Personality and impulsivity as predictors of tobacco use among emerging adults: A latent class analysis

1. Introduction

Many adult users report initiating use of a tobacco product before the age of 18 (CDC, 2014; Lipari & Van Horn, 2017; Sharapova et al., 2018). However, the transition from youth (ages 13–17) to emerging adulthood (ages 18–29; Arnett, 2014) is a critical developmental period whereby individuals experience many lifestyle changes and establish lifelong health-related behaviors (Arnett, Zukauskiene, & Sugimura, 2014). Emerging adulthood is characterized by changes in residence, relationships, financial status, and education (Arnett, 2000) and emerging adults gain new roles, responsibilities, and freedoms that are associated with increased engagement in a range of health-risk behaviors, including tobacco use (Lipari & Van Horn, 2017; SAMHSA, 2017; Wang et al., 2018). The tobacco industry has also targeted emerging adults specifically (Ling & Glantz, 2002) and given this historical targeting along with the legal purchase of tobacco products at 18 years of age in most states, it follows that use of tobacco products increases during the transition from youth to emerging adulthood (Lipari & Van Horn, 2017; SAMHSA, 2017; Wang et al., 2018).

Of concern is that with the rising popularity of alternative tobacco products such as electronic cigarettes (ECIGs), cigars, and hookah (Cullen et al., 2018; Evans-Polce, Lanza, & Maggs, 2016; Harrell, Naqvu, Plunk, Ji, & Martins, 2017), as well as openness to using such products among emerging adults (Mays et al., 2016), experimentation may lead to continued use and transitions between product types (Best, Haseen, Currie, Ozakinci, Mackintosh, & Stead, et al., 2017; Chaffee, Watkins, & Glantz, 2018; Cooke et al., 2016; Mays et al., 2016). Additional concerns arise when considering polytobacco use (i.e., use of two or more products concurrently), which is prevalent among emerging adult tobacco users (30.0–66.4%; Lisha, Thrul, & Ling, 2019; West et al., 2019), and may lead to more detrimental effects on the developing brain (Yuan, Cross, Loughlin, & Leslie, 2015) and later cessation attempts (Messer et al., 2015). Thus, prevention efforts need to be developed to reduce tobacco initiation among this age group. Although identifying individual risk factors are critical for these efforts (Dierker, Avenevoli, Goldberg, & Glantz, 2004), little is known about those factors outside of demographic (e.g., gender, race; Erickson, Lenk, & Forster, 2014; Evans-Polce et al., 2016; Harrell et al., 2017; Lisha et al., 2019) and tobacco-related characteristics (e.g., dependence level, access; Harrell et al., 2017; Lisha et al., 2019; Yu, Sacco, Choi, & Wintemberg, 2018).

Two potentially relevant factors may be personality and impulsivity given that these characteristics predict a multitude of health behaviors and risk for mortality (Bogg & Roberts, 2004; Hampson & Friedman, 2008; Turiano, Chapman, Gruenewald, & Mroczek, 2015). The Big Five model of personality includes the following characteristics—conscientiousness (e.g., goal-directed and in control), agreeableness (e.g., not hostile), neuroticism (e.g., emotional instability), extraversion (e.g., outgoing and sociable), and openness to experience (e.g., creative; McCrae & Costa Jr., 1997). Adolescent and adult cigarette smokers are significantly less conscientious, less agreeable, more neurotic, more extraverted, and more open to experience compared to non-smokers (Harakeh, Scholte, de Vries, & Engels, 2006; Malouff, Thorsteinsson, & Schutte, 2006; Terracciano & Costa Jr., 2004; Turiano, Whiteman, Hampson, Roberts, & Mroczek, 2012; Zvolensky, Taha, Bono, & Goodwin, 2015). Fewer studies have investigated the associations between personality and alternative product use, although existing work is in line with that reported for cigarette smokers. Specifically, users of smokeless tobacco are more extraverted and neurotic (Foreyt, Jackson, Squires, Hartung, Murray, & Gotto Jr., 1993), whereas cigar users are less conscientious (Brikmanis, Petersen, & Doran, 2017), than are nonusers of these products.

Impulsivity is defined as a tendency to act out in response to impulses and overlaps with some aspects of neuroticism and conscientiousness (Hofmann, Friese, & Strack, 2009; Um, Hershberger, Whitt, & Cyders, 2018). Impulsivity is a multidimensional construct that can be broken down into specific domains: lack of premeditation (i.e., acting without thinking), lack of perseverance (i.e., inability to remain focused), positive urgency (i.e., acting rashly in response to positive affective state), negative urgency (i.e., acting rashly in response to negative affective state), and sensation seeking (i.e., seeking out novel/thrilling experiences; Cyders, Smith, Spillane, Fischer, & Annus, 2007). Individuals that are more impulsive tend to engage in behaviors that provide short-term rewards at the expense of long-term benefits, including use of tobacco products. Specifically, lack of premeditation (Brikmanis et al., 2017), greater positive and negative urgencies (Doran & Trim, 2015; Leventhal et al., 2016), and greater sensation seeking (Brikmanis et al., 2017; Doran & Trim, 2015) are associated with cigarette, cigar, and ECIG use among emerging adults and adolescents. Polytobacco users may also be more impulsive than single product users (Leventhal et al., 2016).

Associations between personality/impulsivity characteristics and tobacco use are studied primarily using variable-centered approaches, in which relations are quantified using separate models. Person-centered approaches have the advantage of assessing how tobacco use behaviors interact and may better reflect the complexity of such behavior. One type of person-centered approach is Latent Class Analysis (LCA), which models latent subgroups within a population by clustering participants into classes based on their responses to specific questions. With the identification of tobacco use classes, various individual difference factors can be used to predict group membership. Understanding how these factors predict classes of tobacco use may aid in identifying those that are most likely to engage in the most problematic profiles of tobacco use (e.g., use of multiple products) and provide insight into which factors to potentially target during prevention efforts. Thus, the present secondary data analysis was designed to 1) use exploratory LCA to identify classes of emerging adult tobacco users and 2) to determine whether demographics, personality, and/or impulsivity characteristics predict class membership identified through exploratory LCA.

2. Materials and methods

3. Results

4. Discussion

This secondary data analysis was conducted to identify classes of tobacco users among emerging adults and to subsequently assess whether personality or impulsivity characteristics are associated with class membership. Consistent with prior work (Cooke et al., 2016; Erickson et al., 2014), a majority of the emerging adults in the current sample reported limited lifetime or current product use, and LCA combined these participants into the largest class (78.1% Nonusers). There was sufficient variability in product use patterns for the remaining 21.9% of the sample, leading to the identification of two discrete classes of users. The largest of these classes was Polytobacco Users (12.1%), followed by Experimenters (9.9%). High probabilities of lifetime ECIG use were evident in both classes, suggesting that experimentation with ECIGs is common among emerging adults (Best et al., 2017; Cooke et al., 2016; Mays et al., 2016). Notable is that cigarette use is declining among emerging adults accompanied by an increase in use of alternative products (Cullen et al., 2018; Evans-Polce et al., 2016). Indeed, lifetime cigarette use was more prevalent for Experimenters than Polytobacco Users in the current study, though current use of cigarettes was uncommon for both classes. Instead, those in the Polytobacco Users class engaged in higher rates of alternative product use, including use of smokeless tobacco, waterpipe, large cigars, and small cigars/cigarillos.

The three classes identified by LCA were also diverse in regard to demographic characteristics. Not surprisingly, males were at increased odds of being Polytobacco Users as compared to Nonusers and Experimenters. These findings replicate a large body of research demonstrating that males are more likely to use smokeless tobacco (Agaku & Alpert, 2016; Erickson et al., 2014) and cigars (Agaku & Alpert, 2016; Phillips et al., 2017), and also to engage in polytobacco use (Phillips et al., 2017; Soneji, Sargent, & Tanski, 2016), relative to their female counterparts. Participants of higher SES were also less likely to be Experimenters and more likely to be Nonusers, though other differences in SES between classes were not observed. Among adolescent samples, use of traditional tobacco products (cigarettes, smokeless tobacco) is associated reliably with lower SES (Wellman et al., 2016; Wellman et al., 2018). Findings for other products (little cigars, cigarillos, ECIGs) are mixed, however (Krishnan-Sarin et al., 2019; Simon et al., 2017). In contrast to this previous work, our measure of SES was subjective (Goodman et al., 2001). A myriad of assessments exist for SES (Shavers, 2007), and more objective measures, such as parental income, may have revealed differences between non- and current users. Still, emerging adults may not be the best reporters of parental income (Svedberg, Nygren, Staland-Nyman, & Nyholm, 2016).

After controlling for demographic factors, personality and impulsivity characteristics were associated significantly with class membership. With regard to personality, agreeableness, conscientiousness, neuroticism, and extraversion varied across classes, indicating that such individual-difference factors play a role in initiation and continuation of nicotine/tobacco product use (Foreyt et al., 1993). Consistent with studies using variable-centered analytic approaches (Harakeh et al., 2006; Malouff et al., 2006; Terracciano & Costa Jr., 2004), individuals scoring lower on agreeableness were more likely to be in classes associated with current product use (Polytobacco Users) and less likely to be Nonusers or Experimenters. There are significant social pressures against experimenting with and continuing to use nicotine/tobacco products (Castaldelli-Maia, Ventriglio, & Bhugra, 2016), and individuals who score lower on agreeableness tend to be rebellious, with lower needs for social approval than those scoring higher on this trait (Terracciano & Costa Jr., 2004).

Arguably the most consistent finding in the personality/tobacco literature (Harakeh et al., 2006; Malouff et al., 2006; Zvolensky et al., 2015), individuals scoring higher on conscientiousness were more likely to be Nonusers than Experimenters or Polytobacco Users in the current study. Those that are more conscientious tend to take fewer health risks, engage in more beneficial health-related behaviors (e.g., healthy diet, regular exercise), and consequently live longer than their counterparts (Bogg & Roberts, 2004; Hampson & Friedman, 2008). Further, two facets of impulsivity were associated with tobacco class membership, with Nonusers scoring lower than Experimenters and Polytobacco Users on measures of negative urgency and lack of premeditation. Given that negative urgency and premeditation overlap with aspects of conscientiousness (Hofmann et al., 2009; Um et al., 2018), these results highlight the critical role of self-control in abstinence from nicotine/tobacco use.

Higher levels of neuroticism and extraversion are consistently associated with cigarette smoking and smokeless tobacco use among adolescent, adult, and college student samples (Foreyt et al., 1993; Harakeh et al., 2006; Malouff et al., 2006; Zvolensky et al., 2015). In this vein, Polytobacco Users and Experimenters scored higher on measures of extraversion and neuroticism, respectively, compared to Nonusers in the current study. It is possible that those scoring higher on these traits experiment with or use nicotine/tobacco for stimulation or to reduce tension/anxiety (Eysenck, 2012; Gonzalez, Zvolensky, Vujanovic, Leyro, & Marshall, 2008). Of course, reasons for experimentation or use cannot be determined from the present data and future work will be needed to evaluate this possibility. Notably, openness to experience did not predict latent class membership in the present study. The relation between openness and nicotine/tobacco use may be unreliable (Terracciano & Costa Jr., 2004; Zvolensky et al., 2015), and thus requires further systematic study, especially with the availability of new tobacco products such as ECIGs.

Together, findings regarding personality and impulsivity highlight that distinct types of emerging adults belong to each tobacco use class. Given that individual-difference factors play an important role in the effectiveness of prevention and intervention efforts (Dierker et al., 2004), different classes of individuals may require different strategies. Although it was once thought that personality and impulsivity characteristics are stable across the lifespan (McCrae & Costa Jr., 1997; Odum, 2011), recent interventions were developed to change these directly in an effort to improve various health-related behaviors (e.g., Hudson & Fraley, 2015; Roberts et al., 2017; Stein et al., 2016). For instance, episodic future thinking asks individuals to imagine future events to increase the value of delayed consequences and reduce impulsivity (Atance & O'Neill, 2001). Such an intervention is successful in reducing cigarette consumption, at least in the short term (Stein et al., 2016). For personality, self-directed goal setting may be adequate for those that want to change their personality traits, while others may require behavioral or cognitive intervention (Hudson & Fraley, 2015; Roberts et al., 2017). Indeed, the sociogenomic model of personality suggests that small modifications in behavioral, attitudinal, or emotional changes can eventually become habitual and lead to lasting change (Roberts & Jackson, 2008). Assuming that changes in personality/impulsivity translate to improvements in health-related behaviors, grouping individuals together into classes of tobacco use patterns may hold promise for tailoring interventions to tobacco use prevention. Thus, it may be important for colleges and universities to have different types of prevention and intervention programs in place to tailor treatment to specific types of emerging adults. There may also be utility in pre-testing college students and using this information to appropriately target tobacco prevention and intervention programs.

Although the current study provides valuable insight into the complexity of tobacco use behaviors among a sample of emerging adults, results must be interpreted in light of some important considerations. First, the sample included participants that were from a single large, public university and were predominantly white/Caucasian. However, the majority of participants came from areas of West Virginia, Ohio, and Pennsylvania that encompass the greater Appalachian region. This region is notable due to disproportionately high rates of cigarette and smokeless tobacco use (American Lung Association, 2015) as well as the substantial increases in ECIG, hookah, and polytobacco use since 2013 (WV DHHR, 2015). With the rate of tobacco industry marketing expenditures being nearly 12 times higher than the national rate (Campaign for Tobacco Free Kids, 2019), combined with annual funding for tobacco control efforts at only ~17.8% of the CDC-recommended amount (CDC, 2019), future studies need to further explore such at-risk populations to understand factors that predict initiation and continuation of tobacco use.¹ Moreover, the current study focused on the transition into college and thus, participants fell primarily on the lower end of the age range classified as emerging adults. Future research should examine preditors of tobacco use among a wider age range of university-attending and non-university-attending emerging adults, given that tobacco use of those who attend college may differ from those who do not (Lenk et al., 2012).

Although these data were collected relatively recently (2016), tobacco use patterns change quickly (Delk et al., 2019) and the cross-sectional nature of the data do not allow for assessing changes in class membership over time. In addition, a restricted number of product types and levels of use were assessed in the current study. For product types, the item used to assess smokeless tobacco (snuff/dip/chew/snus) use was aggregate in nature; what type(s) of smokeless tobacco products were more popular (traditional (snuff/chew) or newer (snus)) cannot be discerned from the present data. With regard to frequency, lifetime use was evaluated only for cigarettes and ECIGs whereby current use was assessed for all product types. However, as mentioned previously, this secondary data analysis included all available tobacco use items. Additional questions about lifetime use of other alternative products may further divide the Experimenters class into groups of individuals that experiment with specific products. In future studies, it will be important to tease apart more nuanced differences between classes.

Another consideration is that questions regarding ECIG use referred to such devices as “electronic cigarettes” and included example images of first- and second-generation devices only. It is possible that participants responded “no” to these questions based on their familiarity with alternative language (e.g., “vape”; Hinds 3rd et al., 2016) and/or newer devices (e.g., mods, Juul; Barrington-Trimis et al., 2018), leading to underestimation of lifetime/current ECIG use among the current sample. Moreover, these data were collected right before individuals started college; it is possible that tobacco use increased when individuals were out of their homes and exposed to settings where tobacco use is more likely to occur (Lipperman-Kreda, Paschall, Robert, & Morrison, 2018). Finally, relatively small mean and standard deviation values for several personality and impulsivity traits suggest that the majority of participants scored toward the lower end of the scales (e.g., range of 1.70 to 1.95 for negative urgency on a scale from 1 to 4) and that there was little variability between participants. Thus, it will be important to replicate these findings with participants that produce a wider range of values and score at the higher end of these scales. However, the fact that many of these characteristics still significantly predicted latent class membership, despite their restricted range, indicates that they are likely strong indicators of tobacco use patterns.

The present results describe an important role for personality and impulsivity in nicotine/tobacco use behavior among emerging adults. Individuals that engage in use of multiple tobacco products are also at greater risk for use of other substances, such as marijuana and alcohol (Bernstein et al., 2015; Pulvers et al., 2018), and the associations between demographics, personality, or impulsivity and tobacco use observed here are consistent with those reported for other drugs (Chuang et al., 2017; Dash et al., 2019; Slade et al., 2016). Historically, focus has been toward preventing uptake of tobacco use among adolescents; however, given the relatively high prevalence of tobacco initiation during the transition from youth to emerging adulthood (Lipari & Van Horn, 2017; SAMHSA, 2017; Wang et al., 2018), prevention and intervention techniques need to be developed for emerging adults specifically. Because personality and impulsivity characteristics are amenable to change (Hudson & Fraley, 2015; Roberts et al., 2017), it is possible to tailor prevention to those that are most at risk for tobacco use.