1. Introduction
Teacher turnover and attrition pose persistent challenges for education systems worldwide, underscoring the need to better understand the factors that influence teachers’ intentions to leave both their schools and the profession. While these topics have been widely explored in general education research (Guarino et al., Reference Guarino, Santibanez and Daley2006; Ronfeldt et al., Reference Ronfeldt, Loeb and Wyckoff2013), relatively few studies have focused on language teachers, particularly language teachers working in international or transnational contexts. To address this gap, the present study replicates McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) investigation of the relationship between teacher commitment and turnover intentions, shifting the focus from Hong Kong schoolteachers to expatriate L1 English-speaking teachers in East Asia, often referred to as native English-speaking teachers (NESTs).
Following replication typologies outlined by McManus (Reference McManus2024a, Reference McManus2024b) and Porte and McManus (Reference Porte and McManus2019), this study is best classified as a close replication with an extension. To clarify how this replication was carried out, it is helpful to revisit the design of the original study. McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) study had two components: (1) regression analyses predicting turnover intentions based on occupational and organizational commitment, and (2) group comparisons of commitment and turnover intentions across institutional subgroups (e.g. religious vs non-religious schools and EMI vs CMI schools). This closely replicates the original regression analyses using the same theoretical framework, constructs, instruments, and analytic procedures, while modifying only one major feature: the population. The extension component involves new subgroup comparisons of commitment and turnover intentions within the replication sample and direct comparisons to the original sample’s results.
This study is motivated by three interrelated aims: (1) testing the external validity of McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) findings with a different teaching population, (2) addressing a gap in the literature on language teacher attrition, and (3) exploring whether and how institutional settings and teacher qualifications are associated with commitment and turnover intentions among expatriate NESTs. These aims reflect practical and theoretical concerns relevant to language teaching research and policy. At the same time, they align with a broader framework of replication value (Isager et al., Reference Isager, van Aert, Bahník, Brandt, DeSoto, Giner-Sorolla, Krueger, Perugini, Ropovik, van’t Veer, Vranka and Lakens2023) recommended by McManus (Reference McManus2024b), which encourages researchers to select replication targets based on (1) the importance of the original claim, (2) the uncertainty of the evidence supporting it, (3) the potential for replication to reduce that uncertainty, and (4) the feasibility of replication. By replicating a widely cited, theoretically grounded study with substantial predictive power, but with uncertain generalizability to more mobile, transnational teaching populations, the current study meets each of these criteria.
First, McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015) offers a clear, theoretically grounded, and widely cited model of how teacher commitment, particularly affective commitment, predicts intentions to quit. The study was large scale (N = 1,060), based on established constructs and measures, and has shaped thinking about job retention in both education and organizational psychology research. The value of gaining additional clarity about whether these findings extend to other teaching populations is therefore considerable.
Second, there is meaningful uncertainty surrounding the applicability of McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) results beyond the original context. Their study focused exclusively on full-time, local schoolteachers in Hong Kong. In contrast, expatriate NESTs across East Asia usually have shorter-term contracts, reduced institutional integration, and work in diverse institution types (e.g. public schools, universities, and private academies). Whether the same commitment–turnover dynamics hold in this less stable, more transnational population is unknown.
Third, this replication was designed to reduce that uncertainty in a focused and methodologically rigorous way. It applies the same theoretical model, instruments, and regression procedures as the original study, while modifying only the population. At the same time, it extends the original design by exploring new subgroup differences within the replication sample (e.g. institutional settings, teaching qualifications) and by comparing overall commitment and turnover scores to those reported in the original study. While no single replication can fully resolve issues of generalizability, this replication assesses whether the teacher commitment–turnover results established by McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015) extend to expatriate NESTs across diverse institutional contexts in East Asia.
Finally, the replication was feasible to carry out. The original methods were clearly documented, and the instruments could be adapted with fidelity. Because it was survey research, it was feasible for an individual researcher to recruit participants and replicate it without funding. A sufficiently powered sample was obtained (N = 215), and data, instruments, and results have been made publicly available via the Open Science Framework (OSF) (Moodie, Reference Moodie2025) to support open science practices. In this respect, the study meets both conceptual and pragmatic criteria for high replication value.
2. Background
Teacher turnover and attrition are persistent problems for education, including language education, due to their negative effects on student outcomes, instructional continuity, and institutional stability (Ronfeldt et al., Reference Ronfeldt, Loeb and Wyckoff2013; Sulis et al., Reference Sulis, Babic, Mairitsch, Mercer, Jin and King2022). Although related, the two terms describe distinct outcomes: turnover refers to teachers changing schools, while attrition refers to permanently leaving the profession. Both forms of departure create disruptions, but attrition is especially damaging as it depletes professional knowledge and experience.
Attrition is particularly common in the early years of teaching. Day and Gu (Reference Day and Gu2007) noted that teachers typically develop stable levels of professional identity and commitment after four to seven years. However, studies show that many leave the profession before this point, often within the first five years (Guarino et al., Reference Guarino, Santibanez and Daley2006; Ingersoll, Reference Ingersoll2012). These early exits weaken teacher self-efficacy and disrupt school communities, hindering the long-term development of professional expertise (Ronfeldt et al., Reference Ronfeldt, Loeb and Wyckoff2013). This contributes to instability at the institutional level and negatively impacts teachers, students, colleagues, and educational systems more broadly (Ronfeldt et al., Reference Ronfeldt, Loeb and Wyckoff2013; Sulis et al., Reference Sulis, Babic, Mairitsch, Mercer, Jin and King2022). Taken together, these effects highlight why understanding the causes and precursors of teacher attrition is critical.
2.1. McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) study
McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015) contributed to this area by considering how different types of commitment predicted intentions to quit schools and the profession among a large sample of Hong Kong schoolteachers. Their study used established measures from the field of industrial/organizational (I/O) psychology, including Meyer et al.’s (Reference Meyer, Allen and Smith1993) multidimensional model of commitment and Becker and Billings’ (Reference Becker and Billings1993) turnover intention scales.
Meyer et al.’s (Reference Meyer, Allen and Smith1993) commitment model (see also Meyer et al., Reference Meyer, Becker and Vandenberghe2004; Meyer & Herscovitch, Reference Meyer and Herscovitch2001) identifies three dimensions of commitment (affective, normative, and continuance) targeted across two domains (organization and occupation). For this reason, it is often referred to as a 3 × 2 commitment model. Affective commitment refers to employees’ emotional attachment to their organizations and occupations, such as wanting to be in a particular career or workplace because of personal interest and identification with it. Continuance commitment, in contrast, refers to consideration of the costs and benefits of staying with or leaving an organization or profession, such as contemplating the difficulties of changing a career or workplace. Normative commitment, however, refers to an employee’s feeling of loyalty, duty, or moral obligation to continue in a workplace or profession. This model of commitment is the most widely applied one in I/O psychology research, and it has been validated across numerous professions, contexts, and languages (Meyer et al., Reference Meyer, Stanley, Jackson, McInnis, Maltin and Sheppard2012).
McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) study made several important findings regarding the nature of teachers’ commitments and its relation to their turnover intentions. First, they found that affective commitment was positively related with normative commitment, but negatively correlated with continuance commitment and intentions to quit a school and the profession. As they stated:
Of particular interest is the positive relationship between continuance commitment to the organization and turnover intentions, and the negative relationship between affective and normative commitment to turnover intentions. Higher affective and normative commitment predict lower turnover intentions (or higher intentions to remain) and the opposite is true for continuance. (McInerney et al., Reference McInerney, Ganotice, King, Marsh and Morin2015, p. 16)
Thus, in essence, what they found was that those who feel positive affect and a sense of duty and loyalty to their schools and the profession tend to have lower intentions to quit, whereas those who are at a school or profession because of a lack of alternatives tend to have higher intentions to quit. While on the surface these findings may not be surprising, the significant contribution of their study was documenting the magnitude of the effects. In their regression analyses, their models showed that commitment could predict 44.2% of the variance in intentions to quit schools and 45.2% of the variance in intentions to quit the profession. To give this finding a reference point, the median values for regression analyses in applied linguistics research is 32% (Plonsky & Ghanbar, Reference Plonsky and Ghanbar2018). In addition, when controlling for other variables, they found that the affective commitment to organizations alone had a strong, negative relationship with turnover intentions to schools (β = − .56, p < 0.01) and the profession (β = − .45, p < 0.01). In other words, as affective commitment to schools increases, the likelihood of quitting both the school and profession decreases. However, it was somewhat surprising that affective commitment to schools (β = − .45, p < 0.01) was a stronger predictor of quitting the profession than affective commitment to the profession itself (β = − .23, p < 0.01). One interpretation for this would be that conceptually quitting a school and quitting the profession equate to the same thing among many Hong Kong schoolteachers. It would be surprising if this result replicated among language teachers, who tend to have low job stability but high job mobility. In fact, it should be expected in the replication that affective commitment to schools will highly predict intentions to quit schools and affective commitment to the profession will highly predict intentions to quit the profession.
Nevertheless, other research supports the fact that affective commitment is highly relevant for workplace outcomes. For instance, it has been strongly correlated with workplace performance and organizational citizenship (Meyer et al., Reference Meyer, Stanley, Jackson, McInnis, Maltin and Sheppard2012; Morin et al., Reference Morin, Meyer, McInerney, Marsh and Ganotice2015), and it has a hypothesized relationship to student learning outcomes in the field of language teaching (Moodie & Meerhoff, Reference Moodie and Meerhoff2020).
Recent results by Moodie (Reference Moodie2024) provide a useful point of comparison to McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) results regarding affective commitment, showing that occupational well-being similarly predicts intentions to quit schools (β = − .57, p < 0.01) and the profession of ELT (β = − .52, p < 0.01) among expatriate NESTs, but with slightly lower explained variance from the full model (36.8% for leaving the organization; 36.9% for leaving the profession). Replicating McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), therefore, also allows for testing whether commitment predicts turnover intentions as strongly as occupational well-being does among NESTs in Asia.
2.2. Research aims and hypotheses
Although teacher attrition has been widely studied in mainstream education, few studies have examined turnover among language teachers, particularly those in transnational or contract-based roles. Recent reviews identified only 10 to 15 empirical studies on language teacher attrition (Ayar, Reference Ayar and Karpava2023; Mason, Reference Mason2017), with none addressing commitment or turnover intentions among expatriate NESTs. This gap is significant given the instability and diversity of teaching conditions in global ELT. By applying a validated theoretical framework to a distinct group of expatriate NESTs, this replication clarifies whether established models of commitment and turnover extend to more mobile, internationally employed teachers and provides insights into the factors influencing language teacher retention.
The two aims of this study are closely adapted from McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015). The first aim is to examine the relationships between organizational and occupational commitment and intentions to quit workplaces and the profession of ELT. In line with this aim, it will follow the first two hypotheses of McInerney et al.:
Hypothesis 1. Affective commitment to organization and profession is significantly and negatively related to turnover intention. Teachers who endorse a high level of affective commitment will express a weak desire to quit the organization and the profession.
Hypothesis 2. Normative commitment to organization and profession is significantly and negatively related to turnover intention. Teachers who endorse a high level of normative commitment will express a weak desire to quit the organization and the profession. (p. 14)
McInerney et al.’s third hypothesis, which supposed that continuance commitment would also be negatively related to turnover intention, was not supported. In fact, they found that the opposite was true and that continuance commitment had a small positive correlation with intentions to quit. Therefore, the third hypothesis for this study will be revised as follows:
Hypothesis 3. Continuance commitment to the organization and profession is significantly and positively related to turnover intention. English teachers with higher levels of continuance commitment will also indicate desires to quit their organizations and the profession.
The second aim, as in McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), is to examine group differences in commitment and turnover intentions. This aspect represents the extension of the original study. McInerney et al. investigated group differences specific to the Hong Kong context, such as between teachers in religious and non-religious schools, and between teachers in EMI and CMI schools. However, these institutional distinctions are not applicable to the expatriate NEST population examined in this study. Instead, the present study extends McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015) by comparing commitment and turnover intentions between the original and replication samples. The study also examines subgroup differences within the replication sample. Specifically, it compares participants across institutional types (private academies, public schools, and universities/colleges) and by teacher qualification status (those with and without qualifications from their home countries). This comparison builds on recent findings indicating that expatriate ELT faculty with teacher qualifications reported significantly higher affective and normative commitment to both their organizations and the profession (Moodie, Reference Moodie2023). Because profiles dominated by affective and normative commitment have been linked to higher occupational well-being and lower turnover intentions (Morin et al., Reference Morin, Meyer, McInerney, Marsh and Ganotice2015), following up on that smaller study provides an opportunity to examine whether these differences also appear in a broader sample of expatriate English teachers across East Asia. This may offer insight into the role of teacher qualifications as a predictor of commitment and turnover among NESTs. Accordingly, the study proposes three additional hypotheses as part of the extension:
Hypothesis 4. The expatriate English teachers in this replication will have lower levels of occupational and organizational commitment but higher levels of intentions to quit schools and professions than the Hong Kong school teachers in the original study.
Hypothesis 5. English teachers in universities/colleges will be higher in affective commitment to the organization and profession than teachers in public schools and private academies.
Hypothesis 6. English teachers with teacher qualifications from their home countries will be higher in affective and normative commitment to their organizations and the professions than teachers without qualifications, but they will be lower in continuance commitment.
3. Methods
This study replicates McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), which investigated how occupational and organizational commitment predicted teachers’ intentions to leave their schools and the profession. As indicated, it is a close replication with an extension (see McManus, Reference McManus2024b; Porte & McManus, Reference Porte and McManus2019). The close replication component involves changing one variable, the population. The extension component involves additional subgroup comparisons within the replication sample and also cross-sample comparisons with the original. This study adopts the same theoretical framework, measures, and analytic procedures as the original, including confirmatory factor analysis (CFA), reliability assessments of the measures, multiple linear regression, and comparisons of means. The measures, data, assumption checks, and results for the replication are freely available on osf.io (Moodie, Reference Moodie2025).
3.1. Participants
The term expatriate English teacher is often used broadly to describe transnational teachers with diverse linguistic and educational backgrounds, including both first language (L1) and additional language (L2) English speakers. In this study, however, the target population is narrowed down to expatriate L1 English-speaking teachers, commonly referred to as NESTs in the literature, and who are eligible for work visas in East Asian countries such as Japan, South Korea (hereafter Korea), Taiwan, and Hong Kong.
For instance, in Korea, eligibility for an English teaching visa (E-2) requires applicants to be native speakers who were educated in one of seven designated countries: Australia, Canada, the Republic of Ireland, New Zealand, South Africa, the United Kingdom, or the United States (EPIK, 2024). Accordingly, the inclusion criteria for the study required participants to be current L1 English-speaking teachers from one of these countries and teaching in East Asia.
A total of 215 participants completed the survey, including 108 (50.2%) females, 98 (45.6%) males, and 9 (4.2%) who preferred not to state their gender. The average age was 34.3 (SD = 9.1) years old. Their nationalities included Australia (14 or 6.5%), Canada (33 or 15.3%), Ireland (8 or 3.7%), New Zealand (6 or 2.8%), South Africa (15 or 7.0%), the UK (30 or 14.0%), and the USA (109 or 50.7%).
Regarding teaching contexts, 31 (14.7%) were working in private academies, 129 (61.1%) were teaching in public schools, and 51 (24.2%) were teaching in colleges or universities. Thirty-nine (18.1%) held public school teacher qualifications from their home countries, and 169 (78.6%) held ELT qualifications, such as a CELTA or TESOL certificate. The sample was geographically concentrated: 204 participants (94.9%) were teaching in Korea, with only 11 (5.1%) teaching in other East Asian countries. This results in a sample that is heavily weighted toward the Korean expatriate teaching context.
A comparison of demographic information with the original study is presented in Table 1 below. Although the current sample (N = 215) is substantially smaller than in the original study (N = 1,060), it was sufficient to conduct multiple linear regression analyses with adequate statistical power. The smaller sample reflects the scope of this replication project, which was conducted by a single researcher without external funding. In contrast, the original study was part of a large-scale, government-funded research initiative that included multiple interrelated investigations and access to a broader sampling pool.
Table 1. Comparison of basic demographic information in both studies
Note: The degrees indicate the highest levels of education.
Participants in McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) study were full-time schoolteachers in Hong Kong, including those teaching in primary and secondary schools, public and religious schools, and across EMI and CMI settings. Compared to the current sample, they had nearly twice the average years of teaching experience. The proportion of female participants and the percentage of respondents with master’s degrees were also higher in the original study. Although McInerney et al. included income data, this variable had negligible effects and was therefore excluded from the replication.
3.2. Measures
This study used the same commitment measures as McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), based on Meyer et al.’s (Reference Meyer, Allen and Smith1993) 3 × 2 component model. The items were slightly modified for context by replacing references to ‘this profession’ with ‘the English language teaching profession’ and ‘this organization’ with ‘this school.’ Following McInerney et al., the three highest-loading items for each commitment dimension from Meyer et al. were selected. To remind readers, affective commitment refers to an employee’s emotional attachment to their workplace or profession; continuance commitment reflects the perceived costs of leaving; and normative commitment involves a felt sense of duty or obligation to stay.
Turnover intentions were measured using the same scales as McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), originally developed by Becker and Billings (Reference Becker and Billings1993). These instruments assess intentions to quit an organization (four items) and to leave the profession (four items). As with the commitment items, these items were minimally adapted for this context, such as by replacing references to ‘organization’ with ‘school’ and ‘this profession’ with ‘the ELT profession.’ The same seven-point Likert scale was used for all items in both the replication and original study, ranging from 1 (strongly disagree) to 7 (strongly agree).
3.3. Procedures
Unlike the original study, which used representative sampling from a defined population of Hong Kong schoolteachers across specific school types and districts, this replication employed a convenience sampling approach. Participants were recruited through online networks and professional associations for expatriate English teachers. This modification reflects the practical constraints of conducting research with a transnational population for whom no centralized organizational body exists. While this change limits the representativeness of the sample, it is unlikely to meaningfully hinder the regression analyses, as the procedures, instruments, and constructs remained consistent with the original study. Accordingly, this modification is classified as minor change.
For reference, McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015) faxed invitations to 15 secondary and 15 primary schools across various districts in Hong Kong. In contrast, this replication posted banner ads on online forums frequented by expatriate English teachers and circulated calls for participation in relevant Facebook groups. In addition, colleagues were asked to share the study invitation with eligible participants in their networks. While McInerney et al. did not specify their survey platform, this study used Google Forms. McInerney et al. reported a response rate of 52.2%; however, due to the nature of this sampling approach, response rates could not be calculated in the present study.
3.4. Analysis
The analysis closely mirrored McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) study. The first steps involved cleaning the data and then doing preliminary analyses, such as checking item response distributions, confirming the construct validity and reliability of the scales, and checking the assumptions for multiple linear regression. Data cleaning involved recoding categorical variables into numeric formats and screening responses for suspicious patterns (e.g. straight-lining), unusual replies (e.g. ‘attack helicopter’ for gender), and ensuring respondents fit the study’s criteria. Accordingly, 13 (5.7%) cases were removed.
To assess normality, the skewness and kurtosis values of measures were checked. Results indicated that affective commitment to the profession had a slight negative skew (−1.1), but this did not violate the assumption of normality based on residual plots. For reference, McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) accepted range for skewness was ±2. As with McInerney et al., no outliers were detected.
Construct validity was assessed using confirmatory factor analysis (CFA), conducted in Jamovi, an open-source R-based platform. As in the original study, two CFAs were performed: the first examined the full 3 × 2 factor model of commitment (Meyer et al., Reference Meyer, Allen and Smith1993), and the second tested the two-factor model of turnover intentions (Becker & Billings, Reference Becker and Billings1993).
The CFA indicated that there was a problem with item NCP1. Further inspection indicated an error on the original survey, so it was removed from the analysis.
The commitment model showed acceptable to good fit, similar to McInerney et al.’s results (see Table 2). In contrast, the fit indices for the turnover intention scale were mixed. While the model has acceptable fit on the Comparative Fit Index (CFI = .92), the Tucker-Lewis Index (TLI = .88) and Root Mean Square Error of Approximation (RMSEA = .13) indicated suboptimal fit. This may be due to a smaller sample size, convenience sampling, or greater participant heterogeneity in the replication sample. Unlike McInerney et al.’s relatively homogenous sample of full-time local teachers in Hong Kong, the current study included participants from multiple countries and institutional types. Despite the weaker fit on some indices, the overall CFA results support a reasonable relationship between the theorized constructs and observed data.
Table 2. Comparison of CFA results in both studies
Note: All factor loadings were statistically significant at p < .001 in both studies.
Next, assumptions for multiple linear regression were assessed. The Durbin-Watson statistic indicated no evidence of autocorrelation (DW = 2.02, p = .92). Normality of residuals was confirmed via visual inspection of Q–Q plots and supported by the Shapiro–Wilk test (S = 0.995, p = .79). Visual inspection of residual versus predicted value plots indicated approximate linearity and homoscedasticity. No major violations of regression assumptions were detected.
Consistent with McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), two blockwise multiple regression analyses were conducted: one with intentions to quit the school as the dependent variable and one with intentions to quit the profession as the dependent variable. In both models, to control for demographic variables, they were entered in the first block. In the second block, the three organizational commitment factors were entered, and in the third block, the three occupational commitment factors were entered. These analyses correspond to Hypotheses 1 to 3.
To address Hypothesis 4, independent samples t-tests were run for each commitment and turnover intention factor, comparing the current sample of expatriate teachers to the original sample of Hong Kong schoolteachers. Hypothesis 5 was tested using two one-way ANOVAs, one for ACO and one for ACP, to compare participants teaching at different levels (universities, schools, and private academies). To address Hypothesis 6, independent samples t-tests were run, one for each commitment factor, comparing participants with teacher qualification to those without. Although omnibus procedures such as MANOVA were considered, simpler tests were preferred to maintain consistency with McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015).
One limitation of McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) study is that they did not adjust their alpha level to account for multiple comparisons. To address this, the replication will apply a Bonferroni correction, setting the alpha level to p = .006 (i.e. .05/9), based on the nine variables used in the multiple regression analyses.
Another limitation is that McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015) did not benchmark effect sizes. Accordingly, this replication draws on field-specific standards where available. Effect sizes for the multiple regression analyses will be indicated with R2, adjusted R2, and standardized regression coefficients (β). The benchmark for R2 = .32, which is the median value in applied linguistics (AL) research (Plonsky & Ghanbar, Reference Plonsky and Ghanbar2018). For correlations, Plonsky and Oswald’s (Reference Plonsky and Oswald2014) benchmarks are applied, with r = .25 as a small correlation, .40 as a medium correlation, and .60 as a large correlation. As with McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), effect sizes for ANOVAs and t-tests will be given with eta squared (η2). Because no field- or topic-specific benchmarks for eta squared were found, Cohen’s Reference Cohen1988 general benchmarks will be used, with η2 = .01 as a small, .06 as a medium, and .14 as a large effect.
To assess replication success, the study will consider whether findings align with those reported by McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015) regarding statistical significance, directionality, and the magnitude of effects (i.e. R2 and β for the linear regressions). Specifically, a finding will be considered replicated if it achieves statistical significance (p < .006), displays the same directional relationship as the initial study, and yields an effect size that is broadly comparable.
To clearly communicate the similarities and differences between the original and replication designs, Table 3 presents a side-by-side comparison of key study features.
Table 3. Feature comparison in both studies
Note: Minor adaptations were made to survey language to fit the ELT context in the replication. ‘Same’ refers to identical measures and procedures.
4. Results
4.1. Descriptive results
The first part of the analysis replicates the descriptive procedures of McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), including reporting factor correlations, means, standard deviations, and internal consistency estimates (see Table 4). Cronbach’s alpha values in the current study ranged from .67 to .86, which are comparable to those in the original study (.68 to .87). Notably, normative commitment to the organization (NCO) demonstrated higher internal consistency in the replication (α = .83) compared to the original (α = .68), while affective commitment to the organization (ACO) was slightly lower (α = .71 vs. α = .80).
Table 4. Comparison of correlation matrices and descriptive results
Note: *p < .05, **p < .01, ***p < .001.
Two noteworthy differences emerged in the inter-factor correlations. The relationship between affective and normative commitment to the organization (ACO–NCO) was stronger in the replication (r = .62) than in the original study (r = .42), representing a large effect. In contrast, the correlation between affective and continuance commitment to the profession (ACP–NCP) was substantially weaker in the replication (r = .50) compared to the original (r = .84).
4.2. Regression analyses
As with the original study, two blockwise multiple regression analyses were conducted to examine the predictive relationships between commitment dimensions and turnover intentions. One model used intentions to quit the school as the dependent variable; the other used intentions to quit the profession. To control for demographic variables, they were entered in Block 1. Following McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), gender and highest education were included. However, income was excluded due to its negligible effect in the original study and was replaced with workplace type (private academy, public school, or university/college). Organizational commitment variables were entered in Block 2, and occupational commitment variables in Block 3. Model fit results are shown in Table 5.
Table 5. Model fit measures in current study
As shown in Table 5, the demographic factors in Block 1 were statistically insignificant at the adjusted alpha level. However, the addition of organizational commitment variables in Block 2 significantly improved both models: together with demographics, they accounted for 32.0% of the variance in intentions to quit the profession (R2 = .320, p < .001) and 50.1% for intentions to quit the school (R2 = .501, p < .001). In Block 3, the inclusion of occupational commitment further increased explanatory power: the final models accounted for 59.7% of the variance in intentions to quit the profession (R2 = .597, p < .001) and 51.8% for quitting the school (R2 = .518, p < .001).
These results indicate that the 3 × 2 commitment model is more predictive of turnover intentions among expatriate English teachers than with the Hong Kong schoolteachers in McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), where commitment accounted for 44.2% of variance in intentions to quit the profession and 45.2% for quitting the school (see Table 6).
Table 6. Comparison of standardized beta coefficients: commitment and intentions to quit
Notes: *p < .05, **p < .01, ***p < .001. Full results for the regressions will be provided as supplementary online materials. Workplace 2–1 compares teachers in schools (2) with those in private academies (1), and Workplace 3–2 compares those working in universities (3) to those in schools (2).
Table 6 presents the standardized beta coefficients used to evaluate Hypotheses 1 through 3. As predicted in Hypothesis 1, affective commitment to both the organization (ACO) and the profession (ACP) was significantly and negatively associated with turnover intentions. A key divergence between the replication and the original study, however, was the reversed pattern of predictive strength between these two forms of affective commitment. In McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015), ACO was the strongest predictor of intentions to quit both schools (β = − .564, p < .001) and the profession (β = − .448, p < .001). In contrast, as anticipated, the current study found that ACP was the strongest predictor of intentions to quit the profession (β = – .509, p < .001), while ACO remained the strongest predictor of intentions to quit schools (β = − .531, p < .001). This reversal likely reflects contextual differences between the two samples, particularly the greater organizational stability and professional integration of Hong Kong schoolteachers compared to the more mobile and transient employment conditions of expatriate English teachers in East Asia. For Hong Kong teachers, quitting a school may be closely tied to leaving the profession, whereas for expatriate teachers, the two decisions may be more distinct. This is explored in further depth in the discussion.
Hypotheses 2 and 3, which proposed that normative commitment and continuance would negatively predict turnover intentions, respectively, were not supported.
4.3. Comparison of means
Addressing Hypotheses 4 to 6 involved comparisons of means, marking the extension of the original study. Hypothesis 4, which predicted lower levels of commitment and higher turnover intentions among expatriate English teachers compared to Hong Kong schoolteachers, was largely supported. As shown in Table 7, the replication sample scored significantly lower on five of the six commitment measures and significantly higher on both turnover intention measures. The one exception was affective commitment to the profession (ACP), which was slightly higher in the replication but did not meet the adjusted alpha level for statistical significance.
Table 7. T-test results: Replication versus original study
In terms of effect sizes, the largest differences were observed in continuance and normative commitment to the profession (NCP and CCP), both of which showed large effects. Medium effects were found for affective and normative commitment to the organization (ACO and NCO), while continuance commitment to the organization (CCO) and the two turnover intention scales (IQP and IQS) showed small effects. Although the direction of group differences was consistent with Hypothesis 4, the magnitude of differences in turnover intentions was relatively modest.
Additional comparisons are illustrated in Figs. 1 and 2, which provide a visual summary of mean differences and response variability between the replication and original study samples. As expected, the replication sample scored lower on most commitment dimensions and higher on turnover intentions, though these patterns were already established through t-tests. More notably, Fig. 2 reveals consistently greater standard deviations across all eight constructs among expatriate teachers, especially for normative commitment to the organization (NCO), continuance commitment to the profession (CCP), and both turnover intention scales. This increased variability suggests more heterogeneous attitudes toward commitment and quitting across institutional and personal contexts in expatriate ELT. Together, these figures underscore the context-dependent nature of occupational commitment, as well as the challenges of generalizing across diverse employment conditions.
Figure 1. Comparison of means for organization commitment, occupational commitment, and intention to quit scales in the replication and original study. Generated with ChatGPT4o from tabulated data.
Figure 2. Comparison of standard deviations for organization commitment, occupational commitment, and intention to quit scales in the replication and original study. Generated with ChatGPT4o from tabulated data.
Next, Hypothesis 5 proposed that English teachers working in universities and colleges would report higher affective commitment to both their organizations and the profession compared to those in public schools and private academies, based on the assumption that university positions typically offer more favorable conditions. To test this, two one-way ANOVAs were conducted to compare ACO and ACP across the three workplace types. Results indicated no statistically significant differences in ACO (F(2, 70.5) = 0.76, p = .472) or ACP (F(2, 70.0) = 0.47, p = .626) between these three groups of expatriate English teachers, leading to a rejection of Hypothesis 5. These results suggest that institutional setting alone may not meaningfully influence affective commitment levels among expatriate teachers in East Asia.
Hypothesis 6 predicted that expatriate English teachers with formal teacher qualifications from their home countries would report higher levels of affective and normative commitment, and lower levels of continuance commitment, than those without such credentials. This expectation was based on prior findings suggesting that teacher certifications are associated with more favorable commitment profiles (Moodie, Reference Moodie2023). To test this, independent samples t-tests were conducted for each of the six commitment dimensions, comparing teachers with and without qualifications. None of the differences reached statistical significance, leading to a rejection of Hypothesis 6 (see Table 8). These null findings suggest that possession of teaching credentials alone may not reliably predict commitment levels among expatriate teachers in the region, at least when examined across a broader and more diverse sample.
Table 8. Independent samples t-test results for hypothesis 6
5. Discussion
To begin, the descriptive results showed that the reliability statistics in both studies were comparable, indicating that the measures of commitment and turnover intentions had strong to satisfactory internal consistency. While the general correlation patterns were similar, two figures substantively differed between the replication and the original (McInerney et al., Reference McInerney, Ganotice, King, Marsh and Morin2015). First, the correlation (r) between affective (ACP) and normative commitment to the profession (NCP) was lower (−.20) in the replication, while the correlation between affective (ACO) and normative commitment to the organization (NCO) was considerably higher (+.34). These differences are somewhat surprising, and are likely due to complex factors such as differences in sample characteristics, contextual differences, and measurement reliability.
The regression analyses indicated that McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) results were broadly replicated, as indicated by the statistical significance, directionality, and magnitude of effects. As with Hong Kong schoolteachers in the original, the 3 × 2 commitment model significantly predicted turnover intentions among expatriate NESTs in Asia, with effect sizes exceeding both the original study and the AL median benchmark of R2 = .32 (Plonsky & Ghanbar, Reference Plonsky and Ghanbar2018). In the replication, commitment explained 59.7% of the variance in intentions to quit the profession of ELT and 51.8% of the variance in intentions to quit schools. These effects were also stronger than in a recent study on occupational well-being, which explained 36.9% of the variance in both outcomes (Moodie, Reference Moodie2024). These findings emphasize the general predictive power of commitment on teacher turnover and attrition intentions, providing a contribution to language teaching literature.
The hypothesis testing contributes further understanding about how commitment mindsets differ between expatriate ELT professionals and public school teachers, and also about the relationship of commitment and turnover intentions among expatriate English teachers. Hypothesis 1 was supported, reaffirming the negative relationship between affective commitment and turnover intentions. A one standard deviation increase in ACO was associated with a .531 decrease in intentions to quit schools, and a one standard deviation increase in ACP predicted a .509 decrease in intentions to quit the profession. These results confirm the central role of affective commitment in predicting attrition across teaching contexts (Meyer et al., Reference Meyer, Stanley, Jackson, McInnis, Maltin and Sheppard2012; Moodie, Reference Moodie2023; Moodie & Meerhoff, Reference Moodie and Meerhoff2020).
A key difference with McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015) was the reversal in the relative predictive strength of ACO and ACP. In the original study, ACO was the strongest predictor of turnover intentions to both schools and the profession. In contrast, in the replication, ACP was the strongest predictor of intentions to quit the profession, while ACO remained the dominant predictor of school-level turnover. This reversal likely reflects structural differences between the research contexts. For teachers in Hong Kong, many of whom hold stable, long-term civil service positions, quitting a school may be functionally equivalent to quitting the profession. In contrast, for expatriate English teachers, who generally have shorter-term contracts and work in more mobile, transient contexts, quitting a school and quitting the profession are more distinct decisions.
Hypotheses 2 and 3, which posited negative and positive relationships between turnover intentions and normative and continuance commitment respectively, were not supported. The rejection of these hypotheses reinforces the primacy of affective over other types for predicting turnover and attrition, indicated in both the replication and original.
Group comparisons confirmed that expatriate English teachers reported lower commitment and higher turnover intentions than Hong Kong teachers in the original study (Hypothesis 4). The largest differences were observed in continuance and normative commitment to the profession (NCP and CCP), while turnover intention differences, though statistically significant, showed smaller effect sizes. These results likely reflect structural and cultural differences between the two contexts. Becoming a teacher in Hong Kong requires official credentials, including education degrees and teacher licenses. It is also a relatively high-status position with good benefits (McInerney et al., Reference McInerney, Ganotice, King, Marsh and Morin2015). Expatriate NESTs, on the other hand, have relatively lower status and barriers for entering the field. The higher level of continuance commitment to the profession among Hong Kong teachers is likely attributed to the fact that they perceive relatively high costs associated with leaving the profession and would have a harder time finding a job with similar benefits compared to expatriate NESTs. The relatively higher levels of normative commitment are likely in part due to culture differences. Research indicates that normative commitment tends to be slightly higher in Asian contexts than in Western contexts, which is at least partially explained by the former tending to be more collectivist in nature and the latter more individualistic (Meyer et al., Reference Meyer, Stanley, Jackson, McInnis, Maltin and Sheppard2012). Additionally, teaching is associated with normative mindsets, with research indicating that many teachers enter the field wanting to serve their students and communities (Day & Gu, Reference Day and Gu2007; McInerney et al., Reference McInerney, Ganotice, King, Marsh and Morin2015; Morin et al., Reference Morin, Meyer, McInerney, Marsh and Ganotice2015). This, along with the cultural effect, likely explains the higher levels of normative commitment among Hong Kong teachers than the expatriate English teachers in the present study.
Lastly, although Hypotheses 5 and 6 were not supported, the results remain informative. Because of assumed advantages regarding status and working conditions, Hypothesis 5 predicted that university teachers would show higher affective commitment than those in schools or private academies. ANOVA results showed no statistically significant differences, suggesting that workplace type alone does not meaningfully shape affective commitment. Similarly, based on a prior study (Moodie, Reference Moodie2023), Hypothesis 6 expected higher affective and normative commitment among NESTs with formal teacher qualifications. These results did not replicate, indicating that teacher qualifications do not necessarily translate into stronger commitment profiles among expatriate NESTs.
5.1. Implications
The clearest implication of this study is the central role of commitment, particularly affective commitment, in predicting turnover and attrition among language teachers. Not only were the general results of McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015) replicated, but the predictive power of commitment was even greater in this replication with this sample of expatriate NESTs. The results lend theoretical support to the enduring relevance of affective commitment for professional retention highlighted in education and I/O psychology research.
As for practical implications, the results suggest that language teacher recruitment and retention policies should prioritize affective commitment indicators. Hiring committees might benefit from evaluating applicants’ emotional investment in the profession and alignment with institutional values. In workplaces, administrators should also foster affective commitment among faculty, such as by creating conditions where they have a sense of belonging and agency within their organizations and are able to make meaningful contributions, for example, through leadership roles and curriculum design. In terms of professional development, workshops or mentoring programs aimed at developing teachers’ emotional attachment to their profession and organizations could enhance affective commitment among language teachers.
For current and future language teachers, the results suggest that affective commitment is not only beneficial for job satisfaction but may also play a role in career stability and progression. Thus, teachers should seek ways to increase their affective commitment to the profession and find organizations in which they feel a sense of belonging.
5.2. Limitations and future research
Limitations include a smaller sample size than the original study and reliance on convenience sampling, which was heavily skewed toward expatriate teachers in Korea. These limitations likely contributed to the lower CFA results for the measures in this replication compared to the original. Although the general results of McInerney et al. (Reference McInerney, Ganotice, King, Marsh and Morin2015) were confirmed, additional replications are warranted that include a larger sample across a wider range of ELT contexts.
Thus, future replications should include broader, geographically diverse samples to improve generalizability. Additionally, repeated-measures research could employ longitudinal designs to track changes in commitment and turnover, lending insights into how they change over time and circumstance. Intervention studies aimed at strengthening affective commitment are also warranted, which could investigate professional development programs aimed at strengthening teachers’ attachment and sense of belonging to organizations and the profession. Lastly, future work should explore antecedents of commitment itself, including workplace conditions, motivations for teaching, and personality traits, to better understand how strong commitment profiles develop and can be supported.
6. Conclusion
By replicating McInerney et al.’s (Reference McInerney, Ganotice, King, Marsh and Morin2015) research, this study affirms the central role of occupational and organizational commitment in predicting turnover intentions. The results underscore the particular importance of affective commitment in reducing intentions to quit both schools and the profession among expatriate English teachers in Asia. Although limited by non-random sampling and a bias towards participants teaching in Korea, the study nonetheless offers valuable insights into commitment and turnover intentions in transient and transnational language teaching contexts. The implications extend beyond theory, pointing to the importance of fostering affective commitment as a means of enhancing teacher retention. Future research should aim to replicate these findings with more representative samples across diverse contexts, and intervention studies are warranted to explore how organizational and occupational commitment can be effectively developed and sustained among language teachers.
Ian Moodie is a tenured professor in the Department of English Education at Mokpo National University, South Korea. His recent research has focused on language teacher commitment, attrition, and professionalism, with recent publications in Language Teaching, Language Teaching Research, TESOL Quarterly, System, and ELT Journal, among others. He currently serves as chair of the KOTESOL Research Committee and sits on multiple editorial boards.
Open access
