Intention to leave is defined as an employee’s intent or decision to leave an organisation and seek employment elsewhere which often leads to actual turnover behaviour (Janse Van Rensburg et al. 2017). Researchers in the field of employee retention studies (Grobler & Grobler 2016; Kakar, Sofi & Mansor 2019) provide strong empirical evidence which supports the negative relationship between POF and ITL. This negative relationship implies that when the individual experiences greater fit between themselves and the organisation, the ITL the organisation decreases (Kakar et al. 2019; Kooij & Boon 2018). According to the Person–Organisation fit theory posited by Cable and DeRue (2002), the extent to which an employee is considered to be compatible with the organisation is determined by three dimensions that is value– congruence fit (VCF), needs–supplies fit (NSF) and demands – abilities fit (DAF).

Value congruence refers to the employee’s perceptions or beliefs about the congruence between their personal values and culture of the organisation (Cable & De Rue 2002). When the individual believes that their values are compatible or supported by the organisational culture, the employee is likely to have a low desire to leave the organisation. Jin et al. (2018) assert that when an employee’s values and beliefs are aligned with the organisation, there are positive psychological associations that are created between the individual and the organisation which leads to reduced intention to leave. The second dimension of POF considers the NSF which represents an employee’s judgements regarding whether the rewards they receive (pay, benefits, training) in return for their service are aligned with their personal and professional needs. Employees who believe that their needs are not met by their employer are more likely to have a negative association with the organisation which fosters high- turnover intention. DAF which represents the third dimension of POF, refers to a person’s perceptions of the congruence between the demands of their job and their abilities. De Crom and Rothmann (2018), argue that when the employee’s work roles match their abilities and self-concept, they are actively engaged in their jobs and find new purpose in their work which is related to high job performance. On the contrary, when an employee possesses inadequate abilities or their abilities are insufficient for the job, work processes are less efficient and work outputs are of a poor quality. These employees are likely to experience high levels of work-related stress, which exacerbates the ITL.

The three dimensions of POF mentioned above are directly related to turnover intention and these relationships are supported by the Job Embeddedness Theory (JET) (Holtom, Mitchel & Lee 2006; Ngo-Henha 2018). According to the JET, when employees believe that their values are congruent with the organisation, their needs are addressed and their abilities are compatible with their work roles, they settle in their jobs and organisations. Once people make linkages and feel a sense of being settled in their job, organisational fit is experienced and they find it difficult to leave their job or organisation for a new one as this will remove them from their professional and social environment (Holtom & Darabi 2018; Ngo-Henha 2018). Based on the argument presented above, the following hypothesis is proposed:

Research conducted in the field of organisational psychology has documented a consistent negative relationship between JS and ITL (Kristof-Brown, Zimmerman & Johnson 2005; Naidoo 2018; Wheeler et al. 2007; Zhang et al. 2017). The direction of this relationship implies that when people perceive themselves as being satisfied at work, they do not think of leaving their organisation, while those who perceive themselves as dissatisfied consider leaving their organisation to look for work elsewhere (Buitendach & Rothmann 2009; Diedericks & Rothmann 2014). In conceptualising JS, authors have identified two dimensions of JS that is intrinsic and extrinsic JS, which explains why some people may be satisfied with their jobs while other are not (Phuong et al. 2018; Radebe 2018).

Intrinsic JS is recognised as emanating from within the individual as it emphasises psychological value and has subjective meaning. The subjective meaning and psychological value experienced by individuals are gained from the job itself and occur when individuals feel a sense of achievement, recognition, and self-actualisation from the work that they do (Chang et al. 2021). When employees believe that the job meets their emotional, psychological and subjective needs, they experience a holistic form of work-related wellbeing and are less likely to consider leaving the organisation. Similarly, when employees feel demotivated, unfulfilled and no longer value their work, they may consider leaving the organisation to find work that gives them a sense of purpose. This ITL the organisation is the final cognitive step of voluntary turnover and spurs on behavioural action to actually leave the organisation (Beheiri & Ahmed 2018).

The second dimension of JS is extrinsic or external in nature. Unlike intrinsic JS, extrinsic satisfaction stems from the work environment or factors external to the employee such as financial rewards, promotions, opportunities for career development and leadership. Phuong et al. (2018) argue that while some employees may feel the need for psychological fulfilment from their work, most employees tend to be extrinsically satisfied. This can be attributed to the fact that people primarily join an organisation to be compensated for their services and that they do seek career progress. As a result, financial compensation and opportunity for promotion amongst other external factors could play an important role in whether employees feel that their needs are being met by the organisation. An employee that identifies an incongruence between what they offer to the organisation and what they receive in return is likely to contemplate whether it is in their best interest to remain in the organisation. The relationship between JS and ITL is supported by the value-perception theory which postulates that the extent to which an employee feels satisfied with the job is dependent on their belief that the job supplies them with things that they value (Ekpendu, Egbuta & Ikechi-Ekpendu 2019). When employees perceive the employer’s contribution towards them as less valuable, they begin to experience job dissatisfaction, which results in ITL, eventually leading to actual turnover behaviour. Thus, the following hypothesis is proposed:

Given the effective role of POF in directly mitigating turnover intention, researchers (Jin et al. 2018) have argued that understanding the indirect processes through which POF might explain ITL, may offer additional insight on the topic of employee turnover. As a result, this study aimed to understand the role of JS as an explanatory psychological mechanism in the relationship between POF and ITL. It has been established in the literature that value –congruence need– supplies fit and fit between the individual abilities and demands on the job may all directly explain an employee’s purposeful decision to leave the organisation (Redelinghuys & Botha 2016). However, there is evidence which suggests that these dimensions of POF may explain how employees experience intrinsic and extrinsic JS (Redelinghuys & Botha 2016). Lack of congruence between the employee’s and organisation’s values (VCF) and beliefs, is expected to have a negative effect on factors relating to intrinsic JS such as achievement, recognition, responsibility, personal meaning etc. When there is a mismatch between employees’ values and the organisation, they are likely to be less fulfilled and disengaged from their jobs, leading to low JS. Secondly, when an employee perceives low fit between their needs and the organisation’s attempts to meet those needs (Needs–Supplies Fit), both intrinsic and extrinsic JS begins to decrease. When the employee believes that the organisation is unable to meet their external work-related needs such as financial compensation, promotions, supportive leadership etc., the employees begins to experience low extrinsic JS. Additionally, when the employee believes that subjective needs such as the need for fulfilling work, autonomy, sense of purpose, etc. are not being met by the organisation, intrinsic JS begins to decrease. Low intrinsic and extrinsic JS lead to negative job behaviours such a disengagement which may trigger negative thoughts, relating to leaving the organisation. Finally, when employees identify a discrepancy between their skills and abilities and the demands of the job (Demands–abilities fit), intrinsic needs related to achievement, self-respect and responsibility remain unfulfilled. Peng and Mao (2015) added that employees whose abilities do not meet the demands of their jobs, lack confidence, perform poorly at their jobs, experience increased stress and are unhappy in the jobs. The negative experiences result in low levels of JS which will spur on action to leave the organisation. Thus, the following hypothesis is proposed:

The study population consisted of all staff working in the South African SDS. Using a non-probability purposive sampling design, a total of 119 responses were received using an online survey system that is Qualtrics. From the 119 responses, a total of 100 responses (n = 100) were retained as these surveys were fully completed. The selection criteria for the study included employees in the SDS in Gauteng, South Africa and comprised employees across all age groups, genders, occupational groups and occupational levels that is from operational to executive management. Since the aim of the study was to investigate ITL in the SDS, the use of a purposive sampling method allowed the researcher to find relevant people who could provide the necessary information (Saunders, Lewis & Thornhill 2003).

The demographic breakdown of the sample can be found in Table 1. Respondents were primarily women (61.6%), and more than half of the respondents were married (61.6%), which also corresponded with the age groups that is most of the participants were aged between 40 and 55 years (59.60%). In the current sample 71.7% of respondents had a postgraduate qualification that is 4 years and above, while more than half of the respondents indicated that they had been employed in the SDS for 10 years or more (62.2%). A total of 43.3% of the respondents were employed at middle-management level, 23.2% at operational level and 18% at senior- management level. About 30.3% of the respondents indicated that they would like to remain with the current organisation for more than 5 years, while 26.3% indicated that would like to remain with the organisation for less than 3 months.

The study received ethical clearance from a large Gauteng-based public university while gatekeeper permission was received from the National Department of social development in Gauteng. An online survey, consisting of three self-report measures and a biographical questionnaire, was circulated via the employee email system. A link to the questionnaire, the letter of informed consent and the participant information were included in the email communication to all staff members. The completed questionnaires were automatically sent to the Qualtrics online survey system. The researcher did not have access to the participants’ email addresses or personal information during the data collection, while the data collection process was managed by a contact person in the Department of Social Development.

The data analysis was conducted with the Statistical Package for the Social Sciences (SPSS) version 27. Before analysing the data, as related to the proposed hypotheses, the researchers tested for common method bias considering that the instruments used in the study were self-report measures; and the data were collected at one point in time. Cronbach’s alpha coefficients were calculated to assess the reliability of the measuring instruments used in the current study. Pallant’s (2013) guideline of α ≥ 0.70 was used to determine the reliability of the measuring instruments although values of α > 0.80 are preferable. Descriptive and inferential statistics were also used to analyse the sample and responses to the scales. Pearson’s Product Moment Correlation analysis were used to assess the relationships between POF, JS and ITL. Cut-off points of 0.30 and above (medium effect) and 0.50 and above (large effect), were set for the practical significance of the correlation coefficients (Pallant 2013). Hayes’ PROCESS macro model 4 (Hayes 2018) was used to assess the mediating effect of JS on the relationship between POF and ITL.

The authors obtained ethical clearance from the Department of Human Resource Management at the University of Pretoria prior to data Collection. Ethical Clearance Number: MMD/2020/14.

In addition, gatekeeper permission was obtained from the Department of social development.

The author assessed the data to ensure that it was distributed normally and to identify and address any outliers. First, univariate skewness and kurtosis were assessed using SPSS. Based on the research done by Finch and West (1997) and Field (2009), the criteria used to determine whether the data assumed a normal distribution were set at 2.00 for skewness and 4.00 for kurtosis. Both skewness and kurtosis ranged between −0.093 and −1.020, indicating that the data were normally distributed. To assess, multivariate normality of data, Mardia’s kurtosis co-efficient was calculated. The Mardia’s skewness test statistic was 13.28 (p > 0.0001), and the Mardia’s kurtosis test statistic was 73.29 (p > 0.0001), which confirmed that the data met the criterion of multivariate normality.

To assess the presence of multivariate outliers, the Mahalanobis distances were inspected. The assessment revealed one outlier. It was found that while there were slight variations in the co-efficients when the outlier was removed, these variations did not affect whether the proposed hypotheses were accepted or rejected. Additionally, the outlier did not affect the assumptions of normality hence the outlier was retained in the analysis. Several authors (Boguslavsky, 2021; Tenopir et al. 2016; Yang & Berdine, 2016) recommend keeping the outliers in data analysis. The argument is that on some occasions the outliers might simply be valid extreme observations due to random variability and reflect the inherent property of random sampling. The identified outlier was assessed to determine if it was a result of error in data entry or a genuine valid extreme observation in the sample. Because the outlier did not appear to be a result of error in data entry it was retained in the analysis.

While the possibility of common method bias in this study was acknowledged, researchers (Fuller et al. 2016) argue that the concern of common method bias in self-report measures that have proven construct validity is questionable. All measures used in the current study that is TIS-6; The MSQ-20 and the POF Scale are validated measures, which reduces the possibility of method bias. Additionally, the concern of method bias was further minimised as the researchers found that most of the reported correlations found in this study were in line with previous published findings indicating no significant discrepancy.

Table 2 shows the results obtained from descriptive statistics, reliability analysis and Pearson’s product moment correlation analysis. The results revealed that this sample of employees in the SDS reported moderate levels of ITL (M = 3.25; SD = 0.95), moderate levels of POF (M = 3.38; SD = 0.81) and moderate levels of POF sub-dimensions; VCF (M = 3.30; SD = 0.98), NSF (M = 3.11; SD = 1.01) and DAF (M = 3.70; SD = 1.03), with the mean score for DAF being higher than the other two POF dimensions. Results also indicated moderate levels of JS (M = 3.35; SD = 0.76) with moderate levels of intrinsic satisfaction (M = 3.46; SD = 0.79) and extrinsic satisfaction (M = 3.23; SD = 0.80). The mean scores for intrinsic satisfaction was slightly higher than extrinsic satisfaction. It can be concluded that this sample of SDS employees experience moderate levels of ITL, POF and JS. All scales and sub-scales reported reliability co-efficients above 0.70, indicating an acceptable internal consistency of all the measuring instruments used in this study (see Table 2).

Pearson’s product moment correlation analysis was used to test Hypotheses 1 and 2 proposed in this study. As indicated in the correlation analysis shown in Table 2, POF reported a significant negative relationship with ITL (r = –0.65; p < 0.01) (large effect). Person–organisation fit sub-dimensions also reported significant negative relationships with ITL; VCF (r = –0.388; p < 0.01) (medium effect), NSF (r = –0.733; p < 0.01) (large effect) and DAF (r = –0.46; p < 0.01) (medium effect). It appears that in this sample, needs– supply fit reported the strongest relationship with ITL. Results of the correlation co-efficients indicated that POF and all its sub-dimensions reported a negative relationship with ITL, providing support for Hypothesis 1 that is POF is negatively associated with ITL.

Results found in Table 2 indicate that JS reported a significant negative relationship with ITL (r = –0.747; p < 0.01) (large effect). The results also indicated towards significant negative relationships with large effects between JS sub-dimensions and ITL; intrinsic JS (r = –0.697; p < 0.01) and extrinsic JS (r = –0.722; p < 0.01). Results of the correlation analysis indicated that JS and all it sub-dimensions reported a negative relationship with ITL, providing support for Hypothesis 2 that is JS is negatively associated with ITL.

In order to assess Hypothesis 3 that is JS will mediate the relationship between POF and ITL, all the pathways of the proposed mediation model (Figure 1) were assessed using mediation analysis (Hayes 2018). While the Minnesota JS questionnaire has both a unidimensional- and multidimensional- factor structure, the authors opted to use the unidimensional factor structure for the mediation analysis as this is the common practice observed in more than 30 years of JS research (Hirschfeld 2000). Additionally, in a study analysing the factor structure of the Minnesota JS Questionnaire, Boshoff and Hoole (1998) concluded that it is acceptable to use the instrument as a unidimensional measure. Secondly, while the POF Scale has been known for its three-dimensional factor solution, authors (Chen, Sparrow & Cooper 2016; Grobler 2016) have used the scale as a unidimensional measure. Additionally, Grobler (2016) argues that the scores of each dimension can be combined to provide an indication of overall person–environment fit in which a high score suggests a high level of POF in all three dimensions. This indicates that a unidimensional-factor structure is also acceptable as was used in the current study.

In this study, POF was indicated as the independent variable (X); ITL was indicated as the dependant variable (ITL) and JS was indicated as the mediator (JS). In order to run the mediation analysis, the bootstrapping method with bias-corrected confidence estimates were used (Hayes 2018). Using 5000 bootstrap resamples, a 95% confidence interval (CI) of the indirect effect was obtained. The 95% confidence levels indicate that an indirect effect is significant (at alpha = 0.05) and meaningful if zero does not fall within its 95% confidence interval (Pillay 2020). Results of the mediation analysis are presented in Table 3 above.

The first step in assessing the mediation effect was to consider the direct effects proposed in the current study. The mediation analysis confirmed a direct effect between POF and ITL (b = 0.35, SE = 0.10, LLCI = −0.54 and ULCI = −0.16). The mediation analysis also indicated a significant direct effect between POF and JS (b = 0.60, SE = 0.07, LLCI = 0.45 and ULCI = 0.74), and finally the mediation analysis indicated a significant direct effect between JS and ITL (b = −0.70, SE = 0.10, LLCI = –0.90 and ULCI = –0.49). Once the direct effects as per the pathways were established, attempts were made to analyse the indirect effects in which POF influences ITL through its influence on JS.

The total indirect effect (b = −0.42, SE = 0.08, LLCI = −0.58 and ULCI = −0.25) as indicated in Table 3 above, was significant as zero did fall in between the confidence intervals. This provided support for Hypothesis 3 that is JS mediates the relationship between POF and ITL. It must be noted that the relationship between POF and ITL remained significant in the presence of JS which indicates that POF has a negative association with ITL independent of the indirect effect through JS. This suggests a partial mediation effect. However, it appears that the size of the indirect effect of POF on ITL, through JS, is slightly larger than the direct effect of POF on ITL on its own. Results may indicate that JS as an explanatory mechanism, adds significant value to the relationship between POF and ITL and that POF may account for more variance in ITL when JS is present in the relationship. The results for the mediation model are shown in Figure 1.