3.2.2.1 Methods of Data Collection. The aim of Study 2, in line with the exploratory sequential design, was to “refine and extend the qualitative findings” (Creswell, 2015, p. 546) of Study 1 and to test those findings in terms of their distribution in the population (Dörnyei, 2007). A questionnaire was deemed to be the most suitable instrument to use for this study, as it allowed for the collection of numerical or directly quantifiable data from a large group of people, which could then be used to determine the relationship between predefined categories (Dörnyei & Taguchi, 2010).

3.2.2.2 The Development, Piloting and Validation of the Questionnaire

3.2.2.2.1 Development of the Instrument. The questionnaire, written in Hungarian, began with an introduction inviting participants to share their DI and TEDI beliefs. They were assured of anonymity, voluntariness, and that there were no right or wrong answers.

 

In addition, the questionnaire aimed to understand what importance EFL teachers attach to various learner differences in TEFL as well as to certain challenges and enablers commonly associated with the implementation of DI. As my main objective with this measurement was to obtain a general understanding of participants’ overall judgement of these constructs (Poon et al., 2002), and these constructs were specific, narrow in scope (Fuchs & Diamantopoulos, 2009), and factual rather than non-factual (Dörnyei, 2007), I opted for the use of single-item scales instead of multi-item scales, as their use is considered as an acceptable form of measurement under the mentioned circumstances. Respondents rated the importance of each of these aspects on a 5-point scale, and they could also mention further aspects in open-ended questions following each of these three topics, which were as follows:

 

While the above single questionnaire items could have been treated as background variables, in the present study I opted for treating them deliberately as thematic elements as they directly address two sub-research questions. Beliefs about addressing learner differences corresponds to RQ 1.2 and focuses on participants’ conceptualisation of DI by identifying which learner differences they believe are most important to address, while Perceived challenges of DI and Perceived enablers of DI correspond to RQ 2.2, as they measure the importance participants attach to specific challenges and enablers in implementing DI. The grouping of the questionnaire items was informed both by the existing literature (e.g., Letzel et al., 2020; Melesse, 2015; Santangelo & Tomlinson, 2015; Tzanni, 2018) and by the findings of Study 1, which addressed the same sub-research questions (RQ 1.2 and 2.2).

3.2.2.2.2 Piloting. The instrument was piloted in the spring of 2023. I employed a combination of non-probability sampling methods, specifically convenience and snowball sampling as suggested by Dörnyei (2007), to capture as wide a range of perspectives as possible from the Hungarian primary and secondary school EFL teacher population. First, I sent the link to the questionnaire to my teacher colleagues and then asked them to recruit further participants by distributing the link in their professional circles.

3.2.2.2.3 Reliability Analysis. The aim of the pilot study was to check if the questionnaire constructs produce reliable results. To this end, for each of the 11 scales, the Cronbach’s Alpha internal consistency reliability coefficients were calculated and unrotated principal component analyses (PCA) were ran to see if the grouped items loaded onto the same dimension (Székelyi & Barna, 2002). In all cases, the Cronbach’s Alpha value reached the .6 threshold (Dörnyei and Csizér, 2012). Regarding factor loadings, in the first round of PCA, 10 scales produced a single component, while in the case of one scale (differentiation of the learning environment) I had to remove one item from the scale (“I strive to ensure that my students always work with different partners”), so that the remaining items would load onto a single dimension. After deleting this item, the second round of PCA revealed uni-dimensionality for all scales. Table 8 presents the results of the reliability analysis.

 

 

As one item had to be deleted from the construct of differentiation of the learning environment, I decided to include two new items for this scale for the main study (“I take deliberate efforts to make myself approachable and available to my students”, “I follow up privately on conflicts and problems”), adapted from Santangelo and Tomlinson’s (2012) questionnaire measuring teachers’ DI practices. With this addition, the modified questionnaire was now considered as the final version to be used in the main study. The English translation of the questionnaire items is attached in Appendix F.

 

3.2.2.3 Participants and Data Collection Procedures. The data collection in the main questionnaire study was administered online from May to July 2023. Similar to the pilot study, a mixture of non-probability sampling methods (Dörnyei, 2007), namely convenience and snowball sampling were used to make sure that the participants reflect the diversity of the population of Hungarian primary and secondary school EFL teachers as much as possible. First a database of publicly available email addresses of EFL teachers from primary and secondary schools in Hungary was compiled (comprising altogether almost 1000 addresses), and calls for participation, along with the link to the questionnaire, were sent out to these teachers (the text is attached in Appendix G). The link was also shared on social media platforms, and the author requested their former colleagues, some of whom were the heads of the English departments at their respective schools, to distribute the link among their colleagues as well.

3.2.2.4 Methods of Data Analysis. Data were analysed using the Statistical Package for the Social Sciences (SPSS) version 29. Before running descriptive and inferential statistical procedures, similar to the pilot study, the reliability of the multi-item scales was checked. To this end, Cronbach’s Alpha coefficients were calculated for each of the 11 scales to determine internal consistency, and unrotated principal component analyses (PCA) were performed (Székelyi & Barna, 2002) to check whether the items of the scale loaded onto the same dimension. With one exception (differentiation by learning profile), all scales met or exceeded the minimum acceptable Cronbach’s Alpha threshold of .6, as suggested by Dörnyei and Csizér (2012). The first round of PCA indicated that seven scales formed a single factor, while four scales (differentiation of the learning environment, differentiation by readiness, differentiation by learning profile, cooperation with stakeholders) required the removal of one or two items to achieve a single dimension. After these items were deleted, a second round of PCA confirmed that all scales showed uni-dimensionality (Appendix F presents the questionnaire items, with the items deleted from the analysis highlighted in italics). The finalisation of the constructs is presented in Table 9 below.

 

 

Next, descriptive statistical methods (measures of central tendency, skewness, kurtosis) were used to test the normality of the data. As can be seen in Table 10, most scales showed normal distribution with their skewness and kurtosis values being within the acceptability range of ± 2.0 (George & Mallery, 2016).

 

 

Some scales did have values outside the ±2.0 range, but, as the sample size was above 200 (N = 212), these deviations from normality were not expected to significantly impact the results, in line with Tabachnick and Fidell’s (2007) observation that such non-normalities in large samples are unlikely to “make a substantive difference in the analysis” (p. 80). Besides, the kurtosis level of 4.84 in the scale Beliefs: Importance of DI is also qualitatively justifiable as it is in line with the expectation that teachers would generally consider DI to be an important element of their pedagogical practices. In light of these considerations, it was deemed appropriate to use parametric tests for the data analysis.