0.05).