From self-defeating to other defeating: Examining the effects of leader procrastination on follower work outcomes.

This research examines the influence of leader procrastination on employee attitudes and behaviours. While previous studies have typically viewed procrastination as a form of self-defeating behaviour, this research explores its effects on others in the workplace. In Study 1, using data collected from 290 employees, we demonstrate the discriminant and relative predictive validity of leader procrastination on leadership effectiveness compared with laissez-faire leadership and directive leadership. In Study 2, based on dyadic data collected in three phases from 250 employees and their 23 supervisors, we found that leader procrastination was associated with follower discretionary behaviour (organizational citizenship behaviour and deviant behaviour). Additionally, job frustration was found to mediate the relationship between leader procrastination and follower outcomes. The quality of the leader-follower relationship, as a boundary condition, was shown to mitigate the detrimental effects of leader procrastination. Together, the findings suggest that leader procrastination is a distinct form of negative leadership behaviour that represents an important source of follower job frustration. PRACTITIONER POINTS: Leader procrastination is different from laissez-faire and directive leadership and can be detrimental to followers.Job frustration mediates the relationship between leader procrastination and follower discretionary behaviour.Organizations should facilitate high-quality LMX relationships as a method for mitigating the negative effects of leader procrastination.

Background

Procrastination refers to the tendency to ‘voluntarily delay an intended course of action despite expecting to be worse off for the delay’ (Steel, 2007, p. 66). To date, research on procrastination at work has focused largely on the implications for the procrastinator, showing negative effects on outcomes, such as job performance and subjective well‐being (e.g., Tice & Baumeister, 1997). While such attention to self‐defeating behaviour is important, it neglects the fact that procrastination may have broader implications for others in the workplace.

To address this limitation, we suggest that a valuable starting point is to examine the impact of leader procrastination on followers. Procrastination typically emerges in times of pressure and when action is needed, which are two prominent features of the leadership role. We propose that leader procrastination is different from other types of leadership behaviour, such as laissez‐faire leadership (Wong & Giessner, 2016) and directive leadership (Pearce & Sims, 2002). Furthermore, we argue that leader procrastination is likely to influence followers, as followers are often highly dependent on their leader to obtain access to both tangible and intangible resources (e.g., Liden, Erdogan, Wayne, & Sparrowe, 2006). By exploring a boundary condition, we also highlight the potential for leader–member exchange (LMX) quality to mitigate the deleterious effects of leader procrastination. Finally, we analyse whether job frustration mediates the relationship between leader procrastination and follower discretionary behaviour, namely organizational citizenship behaviour (OCB) and deviant behaviour. Figure 1 visually displays this conceptual model.

Figure 1

Hypothesized model.

Theory and hypotheses

Leader procrastination can be positioned as a passive form of negative leadership behaviour (Leary et al., 2013). As such, leader procrastination shares similarities with laissez‐faire leadership, another type of negative but passive leadership behaviour. Similar to procrastination, laissez‐faire leadership involves the delay or absence of decision‐making (e.g., Zwingmann et al., 2014). However, laissez‐faire leadership represents the complete absence of leadership, with no involvement or attempt to motivate followers. This is conceptually different from the notion of procrastination, which more specifically refers to delaying a course of action. Leader procrastination also has some conceptual overlap with directive leadership (Pearce & Sims, 2002). In many ways, directive leadership represents the antithesis of procrastination as it describes leaders who are decisive. However, unlike procrastination, directive leadership represents a style of leadership that relies on position power to achieve results and is a more active form of leader behaviour than procrastination. Based on these conceptual differences the following is hypothesized:

Hypothesis 1: Leader procrastination will have relative predictive validity over (a) laissez‐faire leadership and (b) directive leadership on leader effectiveness.

We expect that, like other forms of negative leadership, leader procrastination will have deleterious effects on follower behaviour. Leader procrastination is likely to interfere with followers’ jobs and, ultimately, their ability to attain important work‐related goals. This is because followers are often dependent on their leader for the resources necessary to do their job well (Liden et al., 2006). Thus, we posit that leader procrastination will be particularly likely to elicit feelings of job frustration. Intense responses can occur in situations that threaten goal achievement (Weiss & Cropanzano, 1996). As described by affective events theory, employees often draw from affective experiences when constructing job attitudes and subsequent behaviour. In support of this contention, Avey, Wu, and Holley (2015) showed that job frustration that resulted from abusive supervision was associated with deviant behaviour. Hence, the following is hypothesized:

Hypothesis 2: Frustration will mediate the relationship between leader procrastination and (a) leader‐rated OCB, and (b) leader‐rated deviant behaviour.

While we predict that leader procrastination will elicit feelings of frustration, such experiences may be reduced when procrastination occurs within an otherwise positive leader–follower relationship. Research outside of the workplace highlights that relationship quality promotes forgiveness in relationships (e.g., Fincham, Paleari, & Regalia, 2002). Similarly, we argue that followers will be more forgiving of leader procrastination when it occurs within a high‐quality LMX relationship. Within such relationships, followers may attribute procrastination to external rather than internal causes, making it easier to forgive such behaviour and reducing the level of frustration felt as a result. Thus,

Hypothesis 3: LMX quality will moderate the strength of the mediated relationship between leader procrastination and (a) leader‐rated OCB and (b) leader‐rated deviant behaviour via job frustration.

Method

The main goal of Study 1 was to demonstrate the discriminant and relative predictive validity of leader procrastination compared with other leadership constructs with conceptual overlap, namely, laissez‐faire leadership and directive leadership. Data were collected via a Qualtrics Panel of 290 US adults, 64% of whom were female with an average age of 35 years. To help ensure the quality and relevance of the sample, participants were required to be full‐time working adults. Several attentional filters were included to ensure that the respondents were paying attention when completing the survey.

Measures

We measured leader procrastination using a 10‐item adapted version of the scale by Díaz‐Morales, Ferrari, and Díaz (2006); a sample item includes ‘my manager delays making decisions until it's too late’ (α = .96). Laissez‐faire leadership was measured using a 4‐item scale by Wong and Giessner (2016); a sample item is ‘My manager does not bother me when I do not bother him/her’ (α = .85). Directive leadership was measured using a 9‐item scale by Pearce and Sims (2002); a sample item is ‘My manager establishes the goals for my work’ (α = .90). Leader effectiveness was measured using four items (Piccolo et al., 2012): a sample item is ‘My leader is effective in meeting my job‐related needs’ (α = .82).

Results

We conducted a series of confirmatory factor analyses (CFAs) using MPLUS 7 (Muthén & Muthé& Muthén, 2012–2015) to examine the distinctiveness of the variables used in our study. As can be seen in Table 1, the model that allowed the various items to load onto their respective factors produced a better model fit (χ2 = 1354.88, df = 318, p < .01; comparative fit index [CFI] = .83; root mean square error of approximation [RMSEA] = .10) than any of the models in which the scales were combined. This was confirmed by chi‐squared difference tests, which were all significant. However, while the four‐factor model produced the best model fit for our data, the CFI and RMSEA values are both outside the acceptable range (see Hu & Bentler, 1999). Given the high correlation between leader procrastination and laissez‐faire leadership (r = .69), we conducted Fornell and Larcker's (1981) test for discriminant validity, finding that the square root of the average variance extracted (AVE) for leader procrastination (.63) and laissez‐faire leadership (.56) exceeded the maximum shared variance (MSV) (.48) between the latent factors. This provides support for the discriminant validity of the two scales. The correlation between leader procrastination and directive leadership was smaller (r = −.32) and, again, the AVE for both leader procrastination (.69) and directive leadership (.34) exceeded the MSV (.10). To test the incremental validity of leader procrastination, we conducted multiple regression analysis using SPSS (version 24; IBM SPSS, Armonk, NY, USA). We first entered laissez‐faire and directive leadership into a regression model with leadership effectiveness as the dependent variable. In the next step, we added leader procrastination into the model as an additional independent variable. The results showed a significant negative association between leader procrastination with perceptions of leader effectiveness, b = −.25, t (289) = −3.50, p < .01. Directive leadership was also significantly negatively related to leadership effectiveness, b = −.23, t (289) = −3.32, p < .01, whereas laissez‐faire leadership showed no significant association. Furthermore, the addition of leader procrastination added explanatory power, indicated by a change in R‐squared from .03 to .07. Therefore, support was found for Hypothesis 1.

Table 1

Study 1 confirmatory factor analyses results

Model	χ2	df	CFI	RMSEA	Chi‐squared test
Four‐factor model	1354.88	318	.83	.10
Three‐factor modela	1520.24	321	.80	.11	165.36 (3)**
Three‐factor modelb	2372.68	321	.66	.15	1017.80 (3)**
Three‐factor modelc	1743.72	321	.76	.12	388.84 (3)**
Three‐factor modeld	2018.48	321	.72	.14	663.60 (3)**
Three‐factor modele	1742.82	321	.76	.12	387.94 (3)**
Three‐factor modelf	2229.48	321	.68	.14	874.60 (3)**
One‐factor model	2884.79	324	.58	.17	1529.91 (3)**

CFI = comparative fit index; TLI = incremental fit index; RMSEA = root mean square error of approximation.

aThree‐factor model combines leader procrastination and laissez‐faire leadership.

bThree‐factor model combines leader procrastination and directive leadership.

cFive‐factor model combines leader procrastination and leadership effectiveness.

dFive‐factor model combines directive leadership and laissez‐faire leadership.

eFive‐factor model combines directive leadership and leadership effectiveness.

fFive‐factor model combines leadership effectiveness and laissez‐faire leadership.

** p < .01.

Data were collected from employees working at a Chinese textile manufacturing company located in Zhejiang Province. At time one, 300 employees were invited to provide demographic information and rate leader procrastination. At time 2, 3 weeks later, they were required to rate their perceptions of LMX and their frustration. An additional 3 weeks later, the direct supervisor rated employees’ OCB and deviant behaviour. A total of 250 employees and 23 supervisors completed the surveys, representing a response rate of 83%.

For all multiple‐item scales, participants rated each item using a 5‐point Likert scale, where 1 = strongly disagree and 5 = strongly agree. Leader procrastination was measured using a 10‐item adapted version of the scale by Díaz‐Morales et al. (2006) (α = .88). Frustration was measured using three items taken from a measure developed by Peters and O'Connor (1980); a sample item is ‘Trying to get my job done is a very frustrating experience’ (α = .92). We measured LMX using the LMX‐7 scale (e.g., Liden et al., 2006); a sample item is ‘I would characterize my working relationship with my manager as very good’ (α = .88). Deviant behaviour was measured using a four‐item scale designed by Aquino, Lewis, and Bradfield (1999); a sample item is ‘This employee calls in sick when not really ill’ (α = .85). OCB was measured with 11 items taken from the scale developed by Williams and Anderson (1991); a sample item is ‘This employee helps others who have been absent’ (α = .94). Gender, age, and dyadic tenure were used as control variables.

Measurement evaluation

Prior to analysis, we conducted a series of CFAs to support the distinctiveness of the variables in our study. Although the model fit was below acceptable levels (Hu & Bentler, 1999), the hypothesized five‐factor model was found to provide a better level of fit (χ2 = 1286.83, df = 517, p < .01; CFI = .86, RMSEA = .08) when compared to any other competing model, such as a single‐factor solution (χ2 = 3501.62, df = 560, p < .01; CFI = .48; RMSEA = .15). Fornell and Larcker's (1981) test demonstrated that the AVE for leader procrastination (.35), LMX (.57), and job frustration (.83) exceeded the MSV (.12) between these latent factors, providing support for discriminant validity.

Hypothesis testing

The means, standard deviations, and correlation coefficients are presented in Table 2.

Table 2

Study 2 descriptive statistics, correlations, and reliability estimates

Variables	Mean	SD	1	2	3	4	5	6	7
1. Leader procrastination	2.88	0.86	(.88)
2. Frustration	2.62	1.33	.34**	(.92)
3. Leader–member exchange	4.58	0.98	−.04	−.15*	(.88)
4. Deviant behaviour	2.67	1.3	.39**	.33**	−.26**	(.85)
5. OCB	4.40	0.89	−.43**	−.32**	.28**	−.71**	(.94)
6. Gender	0.45	0.50	−.01	−.07	−.04	.01	.06	–
7. Age	33.81	7.92	−.03	−.05	.21**	−.04	.09	.06	–
8. Dyadic tenure	4.68	4.01	.03	−.05	.22**	−.03	.10	.06	.65**

N = 250. Gender: 0 = female, 1 = male.

Values in parentheses indicate scale reliabilities (i.e., Cronbach's Alphas).

*p < .05; **p < .01.

As the participants consisted of individuals nested within teams (N = 23 teams), we tested our hypotheses using a multilevel model that included both the individual (follower) level and the team level (see Table 3). We did not aggregate variables to the team level but rather analysed data at the individual level using a method that simultaneously considered the variations between individuals and between teams. Accordingly, we tested the random coefficient models using multilevel regression analysis, employing SPSS (version 23; IBM SPSS) software and using its mixed analysis function.

Table 3

Study 2 multilevel analysis: Effect of leader procrastination (X) on job frustration (M) and effect of job frustration on follower outcomes (Y)

	Model 1 OCB
	Job frustration			OCB
	Coefficient	SE	T	Coefficient	SE	t
Intercept	2.65**	.41	6.45	4.40**	.30	14.52
Age	0.00	.01	0.05	0.00	.01	0.10
Gender	0.16	.16	0.98	−0.06	.09	−0.71
Tenure	−0.01	.03	−0.57	0.01	.02	0.79
X
Leader procrastination	0.48**	.09	5.10	−0.30**	.06	−4.81
Z
M
Job frustration				−0.08*	.04	−2.14

*p < .05; **p < .01.

The results show that leader procrastination was positively and significantly related to job frustration (see Table 3). Furthermore, job frustration was positively and significantly related to deviant behaviour, y = .11, t (236) = 2.38, p < .05, and negatively and significantly related to OCB, y = −.08, t (232) = −2.14, p < .05. To test the significance of these mediated pathways, we calculated 95% Monte Carlo confidence intervals by bootstrapping with 20,000 repetitions. For deviant behaviour, a significant indirect effect of .05 was found, as the 95% confidence intervals did not include zero (LL = .01, UL = .10). Similarly, for OCB, a significant indirect effect of −.05 (LL = −.08, UL = −.00) was found. Therefore, full support for Hypothesis 2 was found.

We further examined the model described above with the inclusion of LMX (Z) as the moderator variable. Both the independent and the moderator variables were grand mean centred before creating the interaction term. As seen in Table 3, the interactive effect of leader procrastination and LMX on job frustration was significant in the analysis related to both deviant behaviour and OCB. To facilitate interpretation, we plotted the simple slopes for two values of LMX. As predicted, Figure 2 shows a stronger positive slope at lower levels of LMX, γ = .64, t (183) = 5.19, p < .01, compared with higher levels of LMX, γ = .30, t (221) = 2.23, p < .05. Thus, high levels of LMX were found to reduce the magnitude of the negative effects of leader procrastination on job frustration.

Figure 2

Moderating effect of leader–member exchange (LMX) on the relationship between leader procrastination and job frustration.

In support of Hypothesis 3, we found a stronger indirect effect between leader procrastination and deviant behaviour at lower levels of LMX (.07; LL = .01, UL = .13) compared with higher levels (.03; LL = .00, UL = .08), which was caused by higher levels of job frustration. Similarly, for OCB, significant mediation was found at low levels of LMX (LL = −.10, UL = −.00), with an indirect effect of −.05. Thus, evidence of mediation was found. At high levels of LMX, no evidence of mediation was found (LL = −.06, UL = .00).

Discussion

Theoretical implications

Overall, the findings provide three key contributions. First, rather than focusing on the intrapersonal costs of workplace procrastination, we explore the consequences of leader procrastination on followers. Our results show that perceptions of leader procrastination were negatively associated with followers’ levels of discretionary behaviour, both positive (i.e., OCB) and negative (i.e., deviant behaviour). Second, we extend the nomological network of variables related to leader procrastination by exploring leader procrastination as a novel antecedent of follower job frustration. Our focus on frustration helps extend the knowledge of its role within the workplace, while answering calls to provide more research on emotions (e.g., Gooty, Connelly, Griffith, & Gupta, 2010).

Third, given the detrimental impact of leader procrastination, research exploring the factors that reduce these negative effects has particular relevance. We predicted that LMX quality would mitigate the link between leader procrastination and job frustration as followers with a high‐quality relationship would be more forgiving of such leader behaviour. The results support this moderating effect, showing that LMX quality attenuates the relationship between leader procrastination and follower job frustration. Importantly, however, we found that LMX only served to reduce but not eliminate the link between leader procrastination and job frustration.

Practical implications

Several practical implications can be garnered from this research. As some leaders may be engaging in procrastination unintentionally, encouraging leaders to engage in feedback interventions with their followers (such as 360‐degree feedback) may be one way to increase the leaders’ awareness of their own behaviour. To avoid feelings of job frustration if their own goal achievement is impaired by leader procrastination, followers could attempt to determine the reason for this behaviour and assist their leader in making decisions. This could take the form of shared leadership (Pearce & Sims, 2002). As the present findings also suggest high‐quality LMX relationships may serve to guard against the negative implications of leader procrastination, organizations should also consider providing training to help facilitate LMX quality within the workplace.

Limitations and future research directions

While efforts were made to minimize the impact of common method variance through separating data collection by several weeks, only a truly longitudinal design can fully address this issue. Although Fornell and Larcker's (1981) test showed support for discriminant validity, it is important to note that in both studies, the CFAs demonstrated a model fit that was below acceptable levels (Hu & Bentler, 1999). This casts some doubt as to how well the measured variables represent the latent constructs. Finally, procrastination may be influenced by the environment in which leadership is enacted. A high‐stakes climate, for example, in which the decisions made have major implications, may make the leader more cautious about their decisions, and more prone to delay. Moreover, in high power distance cultures, leader procrastination may be deemed more acceptable. Future research should aim to incorporate these broader considerations into their design.