1. Introduction
In 2023, a Canadian accountant was ordered to repay their former employer CAD 2600 for time theft tracked while working remotely [
1]. With the COVID-19 pandemic accelerating the shift towards remote work, addressing employees’ time theft has emerged as a focal issue worldwide across industries and in behavioral science research. However, beyond monitoring and punishment, is it possible for leaders to encourage employees to actively reduce time theft?
Time theft refers to employees’ tendency to partake in unauthorized non-work-related activities during paid work hours, including behaviors such as arriving late, leaving early, making personal phone calls, or shopping online during work [
2,
3,
4]. The term ’theft’ is used because employees are compensated for their work hours [
5]. Research on time theft originates from two trajectories. First, as studies on counterproductive behaviors and workplace deviant behaviors evolve, scholars have begun to explore less harmful theft behaviors, including time theft [
6,
7]. Although a minority of studies suggest that a short unobtrusive break can help restore employees’ concentration [
8], organizations typically view time theft as a form of productivity disruption or deviance [
9,
10,
11]. Second, the early 21st-century saw rapid internet growth, enhancing workplace productivity but also sparking concerns over its negative impacts. Scholars from the psychology, management, and computer science fields delved into cyberloafing (also known as cyberslacking or cyberbludging), which involves employees engaging in personal internet-based activities during work hours [
11,
12,
13,
14]. Building on both trajectories, time theft emerges as an independent concept focused on comprehensive non-work-related behaviors rather than web-based off-task behaviors during work hours, potentially including cyberloafing, shirking, job neglect, social loafing, free riding, etc. [
4,
5]. The rise of remote work due to the COVID-19 pandemic has again prompted discussions on how to better manage employees’ time theft.
In general, there are two approaches to reducing time theft: preventive measures beforehand and corrective actions afterward. While punitive measures are often employed by organizations to deter employees from time theft, the effectiveness of this ’cat and mouse game’ is usually seen as unsatisfactory [
5]. This is partly because time theft, unlike other counterproductive behaviors, is hard to detect [
3], especially as the majority of employees disapprove of time theft tracking [
15]. Additionally, strict control from leadership can further lead to employee emotional exhaustion, which in turn exacerbates time theft [
16]. Moreover, in certain instances, addressing time theft can be more time-consuming than ignoring it due to the risk of legal complications [
5]. Another method to reduce time theft is through preventive measures. Existing research has identified massive antecedents of time theft, which can be categorized into two types: one views time theft as a nonaggressive response to work-related antecedents, such as after-hours electronic communication [
17], authoritarian leadership [
16], and workplace ostracism [
14], while the other sees time theft as a purposeful action, e.g., spending work time on personal matters [
18]. Ideally, understanding and intervening in these antecedents can effectively mitigate time theft. However, the complexity of the antecedents makes intervention challenging. For example, both authoritarian and laissez-faire leadership can lead to employee time theft [
16,
19], placing leaders in a dilemma of not being too strict or too lenient. Moreover, the pervasive nature of time theft and the purposeful features of time theft in certain instances illustrate the difficulty of reducing it solely through cause control [
2,
5].
To address the limitations in current research on preventing employee time theft, our study leverages role theory to understand time theft from the perspective of role and explores how to encourage employees to perform their work roles in accordance with organizations’ expectations in the workplace. Role theory suggests that individuals juggle multiple roles [
20], and employees’ time theft can be seen as a conflict between work and non-work roles. Existing studies have argued that the internalization of work role expectations by an employee is influenced not only by the centrality of their work role but also by role ambiguity [
21]. Furthermore, information related to work roles is typically obtained through interactions with others [
20], especially leaders, who are key figures in the workplace with close interactions with employees. In organizations, leaders often have a significant impact on shaping employee behavior through their feedback [
22]. Leaders’ feedback can be classified into close monitoring feedback and developmental feedback, with the latter providing helpful and useful information to employees and assisting in their future learning and growth [
23]. Previous studies indicate that developmental feedback can convey general information to employees beyond specific task information, including relational roles, cultural norms, and both in-role and extra-role expectations [
24], and can additionally enhance employees’ intrinsic motivation by fostering personalized development [
25,
26]. Consequently, leaders’ developmental feedback has the potential to reduce role ambiguity and elevate the centrality of work roles among employees, ultimately mitigating their theft of work time.
To validate whether supervisor developmental feedback can prevent employee time theft, our study proposed a serial multiple mediation model to capture employees’ attitudinal reactions. Specifically, helpful information provided by leaders may enhance employees’ perception of being organizational insiders, subsequently improving their work passion and ultimately reducing time theft. Furthermore, as employees might question leaders’ word–deed inconsistency [
27], our study introduced perceptions of leader hypocrisy as a boundary condition. The moderated serial multiple mediation model was tested in Study 1. Previous scholars have argued that research on the content of supervisor developmental feedback is lacking [
28]. According to role theory [
20,
21], supervisor developmental feedback on different topics for the same behavior intensity may convey different role expectations to employees, thereby impacting employees’ responses. Hence, Study 2 aimed to further explore the topic structure of supervisor developmental feedback and validate the hypothesized model across various developmental feedback topics.
3. Study 1 Quantitative Study to Validate the Serial Multiple Mediation Model
3.1. Participants
Study 1 employed a quantitative research paradigm. A convenience sampling method was adopted to recruit currently employed individuals to participate in the survey. During the questionnaire collection stage, a total of 600 online questionnaires were distributed, of which 594 were returned. After excluding invalid questionnaires with missing data or inconsistent responses, the remaining 402 valid questionnaires were used for further analysis, resulting in a valid response rate of 67.0%. Before participants took the survey, this study provided standardized instructions, making sure to explain why the survey was being conducted and how it would proceed. Additionally, it was highlighted that any information shared by respondents would be confidential and only be used for academic purposes. To acknowledge participation, respondents were provided with an electronic red envelope containing approximately 0.5 US dollars.
The sample’s demographic features include gender, age, education, job type, industry, tenure, and job level. In summary: (a) 64.2% were female and 35.8% male; (b) regarding age, the majority were in the 20–30 age group (51.2%), followed by the 30–40 age group (37.3%); (c) in terms of education, most held a bachelor’s degree (71.6%), with master’s degree holders being the second largest group (13.7%); (d) concerning job type, a significant majority of employees worked in private enterprise (62.2%); (e) for work experience, the most common range was 4 to 6 years (28.6%); (f) in terms of job levels, there were more regular employees (46.3%) and junior managers (22.1%); and (g) in terms of industry, the sample distribution revealed a relatively balanced representation, featuring a significant presence in the Internet, Information Technology, and Gaming sectors (14.9%), Transportation, Logistics, Trade, and Retail sectors (10.7%), and Healthcare sector (10.4%).
3.2. Measurement Tools
To ensure the reliability and validity of the measurement tools, all of the scales used in this study were authoritative mature scales previously validated in the literature. For Chinese respondents, English scales were translated using the “translation–backtranslation” procedure proposed by Brislin to ensure the accuracy of the scales [
52]. All scales used a seven-point Likert scale, with 1 to 7 respectively representing “strongly disagree” to “strongly agree”. A few scales incorporated reversed items to control response style bias. In summary, all scales utilized in this study have been proven to have satisfactory reliability, with Cronbach’s
coefficient consistently exceeding 0.7. For all scale items, see
Appendix A.
3.2.1. Supervisor Developmental Feedback
This study employed the Supervisor Developmental Feedback Scale developed by Zhou [
23]. Respondents self-reported their perception of supervisor developmental feedback, which comprised three items, e.g., “While giving me feedback, my supervisor focuses on helping me to learn and improve”. The second item of the scale was subject to reverse scoring. The reliability of this scale was good, with a Cronbach’s
coefficient of 0.711.
3.2.2. Time Theft
Based on Lorinkova and Perry’s methodology for measuring employees’ time theft [
53], this study utilized a three-item scale adapted from Bennett and Robinson’s Workplace Deviance Scale [
6]. For example, “I have taken an additional or longer break than is acceptable at workplace”. In this study, the time theft scale demonstrated good reliability, with a Cronbach’s
coefficient of 0.772.
3.2.3. Perceived Insider Status
This study employed a six-item Perceived Insider Status Scale developed by Stamper and Masterson, e.g., “My work organization makes me believe that I am included in it” [
32]. Three items (the third, fourth, and sixth items) were subject to reverse scoring. The reliability of this scale was satisfactory in this study, with a Cronbach’s
coefficient of 0.823.
3.2.4. Work Passion
This study adopted the Work Passion Scale developed by Vallerand et al., consisting of nineteen items [
38]. Among these, fourteen items were employed to measure employees’ current levels of harmonious work passion and obsessive work passion, while the remaining five served as criteria to assess whether an employee had work passion or not. Considering our research objective of introducing the work passion concept in order to explore how much employees value their work, rather than comparing two types of work passion, the latter five items of the Work Passion Scale were employed (for example, “I invest a lot of time and energy into my work”). The fifth item was scored reversely. The scale had good reliability in this study, with a Cronbach’s
coefficient of 0.743.
3.2.5. Perceptions of Leader Hypocrisy
This study employed the Perceptions of Leader Hypocrisy Scale developed by Dineen, Lewicki, and Tomlinson, comprising four items, e.g., “I wish my supervisor would practice what he or she preaches more often” [
48]. The Perceptions of Leader Hypocrisy Scale demonstrated good reliability in this study, and the Cronbach’s
coefficient of the scale was 0.720.
3.2.6. Control Variables
Drawing on prior empirical research on time theft [
4,
16,
19,
53], this study considered seven variables—gender, age, education, job type, industry, tenure, and job level—as control variables to minimize confounding factors and reduce potential biases in the study results. Specifically, gender was controlled using a binary variable (0 = male, 1 = female); age was categorized into five levels, ranging from 19 years and below to 51 years and above; education was divided into five categories, ranging from high school and below to doctoral graduate; job type was categorized into five classes (state-owned enterprises, public institutions, civil servants, private enterprises, and foreign-owned companies); industry was classified into thirteen categories (e.g.,financial industry); tenure was divided into five stages, ranging from less than one year to ten years and above; and job level was categorized into four levels: regular employees, junior managers, mid-level managers, and senior managers.
3.3. Results
3.3.1. Harman’s One-Factor Test
Common Method Bias (CMB) refers to the artificial covariance between predictor and criterion variables caused by common raters, item characteristics, item context (e.g., context-induced mood), and measurement context (e.g., predictor and criterion variables measured at the same point in time) [
54,
55]. In this study, to minimize the impact of CMB, methods such as anonymous self-assessment by employees and online surveys were employed, with the aim of separating measurements spatially and psychologically and protecting the respondents’ anonymity.
Because the data of this study were collected from a single source (i.e., employees’ self-reports), Harman’s one-factor test, which has been commonly used in behavioral integrity and workplace deviance studies, was further employed to address the issue of CMB [
27]. If a single factor emerges that explains a significant portion of the covariance among the variables, this suggests the presence of a common method factor [
54]. This study employed SPSS 27.0 to conduct Harman’s one-factor testing. The results from the unrotated exploratory factor analysis reveal that the covariance explained by the first factor is 38.87%, which is below the 40% threshold. This finding suggests that no single factor in this study accounts for a substantial proportion of the total variance, indicating limited influence of CMB on the results.
3.3.2. Confirmatory Factor Analysis
To further examine the discriminant validity across supervisor developmental feedback, time theft, perceived insider status, work passion, and perceptions of leader hypocrisy, this study conducted confirmatory factor analysis (CFA) [
56]. Utilizing Mplus 8, this study constructed various factor models based on concepts of variables. The CFA results reveal that the model fit of the five-factor model tends to be most satisfactory, outperforming alternative factor models (see
Table 1). The results indicate that the discriminant validity of this study is good and suitable for hypothesis testing.
3.3.3. Descriptive Statistics and Correlation Analysis
Table 2 displays the means, standard deviations, and inter-correlations among the measured variables in Study 1. The results indicate a significant negative correlation between supervisor developmental feedback and employee time theft (
,
). In addition, there is a significant positive correlation between supervisor developmental feedback and employees’ perceived insider status (
,
). Employees’ perceived insider status is significantly positively correlated with employees’ work passion (
,
), and employees’ work passion is significantly negatively correlated with employees’ time theft (
,
). These findings preliminarily support the relationships among the main variables in the proposed serial multiple mediation model, laying the groundwork for a more in-depth examination of the research hypotheses.
3.3.4. Hypothesis Testing on the Serial Multiple Mediating Effect
After controlling for gender, age, education level, job type, industry, tenure, and job level, this study employed supervisor developmental feedback as the independent variable, perceived insider status and work passion as mediating variables, and time theft as the dependent variable, then conducted hierarchical regression analysis using SPSS 27.0. The results in
Table 3 reveal that supervisor developmental feedback has a significant negative impact on time theft (Model 6,
= −0.487,
), providing support for Hypothesis 1. Additionally, supervisor developmental feedback positively influences perceived insider status (Model 1,
= 0.598,
) and perceived insider status negatively influences time theft (Model 7,
= −0.604,
). When taking both supervisor developmental feedback and perceived insider status into account in the regression model, the results reveal that perceived insider status has a significant negative impact on time theft (Model 8,
= −0.481,
) and supervisor developmental feedback continues to have a significant negative effect on time theft (Model 8,
= −0.199,
). Thus, perceived insider status plays a partial mediating role in the relationship between supervisor developmental feedback and time theft, supporting Hypothesis 2. The hypothesis testing for the mediating role of work passion between supervisor developmental feedback and time theft followed a similar process. Supervisor developmental feedback positively predicts work passion (Model 3,
= 0.533,
) and work passion negatively predicts time theft (Model 9,
= −0.668,
); moreover, when taking both supervisor developmental feedback and work passion into account in the regression model, the results show that work passion has a significant negative on time theft (Model 10,
= −0.564,
) and the impact of supervisor developmental feedback on time theft remains significantly negative (Model 10,
= −0.186,
), supporting Hypothesis 3. In addition, the results of Model 11 affirm the serial multiple mediating effect between perceived insider status and work passion, preliminarily supporting Hypothesis 4.
Then, this study further employed Model 6 from the SPSS PROCESS macro proposed by Hayes and conducted 5000 bootstrap resamples to validate the serial multiple mediating effects [
57]. The results in
Table 4 reveal that the mediating effect of employees’ perceived insider status on the relationship between supervisor developmental feedback and employee time theft is significant (Effect = −0.288, 95% CI = [−0.376, −0.210]). The mediating effect of employees’ work passion on the relationship between supervisor developmental feedback and employee time theft is significant as well (Effect = −0.301, 95% CI = [−0.375, −0.236]). Furthermore, employees’ perceived insider status and work passion serve as a serial multiple mediating effect between supervisor developmental feedback and employee time theft (Effect = −0.382, 95% CI = [−0.477, −0.302]); thus, Hypothesis 4 is confirmed.
3.3.5. Hypothesis Testing on the Moderating Effect of Perceptions of Leader Hypocrisy
To test the moderating effect of perceptions of leader hypocrisy on the relationship between supervisor developmental feedback and perceived insider status, this study utilized SPSS PROCESS macro Model 1 and conducted 5000 bootstrap resamples [
57]. As shown in
Table 2, perceptions of leader hypocrisy negatively predict perceived insider status (Model 2,
= −0.185,
). However, the interaction term between supervisor developmental feedback and perceptions of leader hypocrisy on the negative impact on perceived insider status is not significant (Model 2,
= −0.042, 95% CI = [−0.103, 0.018]). Therefore, this study rejects Hypothesis 5. As Hypothesis 5 is a prerequisite for Hypothesis 6, this study preliminarily rejects Hypothesis 6. Furthermore, this study employed SPSS PROCESS macro Model 83 to evaluate the significance of the moderation effect on the serial multiple mediation model [
57]. The results in
Figure 1 reveal that the difference in the moderation effect of perceptions of leader hypocrisy on the serial multiple mediation relationship between supervisor developmental feedback, perceived insider status, work passion, and time theft is not significant (95% CI = [−0.010, 0.032]). Consequently, Hypothesis 6 of this study is rejected.