Grade R teachers working in 10 educare centres or preschools in the Western Cape, Cape Town area were recruited as the respondent group to complete the protocols of the E3SR. The teachers were full-time employees who currently taught Grade R. They had to be familiar with the child’s behavioural patterns, abilities and general traits across environments through their day-to-day interaction with the child in the preschools. Seventeen teachers gave consent to participate in the pilot study. A total of 330 protocols were received. The preschools included one alternative (n = 36), one private (n = 71), three governmental (n = 201) and five community-based (n = 22) preschools. All protocols were obtained from children between the age of 6 and 7 years. The demographic profile of the children on whom the protocols were based is presented in Table 1.

Table 1 shows that in this sample, 64% of children were male and 36% were female. English was the most frequently spoken first language (56%), followed by Afrikaans (37%) and Xhosa (6%). Other primary languages that were specified as mother tongues included French, Congolese and other South African languages, Sepedi or Zulu (2%).

The E3SR is a strength-based screening instrument designed to screen for emotional and social readiness in preschool children in South Africa, and was used for data collection (Munnik, 2018). The E3SR used a questionnaire format with a five-point Likert scale (Never, Rarely, Some of the Time, Most of the Time and Almost always). The instrument was used in a summative way. The instrument has clear instructions on how to complete the questionnaire. The E3SR consists of two sections.

Section A: The Demographic section includes questions on the demographics of the child, such as the child’s chronological age, birth order, gender, ethnicity, language of instruction in school, home language, and history of illness or disability. Information of the respondents is also recorded, such as the length of time that the child was known to the teacher, a rating on how well the child is known and if the child has been referred for special interventions.

Section B: This section included items for each of the domains (2) and sub-domains (9). The emotional competence domain comprised of 31 items across five sub-domains, namely, Emotional maturity, Emotional management, Positive Sense of self, Independence, and Mental well-being and Alertness. The social competence domain comprised of 25 items across four sub-domains, namely, Social skills, Pro-social behaviour, Compliance with rules and Communication. The theoretical and operational definitions for each scale and subscale with their personal attributes can be accessed from the unpublished doctoral thesis of Munnik (2018).

A composite score can be calculated for the full scale, each domain and sub-domain. The full-scale score reflects the level of readiness to enter mainstream education on the emotional–social level. The domain scores reflect the level of readiness on an emotional or social level.

A stratified sampling frame of preschools or educare centres registered under the Social Welfare Act in the Cape Town Metropolitan area was established. Socio-economic status (SES) was used to stratify the sample into high, middle, and low geographical areas. Schools within these areas were invited to take part in this study. The recruitment process entailed a multi-layered stakeholder consultation. Firstly, an invitation was sent to the principals of the preschools, which included the proposed purpose of the pilot and an outline of what their involvement would entail. The principals discussed the invitation with the teachers and identified teachers who expressed their interest to participate. The research team then contacted the identified teachers for recruitment purposes. Willing principals and teachers constituted willing schools. A meeting was convened with parents of preschool children at ‘willing schools’. The aim of the meeting was to explain that the school was participating in a pilot study and the value of the E3SR. Parents were also informed that the information will be provided by the teachers as part of an administrative process as described above. Thus, principals, parents and teachers had to agree in order for the school to be included.

Meetings between the research team, teachers and the principals were scheduled at identified settings. The main purpose was to give an outline of the research (pilot study) and to clarify what teachers’ involvement would entail. Teachers were invited to ask questions about the study, the test administration,the layout of the E3SR and about the dissemination strategy. The questionnaires were delivered to an identified teacher at each preschool for completion. The respondents (teachers) were able to contact a nominated researcher at any time to discuss reservations or difficulties that arose during the pilot study. These steps increased compliance with the administration guidelines and by extension, the reliability of the data. The completed questionnaires were collected as soon as the teachers indicated that they have completed the questionnaires.

The data set included 330 protocols for children aged 6–7 years old. Protocols received from two independent schools were 107 (32.4%); 201 (60.9%) protocols were received from three governmental schools and 22 (6.7%) protocols from five community-based settings. The number of protocols in the data set (n = 330) exceeded the threshold requirements on the number of cases per item and the overall threshold for robustness recommended by DeVellis (2016). The minimum cases per item ratio for validation studies should be five cases per item up to 300 cases after which the ratio can be relaxed. The pilot E3SR consisted of 56 items, which set the minimum threshold at 280 (5 × 56 = 280). The recommended threshold sample size of 300 was exceeded, which increased the robustness of the analysis in this validation study (DeVellis, 2016).

Before the multivariate statistical analysis commenced, the assumptions for multivariate statistical analysis and data reduction were tested, as recommended by Field (2013). Normal distribution was assessed using the Shapiro–Wilk test. Bartlett’s test of sphericity was used for testing homogeneity of variance. The Kaiser–Meyer–Olkin test (KMO) assessed sample adequacy.

Statistical analysis was conducted using SPSS (version 25). Internal consistency was assessed with Cronbach’s coefficient alpha. The dimensional structure of the E3SR was assessed by exploratory factor analysis (EFA), as recommended by Henson and Roberts (2006). The main aim of the EFA was to clarify how many items were loaded on the identified factors and to identify a reduced set of factors that would describe the structural inter-relationships amongst the domain and sub-domain scales in the E3SR (Henson & Roberts, 2006). Principal axis factoring (PAF) was used in this study, as the intention of PAF was to explain the common variance amongst variables by means of factors (Henson & Roberts, 2006). The direct oblimin method was used as the rotation method, which allows for factors to be correlated (Laher, 2010). Factor loadings were pushed towards 0 or 1.0 by decreasing the standard errors of the loadings for the variables with small communalities or increasing those of the correlations amongst oblique factors (Kline, 2013).

The interpretation and reviewing of items were performed by inspecting factor loadings, communalities and factor over-determinations (Nunnally & Burnstein, 1994).

Item inspection on the correlation matrix. Items that did not correlate with at least one other item significantly above 0.3 were omitted. Items with few significant correlations with other items above 0.3 were flagged for further inspection consistent with the recommendation from Kline (2013). Measures of Sampling Adequacy (MSA) were also assessed per item, with a threshold of 0.90.

Communalities. Communalities express the shared variance accounted for by all the extracted factors (Field, 2013). High communalities indicate a reliable item. The desirable level of communality was set at 0.40, and communalities should not vary over a wide range (Gaskin, 2016; Osborne, Costello, & Kellow, 2014).

Number of factors to extract. Factor extraction was informed by cross validation of methods: (1) eigenvalues that exceeded 1 and inspection of the scree plot consistent with the recommendation of Henson and Roberts (2006), (2) Parallel analysis (PA) (Horn, 1965) and (3) and the Velicer’s minimum average partial (MAP) test (Velicer, Eaton, & Fava, 2000).

Factor loadings. According to Costello and Osborne (2005), the threshold for factor loadings was set at 0.50. Items that loaded above 0.32 on more than one item were considered as cross-loading. Items that cross-loaded on two components were retained in the component, in which they obtained the highest loading, on condition that they obtained a minimum loading of 0.50 based on the recommendation of Williams, Onsman and Brown (2010). Items that did not load on any factors were removed after examination.

The Humanities and Social Sciences Research Ethics Committee at the University of the Western Cape granted ethical approval during the PhD study and again for the post PhD project, reference number: HS19/24. Permission to conduct research at the preschools were obtained from the Department of Basic Education and the principals.

An information sheet explained what the study entailed. Permission to include protocols in the final assessments was provided by parents. Teachers at participating schools consented to complete protocols as respondents. All participation was voluntary, and the right to withdraw without fear or negative consequence was upheld. All protocols were anonymised. Participants were informed of their rights and recourse if dissatisfied. The learners were the unit of analysis; however, they did not participate directly in the study. However, parents were appropriately informed of the study and consented to the school participation. Teachers recorded their assessment based on observation of the children during the normal course of the execution of teaching responsibilities.

The assumption of normality was violated in this sample for the overall scale, as well as all nine subscales (Shapiro-Wilk = p < 0.05). The results showed that the distribution was positively skewed, and the assumption of normality was not met for this group. This violation was in line with the expected results, as 6- to 7-year-old children are assumed to already have mastered most of the attributes of emotional social readiness in the fourth term of the academic year when data were collected. In other words, the distribution accurately reflected where the cohort should be in terms of the measured competencies. Thus, the non-normal distribution was, in fact, an accurate representation of the target group.

The KMO statistic of 0.96 suggests that sampling adequacy was within the accepted range. All items reported individual MSA values above 0.90, with the exception of Item 3 on the Mental Well-being subscale (MW3) with an MSA value of 0.73, which was still considered to be acceptable according to Field (2013). Bartlett’s test of sphericity indicated that correlations between items were sufficiently large for PAF, suggestive of a correlation matrix and not an identity matrix (χ² (1540) = 18918.98 p < 0.01).

Table 4 provides an overview of the internal consistency for the original and revised scales and subscales of the E3SR. Overall, the scale and subscales of the original E3SR were found to be internally consistent, as demonstrated by alpha levels indicative of good to excellent reliability. The reported internal consistency estimates suggest that the items hang together in a reliable way. The internal consistency results of the revised scale are also found to be excellent. The final Cronbach’s alphas for the E3SR range from 0.92 to 0.95 across the six sub-domains. The revised scale had an overall Cronbach’s alpha of 0.97 (36 items). The emotional competence domain reported a Cronbach’s alpha of 0.95 (22 items), and the social competence domain had a Cronbach’s alpha of 0.94 (14 items).