Altered States or Much to Do About Nothing? A Study of When Cannabis Is Used in Relation to the Impact It Has on Performance

Authored by journals.sagepub.com and submitted by LoreleiOpine

As more local, state, and national governments change laws regarding the legality of cannabis use, it is essential for organizations to understand how the workplace may be influenced by these changes. The current study begins to answer this question by examining the relationship between three temporal-based cannabis measures and five forms of workplace performance. Using data from 281 employees and their direct supervisors, our results indicate that cannabis use before and during work negatively relate to task performance, organization-aimed citizenship behaviors, and two forms of counterproductive work behaviors. At the same time, after-work cannabis use was not related (positively or negatively) to any form of performance as rated by the user’s direct supervisor. We discuss methodological, theoretical, and practical implications for researchers, organizations, and governmental agencies concerned with cannabis use.

Once considered “. . . the menace which is destroying the youth of America,” and “the real public enemy number one” (Esper & Gasnier, 1938), cannabis’ place in society has drastically changed over the last 50 years. Originally viewed as an intoxicant used only by deviant members of society, cannabis is now the most widely used illegal substance in the United States, Canada, and Europe (Beck & Legleye, 2008; Frone, 2006; Hajizadeh, 2016); this popularity translates into millions of employees who consume cannabis. Given such a state of affairs, it should be of little surprise that organizations spend billions of dollars each year (Frone, 2006; Normand et al., 1990) addressing what many believe is a problem (Bass et al., 1996; Harris, 2004; Lyons et al., 2016). Whereas some might believe steps taken by governments and organizations to screen and/or prevent cannabis use are supported by empirical evidence, “such commentary is often based on speculation, vested interests, and pseudo-science rather than on objective interpretation of all available scientific data” (Frone, 2008a, p. 519). In fact, much existing research on cannabis in relation to the workplace, which is rather dated (e.g., Colaiuta & Breed, 1974; Mangione & Quinn, 1975; McDaniel, 1988), finds mostly mixed results with weak effect sizes (Frone, 2008a; Harris, 2004; Normand et al., 1994; Olbina et al., 2011).

Dated research that offers limited empirical evidence is a serious concern for organizations, governments, and society, but perhaps just as serious as these concerns are methodological deficiencies within the study of cannabis. Existing research illustrates this point as cannabis is frequently grouped with other substances in a catch-all category despite meaningful pharmacological and physiological differences (Frone, 2008a; Galaif et al., 2001; Harris, 2004; Normand et al., 1990). As evidence of this, one of the most recent studies exploring cannabis use in relation to the workplace published in a premier management journal grouped cannabis and cocaine in a single category (see Galaif et al., 2001). And whereas some scholars recognize substance use research focuses largely on alcohol consumption, they nevertheless conclude similar results would be found with other substances such as cannabis (Harris, 2004). Such conclusions are certainly open to debate, but one notable finding within the alcohol literature that has yet to be incorporated into other substance explorations relates to measurement. That is, measures that reflect a temporal relation to the workday have proven to be critical in detecting relationships not evident in broad historical measures (Frone, 2008b). A lack of temporal consideration with cannabis research is particularly important for several reasons including the fact that cannabis effects dissipate relatively quickly with fewer residual effects than other substances (Chait, 1990; Lucas et al., 2013; Reiman, 2009). This means that certain types of cannabis use may not have the same detrimental effects as other substances even though workplace drug tests would view all forms of cannabis use (e.g., before-work, after-work, weekend use) similarly.

We designed the current study to address these methodological concerns. In particular, we draw from organizational and epidemiology research to suggest a more nuanced conceptualization of cannabis as it relates to the workplace. This study then uses temporal cannabis items (e.g., how often during the past 12 months have you used cannabis before starting your work day) to predict five forms of workplace performance as rated by the cannabis user’s direct supervisor. To our knowledge, this is the first study using temporal cannabis items in relation to different forms of workplace performance; this is also the first study isolating cannabis in relation to workplace behaviors in nearly 20 years. Both of these aspects of our study represent important contributions because previous research ignoring when cannabis is used in favor of whether or not it is used may mask discrepancies that employees and applicants could exploit in the court of law. A temporal focus could also help resolve contradictory theoretical discrepancies that suggest cannabis is both harmful (e.g., impaired functioning theory, Galiaf et al., 2001) and helpful (e.g., stress-response dampening proposition, Sayette, 1999; self-medication hypothesis, Khantzian, 1985, 1997). The performance expectations of modern employees differ from the expectations placed on employees of prior eras as well, meaning generalized performance studies from decades ago lack the subtleties found in current definitions of performance. By empirically studying untested constructs in a more rigorous manner, this study addresses a critical need at a critical point in time (i.e., a time in which local and federal governments along with managers and organizations are struggling with how to address a societal issue).

Method Participants and Procedure We solicited participation by way of two different methodologies. Students enrolled in upper-level business courses at two universities (one in the Southern and one in the Western United States) were offered extra course credit for submitting the names and email addresses of a full-time employee and their direct supervisor (students were not allowed to participate). We then contacted employees and supervisors directly through email describing a study about workplace stressors, employee downtime, and work-related well-being. We purposefully left out reference to cannabis in the study description and consent form to protect employees and because cannabis use was not a mandatory requirement for study participation. All study procedures were approved by our Institutional Review Board (IRB). Cannabis use is on the rise and used by millions of employees (Frone, 2006; Hajizadeh, 2016; Lytle, 2014), yet the base rate of cannabis use is low relative to alcohol use or other organizational phenomenon of interest (Frone, 2006). As such, we employed a second methodology to help ensure enough variation in our focal variables to detect any potential relationships. This second method including recruiting participants from an online interest group; this interest group has a social media page devoted to cannabis use where we posted a message describing our study and soliciting participation. In exchange for participation in our study, we offered employees and their direct supervisor a US$10 Amazon gift card. Interested employees were directed to a link to complete the survey; once employees completed the survey, we sent them an email invitation describing the study (without reference to cannabis) to forward to their direct supervisor (having employees forward the survey invitation and link to their supervisor as opposed to a direct invitation was required by our IRB). As a result of these two methodologies (and the data quality checks described next), the final sample included data from 281 employees matched with 281 direct supervisors. Reported demographic statistics for employees includes 47% male, 53% female; 16% Black, 65% White, 10% Hispanic, 6% Asian/Pacific Islander, 1% Native American, 2% other; <1% some high school, 13% high school grad/GED, 19% some college, 12% 2-year college degree, 43% 4-year college degree, 13% graduate degree. Employees averaged 36.2 (SD = 11.0) years of age and 6.4 (SD = 6.7) years of organizational tenure. A total of 98% of employees came from the United States (31 states), and the remaining respondents represented six countries. Demographic statistics for supervisors includes 60.5% male, 39.5% female; 12% Black, 70% White, 10% Hispanic, 6% Asian/Pacific Islander, 2% other. The average age of supervisors was 44.3 (SD = 9.2) years; the average time spent supervising the employee participant was 3.9 (SD = 4.6) years. We recorded survey responses, unless otherwise noted, on a 7-point strongly disagree to strongly agree response scale. Employee-Rated Measures Cannabis use As alluded to above, measuring cannabis use is not straightforward. Urinalysis or other medical tests provide objective confirmation of cannabis use, but they only detect the presence of cannabis metabolites not the frequency of nor the time in which (relative to work) cannabis was consumed (Bass et al., 1996; Harris, 2004). An employee who consumes cannabis during a weekend party would return similar testing results as someone who consumed cannabis immediately prior to their work shift. It follows that we asked employees to self-report (a) how frequently they used cannabis over the past 12 months within 2 hours of starting their work shift (0 = I have never used cannabis, 1 = I used cannabis within the last 12 months, but never before starting my work day, 2 = I have used cannabis 1–5 times before starting my workday over the last 12 months, 3 = I have used cannabis about once a week before starting my workday over the last 12 months, 4 = I have used cannabis 2–3 times a week before starting my workday over the last 12 months, 5 = I have used cannabis before starting my workday virtually every day over the last 12 months), (b) how frequently they used cannabis over the past 12 months during their workday (0 = I have never used cannabis, 5 = I used cannabis during my workday virtually every day over the last 12 months), and (c) how frequently they used cannabis over the past 12 months within 2 hours of leaving work (0 = I have never used cannabis, 5 = I used cannabis following my workday virtually every day over the last 12 months). Supervisor-Rated Measures Task performance Supervisors evaluated their employee’s task performance with a 7-item measure by Williams and Anderson (1991). An example item includes this employee “Adequately completes assigned duties.” Coefficient alpha for the measure was .78. OCBs Supervisors assessed their employee’s citizenship behavior with 14 items by Williams and Anderson (1991); seven items described behaviors aimed at helping individuals within the organization (OCBi, e.g., “Helps others who have heavy workloads,” α = .81) and seven items described behaviors aimed at helping the organization itself (OCBo, e.g., “Adheres to informal rules devised to maintain order,” α = .78). CWBs With a 7-point (1 = never, 7 = daily) response option, supervisors assessed CWBs with 10 items by Stewart et al. (2009). This measure includes seven items assessing employee’s production deviance (e.g., “Spent too much time fantasizing or daydreaming instead of working,” α = .90) and three items assessing property deviance (e.g., “Took property from work without permission,” α = .79). The results described below are based on analyses with the original three-item property deviance measure by Stewart et al. (2009); one of these items (“used and illegal drug or consumed alcohol on the job”) may inadvertently conflate results given cannabis use was the focal predictor. We reran our analyses with this item removed and results/conclusions remained the same. Methodological Check The recruitment of participants via social media and through a link forwarded by these individuals opened our study to unknown individuals. We took several steps to alleviate possible concerns about the validity of this methodology. First, embedded within both surveys were multiple instructed response items (please select “strongly agree” to this statement; Breitsohl & Steidelmüller, 2018; Kung et al., 2018; Ward & Meade, 2018). Second, we also recorded the total amount of time taken to complete each survey (Huang et al., 2012). We eliminated 21 participants who missed instructed response items and/or took fewer than 5 min to complete the survey. Third, we emailed employees following the completion of their survey asking them a math question, their age, and their location. The math question was used to ensure respondents were a real person (as opposed to a spammer or robot); their age was used to check against what respondents reported on their survey (if they were not being truthful on their survey responses, a mismatch would flag their survey); their self-reported location was used as an objective check of truthfulness insofar as each survey response recorded the survey takers Internet Protocol (IP) address (Smith et al., 2016). Using a free IP lookup tool, we were able to pinpoint the location of each respondent and compare that location to what was reported via email. Six respondents who missed parts of these steps were excluded from study results. Finally, as a fourth check, we compared the IP address of employees with the IP address of their direct supervisor. While duplicate IP addresses do not necessarily mean the same person completed both surveys (i.e., they could have both used the same network or laboratory computer), we eliminated two employee–supervisor pairs who shared IP addresses just to be safe. Confirmatory Factor Analyses (CFAs) Prior to running study hypotheses, we ran a confirmatory analysis on the data provided by supervisors. Given our sample size and the number of constructs measured, we parceled items together so that each latent factor was explained by three indicators, which should help prevent inflated goodness-of-fit statistics (see Kishton & Widaman, 1994; Landis et al., 2000). Results indicated the data fit the hypothesized five-factor model adequately: χ2(80) = 310.63, p < .001; comparative fit index (CFI) = .92, standardized root mean square residual (SRMR) = .05, root mean square error of approximation (RMSEA) = .10, and that this model fit better than an alternative CFA that combined all citizenship items onto a global OCB construct and all counterproductive work items onto a global CWB construct, χ2(87) = 576.71, p < .001; CFI = .83, SRMR = .09, RMSEA = .14.

Results Table 1 provides the means, standard deviations, and intercorrelations for all study variables. Table 1. Descriptive Statistics and Correlations for Study Variables. View larger version Study hypotheses predicted relationships between three different types of cannabis use (i.e., before, during, and after work) and multiple forms of performance (task, OCB, and CWB). We tested each of these hypotheses using simple ordinary least squares (OLS) regression. Results displayed in Table 2 indicate using cannabis before (b = −.11, p= .001) and during (b = −.14, p < .001) work negatively related to task performance; after-work cannabis use did not relate to task performance. This pattern of results indicates support for Hypothesis 1a and 1b but not 1c. Results in Table 2 also indicate that none of the three forms of cannabis use correlated with citizenship behaviors aimed at individuals within the organization. That said, before- and during-work cannabis use did negatively relate to organization-aimed citizenship behaviors (after-work had no relationship) meaning we found partial support for Hypothesis 2. Finally, Hypothesis 3, which predicted relationships between cannabis use and CWBs, received full support as both before-work (production [b = .21, p < .001], property [b = .29, p < .001]) and during-work (production [b = .23, p < .001], property [b = .41, p < .001]) cannabis use positively correlated with CWBs. Table 2. Simple Regression Results Using Work-Relevant Cannabis Measures and Leader-Rated Criterion. View larger version

Discussion Although it is common for organizations to screen employees and applicants for substances including cannabis and for politicians and societal leaders to make sweeping claims about cannabis, there is virtually no empirical research exploring cannabis use in relation to the modern workplace. This omission is important and unfortunate as modern cannabis (and cannabis use) differs greatly from that explored in the 1970s and 1980s (Morgan et al., 2012; Nicholson et al., 2004; Walsh et al., 2017). To address this gap in organizational and societal knowledge, we proposed a temporal-based cannabis framework that predicts relationships between three different forms of cannabis use (before, during, and after work) and five forms of workplace performance. Results indicate using cannabis before or during work harmed four of five different dimensions of performance rated by the user’s direct supervisor, yet contrary to commonly held assumptions, not all forms of cannabis use harmed performance. In fact, after-work cannabis use did not relate to any of the workplace performance dimensions. This finding casts doubt on some stereotypes of cannabis users and suggests a need for further methodological and theoretical development in the field of substance use. Beyond offering empirical evidence with regard to a topic at the forefront of a national and international debate, this study’s findings highlight the need to move away from broad historical conceptualizations of cannabis (e.g., “have you ever used cannabis”) in favor of more nuanced (e.g., trilateral) work-centric conceptualizations. This represents a substantive methodological contribution in that it is rare to find substance use research exploring anything other than lifetime substance use or use versus nonuse in relation to the workplace. Taking a trilateral approach, as a result, revealed patterns that would have remained hidden had we used a singular conceptualization of cannabis use or a singular conceptualization of performance. Such findings are obviously consequential for scholars and practitioners who might inaccurately believe that it is cannabis use in general that harms workplace behaviors. Our research suggests this is not the case as after-work cannabis use did not predict a single form of performance as rated by one’s direct supervisor. As we noted above when proposing relationships between different kinds of cannabis use and performance, the immediate and next-day after-effects of cannabis differ significantly from other types of substances. Thus, when predicting relationships between substances and aspects of workplace effectiveness, scholars should incorporate a more fine-grained conceptualization of substance use. Related to this point, we also found before-work and during-work cannabis use related to different forms of performance, but this was not a universal effect. Neither form related (positively or negatively) to citizenship behaviors aimed at individuals within the organization. This finding, once again, emphasizes the necessity of more clearly articulating what aspect of cannabis use and what aspect of performance one is referencing when making statements or predictions about substance use in the workplace. The methodological implications of this study offer opportunities for future theoretical development in the area of work-related substance use as several explanations are put forth in the existing literature, but none of these perspectives distinguish cannabis use from cannabis use before, during, or after work. While only an initial exploration into cannabis use (the first we are aware of within the last decade), our results found after-work cannabis use does not correlate with any of the performance dimensions typically incorporated into management research. This finding would have been masked had we used a global measure such as “have you used cannabis in the last 6 months.” This suggests a strong need to expand and reframe existing substance use theories to include explicit reference to the manner and timing in which cannabis is used. Failure to do so could lead scholars or organizational leaders to incorrectly suggest all forms of cannabis use are harmful, which may subsequently support the continuation of broad policies prohibiting all forms of cannabis use. We note this is much more than simply a theoretical issue though as the public’s (i.e., applicants) approval of cannabis has changed dramatically over the last two decades (Swift, 2016). Given differences in cannabis use across races (Normand et al., 1990), organizations need to be prepared to provide evidence in support of their policies; and if, as our research shows, off-the-job cannabis use has little to no impact on workplace performance, organizations will be hard pressed to provide legally defensible justifications for the continuation of policies prohibiting all forms of cannabis use (or time-invariant drug tests). Limitations and Future Research The recruiting of a unique multisource sample is a strength of this research, yet we are mindful of certain study limitations that warrant acknowledgment. Two related potential limitations include concerns over common method bias and self-reported measures of cannabis use. Given the sensitivity of our focal topic (i.e., substance use), we were limited in the types of methodological approaches that would be deemed ethical and acceptable by our IRB. Knowing this would be a valid concern of reviewers and readers, we turned to methodological best practices (see Podsakoff et al., 2003) to help design a study that accounted for ethical concerns while providing the necessary rigor of high-quality research. In the end, we included three exogenous variables provided by one source and performance data provided by a second unique source (i.e., direct supervisors). This does not eliminate methodological concerns, but it should lessen worries that common method bias was the driving force behind study results. We also recognize that self-reported cannabis use may raise concerns with some readers even though existing research indicates self-reported substance use is a reliable and valid methodology for collecting such data (Bass et al., 1996; Frone, 2008a; Harris, 2004). Nevertheless, if concern remains, we direct readers to existing findings that indicate self-report measures sometimes underreport behaviors (e.g., substance use) that are socially undesirable (Aguinis et al., 1995; Harris, 2004). This would mean our data underestimate cannabis use and results are conservative estimates of true relationships. An additional consideration of this study is the use of a between-person model of cannabis use and the workplace. Whereas we asked participants to describe their recent cannabis use with a temporal focus, substance use scholars note substance use is fluid (Allen et al., 2017; Frone, 2008b). This coincides with recent management research that indicates stressors, coping behaviors, and performance also ebb and flow (e.g., Kim et al., 2018). In light of this, future research would be well served to incorporate dynamic models and methodologies (e.g., experience sampling methodologies [ESM]) into the study of cannabis use as it relates to the workplace. This would be particularly helpful in exploring potential antecedents to cannabis use such as psychological contract breach, abusive supervision, or any number of other factors that could fluctuate over time, and as such, influence not only if employees use substances but when they decide to use them (e.g., while on the job immediate after a perceived psychological contract breach). An ESM approach would also help provide a test (and potential advancement) of theory as such a methodology could examine intervening mechanisms between substance use and performance-related outcomes. Results reported in this study indicate a link between before- and during-work cannabis use and reduced performance, but we were left to speculate about the reason (e.g., decreased cognitive functioning) for this decline. A within-person study that assesses concentration, focus, energy, and/or other resources at different points in time may provide an explanation for both our significant and insignificant findings. That is, after-work cannabis use could allow for relaxation and resource recovery to take place which would help with performance-related aspects of one’s work, but it may be indirect such that after-work cannabis users get better sleep, wake up feeling more energized, and this positive state is what sets the stage for improved performance. Because between-person studies mask such dynamics, there is a very real need for future research in this area. One last point in relation to employing different types of methodologies such as an ESM relates to the causal sequence of events. We employed a multi-source cross-sectional design to make assertions about the after-effects of cannabis use, but it is possible that poor performance acts as a stressor that induces substance use. Future research employing longitudinal designs would help clarify the order of events. Having noted that after-work cannabis use could potentially help some aspects of work-life, it is important to remind readers our research did not find a link with any aspect of performance nor does our research say anything about the long-term health implications of cannabis use. Habitual use of cannabis after work for recovery purposes may lead to automatic consumption, an increase in tolerance levels, and risk for substance dependence over time (Liu et al., 2015). As the management sciences are rich with explorations into other forms of coping (e.g., exercise, reading, volunteering; Sonnentag et al., 2008, 2017), future research that compares the short-term and long-term health implications of after-work cannabis use and other recovery activities would provide further guidance in this area.

Conclusion Cannabis use among the general public has exploded over the past decade, yet there is virtually no empirical research within the organizational sciences exploring the performance-related implications of cannabis use for more than two decades. This is clearly problematic as our knowledge and understanding of the workplace has advanced at the same time the types and uses of cannabis have evolved. This study took an initial attempt to address this void by exploring the relationship between different forms of cannabis use and different forms of workplace performance. Results linking certain types of cannabis use (i.e., before work) with workplace performance are important to organizations with substance use policies, but so too are null findings between after-work cannabis use and workplace performance. Ultimately, organizations and governmental bodies need to present evidence in favor of their beliefs and substance policies. We hope this research serves as a foundation for what could be a rich and long field of inquiry in the years to come.

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article. ORCID iD

Jeremy B. Bernerth https://orcid.org/0000-0001-7925-0587 Associate Editor: Thomas Zagenczyk

Because we have no empirical or theoretical reason to believe after-work cannabis use relates to counterproductive work behaviors (CWBs), we offer no formal hypothesis. We do, however, test such a relationship for completeness.

wrestlingchampo on August 22nd, 2020 at 16:18 UTC »

This is another study where I will remind people of another reason why this is important: On the Job and pre-employment drug tests are highly biased toward rooting out cannabis use, on the basis of the active ingredients and metabolites measured are fat-soluble in cannabis, and water-soluble in damn near every other illegal drug or measured intoxicant (Ethanol via BAC testing for alcohol) in drug tests.

This means that a cocaine or heroin user needs to abstain from their drug for ~72 hours to clear most urinary or blood tests for drugs. Cannabis users must abstain for at least a week, if not an entire month to lower their metabolite levels to an acceptable level to pass a drug test. The problem is the active ingredient/metabolite gets stored away in cell vacuoles, and are periodically released over time like triglycerides (Not 100% on the exact mechanism).

This can be frustrating, simply because every employer would (likely) take a stoner over someone addicted to Heroin or Cocaine, or even Alcohol these days.

bpowell76 on August 22nd, 2020 at 13:42 UTC »

How would use AFTER work have an impact on work the next day?

kokoskokos on August 22nd, 2020 at 12:54 UTC »

Hopefully studies like this will change the way cannabinoids are perceived by people, encourage more research and pave the road to legalisation!