Person-vocation and same organization bridge employment intention: Testing the mediating and moderating roles of person-organization fit

2.1 Person-environment fit, continuity and role theories

Continuity theory [22] highlights the different continuity strategies older workers use when facing adaptive situations. More specifically, it suggests that older workers adopt strategies according to their internal and external structures to avoid the experience of disruptive events while aging. According to Lieberman & Tobin [23], continuity strategies can be achieved at two levels. Internally, when the aging person adopts a pattern of behavior allowing him/her to preserve the stability of the inner structure (e.g., values, emotions). Externally, when the aging person successfully preserves his/her usual physical and social environment. Concerning retirement, it has been assumed in previous works (e.g. [24, 25]) that a good work environment can encourage older workers to hold bridge employment positions to act in continuity with their past experiences.

For its part, role theory [26] emphasizes the importance of role identity during role transition. According to role theory, role transitions are affected by the involvement in a specific role (e.g., organizational role). The more the person is involved, the more difficult it is for him/her to give up the given role. In other words, the centrality of work-related identities (e.g., vocational, organizational) may encourage older workers to pursue professional activities once retired (i.e., bridge employment) to avoid the stress caused by the role-identity loss [27]. Evidence from retirement literature confirms the relevance of role theory as a theoretical framework to study retirement intention and behaviors. Indeed, by mobilizing role theory, it has been shown that older employees who have a strong commitment toward their career or organization [3] or perceive a high level of fit with their work environment would be more inclined to pursue bridge employment.

Additionally, P-E fit refers to the existing compatibility between the individual and his/her work environment [28]. P-E fit can be categorized in two ways. First, the supplementary kind of P-E fit reflects the similarity between the person and the work environment (e.g., values congruence at the vocational level). Second, the complementary P-E fit is related to situations where individual attributes are complemented by the work environment (e.g., individual needs are satisfied by the organizational resources) and vice versa.

In the organizational/vocational psychology literature, it has been shown that the perception of fit with the work environment is positively related to many attitudes and behaviors at the workplace (e.g., job satisfaction: [29]; organizational citizenship behavior: [30]; organizational commitment: [31]). Moreover, some studies have explored the direct effect of P-E fit levels on different retirement intentions [15, 16]. Nevertheless, understanding the impact of the P-E fit levels on retirement intentions is still at its beginning, and many questions regarding the mediating and moderating variables of the relationship are still unanswered [15]. Therefore, based on the limits of the previous studies, and by mobilizing a theoretical framework (i.e., continuity and role theories) that has been widely used in the retirement literature [15, 32], we formulate a set of hypotheses that aim to explain how P-E fit levels interact to explain the intention to pursue bridge employment within the same organization.

2.2 Person-vocation fit and the intention to pursue bridge employment within the same organization

Person-Vocation fit (hereafter P-V fit) is conceptualized as a value congruence, which occurs when the employee perceives a match between his/her values and the vocational ones [33]. At this level, it seems essential to distinguish between the P-V fit and the somewhat related Person-Job fit (hereafter P-J fit). Indeed, contrary to P-J fit, which is associated with the hierarchical position, P-V fit occurs when the professional trajectory fits with the personal interests, aspirations, and values. Here, we suggest that P-V fit is positively related to the intention to pursue bridge employment within the same organization.

Indeed, Feldman [34] suggested that employees who identify themselves with a given profession are more likely to hold bridge employment than to retire fully. Previous works showed that older workers who perceive work as meaningful prefer to postpone retirement [35] and that P-V fit is positively related to career bridge employment [15].

According to both role theory [26] and continuity theory [22], we consider that when employees perceive a high level of fit with their vocation, the corresponding role identity becomes central in the older worker’s global identity. Hence, to ensure the continuity of their internal structure (i.e., valued vocational role), the older workers would be more inclined to adopt roles that allow them to ensure internal continuity. Based on the arguments discussed above, we formulate the following hypothesis:

H1. P-V fit is positively related to the intention to pursue bridge employment within the same organization.

2.3 Person-vocation fit and the intention to pursue bridge employment within the same organization: The mediating and moderator role of person-organization fit

P-O fit occurs when a similarity or a complementary between the individual and the organizational characteristics exists [33]. Indeed, the P-O fit could be complementary or supplementary [28]. Complementary fit occurs when the characteristics of the person and those of the organization complement each other (e.g., Individual needs, Organization supplies) [28]. Supplementary fit reflects the degree of similarity between the person and the environment characteristics (e.g., values, personalities, goals) [19]. Further, several studies have shown that the P-O fit perception reinforces the positive attitudes toward the employing organization (e.g., organizational commitment: [19, 36]; intention to leave: [37])

Evidence from the P-E fit literature shows that the different levels are conceptually distinct and related to the workplace attitudes and behaviors [28, 30]. Nonetheless, the different fit levels may relate to each other to predict the employees’ intention to hold bridge employment once retired. Indeed, according to Feldman & Ng [38], P-V fit is an important antecedent of P-O fit. The rationale is that employees who do not perceive a congruence between their values and the vocational ones may not perceive a high level of fit at the organizational level [19, 39]. For example, a faculty member who does not perceive a value congruence with his/her vocation may struggle to achieve a high level of fit at the organizational levels no matter where he or she works.

Further, P-V fit may affect how people perceive P-O fit [19]. For instance, even if the different needs are satisfied by the organization (i.e., P-O fit Needs-Supplies), it would be hard to imagine the individual persisting in a career path if he or she perceives that he does not fit anymore with its values. Only one study has explored P-V fit as a predictor of P-O fit to our best knowledge [19].

In parallel, in this study, we argue that P-O fit is positively related to the intention to pursue bridge employment in the same organization. Indeed, empirical evidence shows that when employees perceive that their employing organizations satisfy their different needs (e.g., developmental opportunities: [40, 41]), they are more likely to keep working during retirement. In addition to the satisfaction of the different needs, values congruence between the older worker and the organization is taken into account while considering retirement decisions. For example, Boumans, De Jong, & Vanderlinden [42] demonstrate that older workers evolving in organizational contexts marked by stereotyped visions (e.g., lack of performance and flexibility, reluctant to change) are more likely to retire.

From a theoretical point of view, both continuity [22] and role [26] theories support the hypothesized relationships between P-O fit and the intention to pursue bridge employment in the same organization. According to continuity theory [22], bridge employment jobs offer the best continuity strategy to preserve satisfying environments (i.e., Needs-Supplies fit). Furthermore, in line with role theory [26], the match between an individual and a specific aspect of the environment makes the corresponding role identity central in the global identity of older workers; bridge employment hence offers a way to preserve a central and valued role identity [43]. On those premises, we formulate the following hypothesis:

H2. P-O fit mediates the positive relationship between P-V fit and the intention to pursue bridge employment within the same organization.

In the current study, it is assumed that the relationship between P-V fit and the intention to pursue bridge employment in the same organization would be strengthened by the perception of P-O fit. Indeed, previous studies have demonstrated that the perceived fit with the organization may reinforce the employee’s relationship with his/her organization. For instance, it has been demonstrated that P-O fit is positively related to positive organizational attitudes (e.g., organizational commitment: [28, 44]; organizational identification: [45]) and negatively to turnover intention [21].

According to role theory [26], an older worker who already perceives a high level of P-V fit may be more inclined to pursue bridge employment within the same organization when he/she perceives a high level of P-O fit. In fact, in this case, in addition to the vocational identity, the older worker may also value the organizational one and would adopt strategies that allow the maintenance of both identities. Thus, the intention to pursue bridge employment within the same organization would be reinforced as it enables preserving two-valued identities, namely the vocational and the organizational ones.

From a continuity perspective [22], an older worker who perceives only a high level of P-V fit may ensure continuity by pursuing bridge employment within or outside the organization (i.e., career bridge employment). Nevertheless, when the perception of P-V fit is combined with the perception of P-O fit, the intention to pursue bridge employment within the same organization would become a better strategy as it offers the most suitable option to ensure continuity. Based on this discussion, the current study argues that P-O fit plays a moderation role in the relationship between P-V fit and the intention to pursue bridge employment within the same organization. As such, the following hypothesis will be tested:

H3. The relationship between P-V fit and the intention to pursue bridge employment is moderated by P-O fit, such that the positive relationship between them will be stronger for older workers who are high in P-O fit.

The conceptual model is presented in Fig. 1.

OPEN IN VIEWER

Fig. 1

Conceptual model.

3.1 Sample and procedure

As the retirement anticipation phase starts by 50 [47], data were collected exclusively from executive employees aged 50 and above. This choice is in line with previous research in the retirement literature (e.g. [13, 48]).

Concerning the sampling strategy, the convenience sample technique has been used. The reason behind this choice is the difficulty to choose randomly from the entire population (i.e., executives aged 50 and above) and the “convenient” character (e.g., easy accessibility, the willingness of the two companies to participate in the study) for the research team to collect data from the banking sector. This technique is widely used in social sciences and business research [49].

For the current research, data was collected from two different banks in France. Before starting data collection, we first introduced the main objectives to the principal HR manager, who subsequently briefed all the potential participants. Further, the data collection was completed in two different waves. At T1, employees were asked about their perception of P.E fit levels and demographic information. Two weeks later (second wave) T2, the employees who filled the survey at T1 were asked about their intention to pursue bridge employment in the same organization. We assigned codes to the participants in the different waves of data collection to match responses over time (T1 and T2). Specifically, we distributed 370 paper-pencil questionnaires to the respondents at T1, and 296 participants responded (the response rate = 80%).

Similarly, after two weeks in T2, we distributed 296 questionnaires, and a total of 258 employees responded (70% of the retention rate from the initial sample). However, 12 responses were deleted due to the missing data and the presence of multivariate outliers. Data was gathered with the permission of top management. However, we have assured anonymity to the participants to obtain their confidence. In the final sample, 73.5% of the participants were aged between 50 and 60 by the time they participated in the study, over 77% were men, and 88% had at least a bachelor’s degree. Further, 88% were married, and 51% declared having more than ten years of tenure within the same company.

3.2 Measures

In this study, existing multi-items adapted from previous research were used. When needed, translation to French has been accomplished according to back-translation technique recommendations [50]. The respondents were asked to rate their agreement level using a Likert scale, with responses ranging from 1 = strongly disagree to 7 = strongly agree. P-V fit was measured with the scales developed by Kennedy [51].

Regarding the P-O fit, we used respectively the measurement scales developed by Kennedy [51] and Cable & DeRue [30] to measure Values congruences’ and Need-Supplies’ dimensions. Finally, the intention to pursue bridge employment within the same organization was measured using Jones & McIntosh’s [3] measurement scales. The internal consistency was estimated using Cronbach’s alpha and AVE (Table 1).

Based on previous research, the analyses of this study included many control variables. Adjustment to retirement was measured with the four-item scale developed by Taylor & Shore [52]. Further, health satisfaction was measured with Adams & Beehr’s measurement scale [53]. Lastly, age and salary were also controlled because previous research [54] have shown that such variables are related to the different retirement intentions.

4.1 Psychometric properties of measurement scales

Internal consistency reliability, convergent validity, and discriminant validity were assessed to check if the different reliability and validity requirements were met. As shown in Table 1, first, internal consistency of the different measurement models was established with composite reliability (C.R) ranging from 0,832 to 0,981 (equal or above 0,7: [55]) and Cronbach’s alpha ranging from 0,804 to 0,971. Standardized loadings (lambda) were used to assess indicator reliability, and they were higher than the threshold of 0,70 [55]. Moreover, the convergent validity criterion was met since the Average Variance Extracted (AVE) of all the constructs exceeded the threshold of 0,50 [56].

Finally, many criteria have been used to establish discriminant validity. First, following Fornell & Larcker [56] criteria, the results (Table 2) showed that the square root of AVE for each construct is greater than the variance shared with all the other constructs. Second, Table 3 shows the existence of the discriminant validity as loadings of each construct (in bold and italic) are higher than cross-loadings. Third, to confirm our results, we used HTMT criteria by using the most conservative threshold level of 0,85 [57, 58]. As shown in Table 4, discriminant validity criterion is met since none of the values exceeds 0.85. Consequently, the results obtained indicate that the requirements of validity and reliability have been met.

As demonstrated in Table 5, the model fit was ensured by examining many criteria. First, in terms of predictive power, the portion of the variance explained in the two endogenous variables: P-O fit and the intention to pursue bridge employment in the same organization was 27,5 and 36,7, respectively. According to many scholars, those values are satisfactory [55, 59]. For example, Cohen [59] noted that the values of 0,02, 0,15, and 0,35 reflect small, medium, and significant effects, respectively. Second, the results showed that the model has good predictive relevance since Stone & Geisser’s Q² values were larger than zero for each reflective endogenous latent variable. More specifically, the results on Table 5 show medium predictive relevance (0,02 = small, 0,15 = medium and 0,35 = immense, [53]). Third, the presence of collinearity issues was assessed through tests for Variance Inflation Factor (VIF). Results indicated that the structural model do not present any multicollinearity since all VIF values range between 1,050 to 1,513, which falls below the recommended threshold of 5 [60]. Finally, we used the Standard Root Mean Square Residual (SRMR) indicator to ensure that the conceptual model fits the data. The results show a significant value of 0,078 (Table 5), which is satisfactory since it is less than 0,08 [61].

The hypotheses were tested using 5000 bootstraps. As shown in Table 6, direct and indirect effects estimates, and 95% bias-corrected bootstrapped confidence intervals are provided for path estimates. Hypothesis 1 tested the direct effect of P-V fit on the intention to pursue bridge employment within the same organization. The result of the regression analysis validated the hypothesis (β= 0,253, t = 3,81, [0,127; 0,383], p < 0,001). Concerning the mediation hypothesis, Hypothesis 2 states that P-O fit mediates the relationship between P-V fit and the intention to pursue bridge employment within the same organization. According to the results, the hypothesized relationship was supported (β= 0.176, t = 4,64, 95% CI [0,105, 0,243]). Finally, the impact of P-O fit as moderator on the direct effect of P-V fit on the intention to pursue bridge employment within the same organization was found to be positive and significant (β= 0,110, t = 2,31, 95% CI [0,013, 0,201]) (Hypothesis 3). Figure 2 shows the interaction effect of P-O fit and P-V fit on the intention to pursue bridge employment within the same organization. The graph shows that when P-O fit is high (one standard deviation above the mean), the perception of PV fit has a higher positive effect on the intention to pursue bridge employment within the same organization than when P-O fit is low (one standard deviation below the mean).

OPEN IN VIEWER

Fig. 2

Interaction diagram P-O fit between P-V fit and the intention to pursue bridge employment within the same organization.

The theoretical model is tested using the machine learning classifiers to see the robustness of the proposed theoretical model [62]. Using supervised learning methods, we can establish a classifier algorithm that establishes predictions with suitable accuracy. Individual Likert scale responses are combined depending on the model representation by reducing dimensions [63]. A dimensionality reduction helps in combining dimensions without incurring significant information loss. Owing to the ordinal nature of the data (7 points Likert Scale), we employ a correlation analysis based on the maximum likelihood estimate. A polychoric correlation matrix is layered on, and dimensions are further combined by a Principal component Analysis approach.

Further, the study arrived at a point with exactly four predictors. With the target variable categorized into a binary outcome, the study has set for a classification setup based on a threshold. Four unique classifiers were deployed, in the form of a Support Vector Machine (SVM), Gradient Boosting, Random Forest and Naive Bayes are used. A 90:10 test train split is used, and a 10-fold cross-validation is done to select the best fit and quantify the model in Table 7. Based on the different classifiers, SVM exhibited the best accuracy with minimum error values, as shown in Table 7.

For the proposed model, the prediction of hypotheses was made through training and test of the dataset. Further, to test the model’s ability using K-cross Validation [64]. This test is done to generalize that the model is free from overfitting and selection bias. Table 8 shows that the models have an accuracy of 75%, proving that the model has no underfitting and overfitting [64] as is in the prediction.