Using Latent Profile Analysis to Assess College Students' Attitudes about Underage Drinking and Prescription Stimulant Misuse.
Misusing alcohol can result in significant negative academic consequences, including poor performance on an exam or missing class due to a hangover. In addition to the potential negative consequences that impact students' academics, however, there are a number of other serious consequences related to alcohol misuse. The National Institute on Alcohol Abuse and Alcoholism (NIAAA) estimates that over 1500 college students die each year due to the effects of alcohol, 700,000 students are assaulted by a peer who has been drinking, and almost 100,000 instances of sexual assault related to alcohol use occur on college campuses (NIAAA, 2019).
Although the misuse of alcohol by college students is not new, the misuse of prescription stimulants represents a more recent concern. Typically used for the management of ADD and ADHD, stimulants including methylphenidate (Ritalin) and dextroamphetamine (Adderall) are increasingly being consumed by students without a medical prescription, primarily for use as a "study aid," and most frequently obtained from peers having a prescription for the drug. Although estimates of misuse vary from study to study, a recent meta-analysis reported that 17% of college students misuse stimulants prescribed for another individual (Benson, Flory, Humphreys & Lee, 2015), with White males, members of the Greek system, and students using other substances most likely to misuse stimulants (Arria, Wilcox, Caldeira, Vincent & Garnier-Dykstra & O'Grady, 2013; Dussault & Weyandt, 2013; McCabe, 2008; Rabiner, Anastopoulos, Costello Hoyle, McCabe & Swartzwelder, 2009).
Because stimulants increase vigilance and attention and decrease the need for sleep, students are able to stay awake for longer periods of time to study for an upcoming exam or work on a project or paper with a looming deadline. Other reasons for misusing stimulants include weight management and for recreational purposes, often combining the drug with other substances (Benson et al., 2015). In addition to concerns over students sharing their prescribed medications with other students, the misuse of these drugs can result in a host of potentially harmful consequences, including cardiac irregularities, delirium, physical dependency, and hallucinations (Morton & Stockton, 2000).
Despite the prevalence of alcohol and prescription stimulants on college campuses, not enough work has been done investigating student attitudes regarding alcohol or prescription stimulant misuse, how the reasons for using alcohol or stimulants affect attitudes about the drug being used, or how attitudes might be affected by the frequency with which someone uses alcohol or prescription stimulants.
The present preliminary study examined attitudes about underage alcohol and prescription stimulant misuse in a sample of undergraduate students using fictional scenarios. Because the dangerous consequences of alcohol misuse are often seen more in underage students as the adjustment to college is taking place (NIAAA, 2019), we chose to focus on underage alcohol misuse in our study. We predicted that although both men and women in our study would view underage drinking and prescription stimulant misuse as illegal, males would view underage alcohol and prescription stimulant misuse as less of a concern and would view their use as more justified than would women in our study.
Eighty-one college undergraduates (71% female) participated in the study. Eighty-three percent of our sample was either first- or second-year students. Fifty eight percent of our sample identified as Caucasian, 18% as Asian, 11% as Black, 11% as LatinX and two percent as "other." Participants were given course credit for their participation.
Eight scenarios were created that varied by gender of the actor (male, female), drug being used (underage alcohol, prescription stimulant), and reason for use (increased sociability or stress reduction for alcohol; study aid or weight management for stimulant). Each participant completed four of the eight created scenarios, randomly assigned to include two of each gender, two of each drug (alcohol, prescription stimulant) and two of each reason for either alcohol or stimulant. Scenarios were followed by a series of questions designed to assess attitudes about the content of the scenario. Questions were responded to using a 5-point Likert Scale, with 1 indicating "strongly disagree," 5 indicating "strongly agree" and 3 indicating "no opinion." Questions assessed participants' attitudes regarding the legality of the drug being used, ("x is breaking the law"), the justification for taking the drug ("because of his demanding schedule, x is justified in taking Adderall to help him study for the important exam"), the frequency with which the actor takes the drug ("because x only uses Adderall occasionally, there is no need for concern"), and whether or not characters in the scenario who were abusing prescription stimulants as study aids were cheating ("because x is using the Adderall to help better her/his exam grade, x is cheating"). A sample of one of the scenarios is provided in the Appendix; all scenarios can be obtained by contacting the author.
Approval by the departmental Human Subjects Review Board was obtained prior to participant recruitment or data collection. Participants completed an Informed Consent form upon entering the lab. They were then given a packet containing demographic questions and the scenarios. Upon completion of the scenarios, participants were given a Debriefing Form that explained the purpose of the study and encouraged contacting the experimenter with any questions the participant might have.
This study used Latent Profile Analysis (LPA) in an attempt to find reliable patterns in responses to the scenario questionnaire. LPA is capable of identifying mixtures of distributions of one or more variables.
It is similar to cluster analysis but identifies classes according to a hypothesized model whereas cluster analysis does not (Pastor, Barron, Miller & Davis, 2007). LPA is based on the assumption that some sets of variables are not consistent with a single probability distribution, but rather with a mixture of distributions, thereby accounting for latent or hidden heterogeneity in the data. Thus, a mixture analysis is a person-centered approach used when it is hypothesized that a population contains multiple groups but lacks the information about group membership. It is a categorical latent variable model (in the similar way that factor analysis is a continuous latent variable model) such that the categorical variable is hypothesized to be the source of one or more observed indicators. The categories of this variable are referred to as classes that tend to generate different patterns of responses (or mixtures) for the observed indicators. The mixture analysis estimates parameters for a specified number of these groups (Pastor & Gagne, 2013). To the extent this assumption is correct, a model with two or more classes of participants will fit the data better than just a single distribution.
In the current study, it is hypothesized that a single latent categorical variable can adequately account for the multivariate distribution of the observed indicator variables shown in Table 1. These variables were derived from an exploratory factor analysis (EFA) using Revelle's (2018) psych package in the R programming language (R Core Team, 2019) that suggested four factors for 20 items of the scenario questionnaire. Specifically, the development of the scenarios was based on assessing students' beliefs regarding core dimensions of alcohol and stimulant use. For example, items of the scenarios included beliefs regarding justification of alcohol and stimulant consumption based on reducing stress or social benefits (alcohol) or coping with a demanding schedule or weight loss (stimulants). Other items assessed students' beliefs regarding whether they or others were breaking the law by consuming alcohol or stimulants. Thus, given a relatively large number of items specifying a smaller number of hypothetical dimensions, an EFA was needed to discover the nature and number of factors underlying the responses to the 20 items of the scenarios.
The items upon which the LPA indicators are based are shown in Table 2. Four of the LPA indicators were created as unit-weighted sums of the items that make up each respective factor defined by the factor loadings shown in Table 1. These factors, with categorical omega values ([[omega].sub.c]) and 95% confidence intervals (Kelley, 2017; Kell & Pornprasertmanit, 2016) as measures of internal consistency, are shown in Table 1. An item assessing an attitude of cheating and two items assessing the belief that students in the scenarios have an alcohol problem had low weights on the four factors and were thus included as separate item indicators in a latent profile analysis discussed below. Note that the responses to the two items assessing perceptions of alcohol problems were summed to make a single indicator.
Following Nylund, Asparouhov, and Muthen (2007), evaluation of models with different numbers of classes was done using the Bayesian Information Criterion (BIC), Lo-Mendell-Rubin (LMR), and the parametric bootstrap likelihood ratio (BLRT) tests. The results, shown in Table 3, suggest that a two-class solution fit the data best. Specifically, the minimum BIC occurs for two classes and the LMR and BLRT tests are not statistically significant in the comparison of three versus two classes, suggesting that the more parsimonious two-class solution should be retained. In addition, a four-class solution produced multiple errors related to the fact that the fourth class contained only 3 students. This suggested that the four-class solution extracted too many classes. Thus, given the superiority of the BIC criterion (Nyland, et al, 2007) combined with the result the LMR and BLRT tests suggested no difference between the two- and three-class solution, the two-class solution was retained. Finally, the entropy of the two-class solution suggested that there was an adequate degree of separation between the classes for the two-class solution.
Data are shown in Table 4 and are plotted in Figure 1. Group 1 disagreed that students are cheating when abusing a prescription stimulant for studying purposes (mean = 2.0/5) and they had no opinion regarding justification for using alcohol or stimulants, or that students were breaking the law if misusing alcohol under the legal age or abusing stimulants for either weight loss or studying advantage (means = 3.0/5). Group 2, however, tended to disagree (approaching "strongly disagree") with respect to justifying stimulant or alcohol abuse, and tended to agree that students in the scenarios are breaking the law.
The first class "Alcohol or stimulant use is not necessarily a concern" is the smaller of the two groups (n = 14) and is composed of seven men. The second class, "Alcohol or stimulant use may be a concern" (n = 67) is comprised primarily of women (n = 51). The odds that women are in this group are approximately 3.2 times greater than the odds that women are in the first group. While this is a large odds ratio, it is not statistically significant (p =.233). This lack of significance would suggest that the sex of the participant has no effect on attitudes; however, the small sample size confounds the interpretation of the influence of sex on attitudes. Although the results seem to suggest that women may be more cautious in their outlook on alcohol or prescription stimulant abuse, further research is needed to substantiate this hypothesis.
The present preliminary study examined attitudes about underage alcohol and prescription stimulant misuse in a sample of undergraduate students. Fictional scenarios were created that varied by gender of the actor, drug, and reason for use. Following each scenario, participants answered a series of questions that asked about whether the actor was seen as breaking the law, cheating, or if the reason for use was justified. Using Latent Profile Analysis, two distinct groups of participants were identified based on their responses to the questions.
The first group had no opinion regarding the legal consequences of underage drinking or misusing prescription stimulants and disagreed that students were cheating when using non-prescribed stimulants as "study aids" (Table 3, Figure 1). These data offer the possibility that as prescription stimulant misuse becomes more prevalent on college campus (e.g., Benson et al., 2015), the drug may be perceived in the same way as is underage alcohol; that is, as part of the "normal" college experience and as a rite of passage for college students. Related to this possibility is previous research that found that, despite students' knowledge of campus regulations regarding alcohol use, the majority of students disregarded these policies, with close to 80% of students drinking on campus (Marshall, Roberts, Donnelly & Rutledge, 2011).
Group 2, the larger of the two groups, comprised primarily of women, showed a different set of responses to the scenario questions. This group recognized that students engaging in underage alcohol consumption or using a friend's prescribed stimulant were breaking the law and strongly disagreed with the justification for using prescription stimulants for weight gain or as "study aids." It is noteworthy that women comprised the majority of this group and expressed a more cautious view regarding underage alcohol use and prescription stimulant misuse. The suggestion that women may be more cautious in their views of stimulant misuse must be considered carefully, however, as more work is needed to confirm this observation. However, in light of previous research supporting gender differences in risk behavior engagement and the finding that males misuse prescription stimulants more than women (Arria et al., 2013; Benson et al., 2015; Dussault & Weyandt, 2013), the present preliminary study offers further support for males having views that are less cautious concerning underage alcohol and prescription stimulant misuse and that these attitudes might translate into risky health behaviors. When taken collectively, these data might offer some direction for colleges concerned about alcohol misuse on campus to take by offering programming and educational opportunities targeting male students prior to beginning the first year of studies.
We used Latent Profile Analysis to identify multiple distributions within our sample. The analysis is useful to identify homogenous subgroups of participants in a larger heterogenous sample. The present preliminary study employed Latent Profile Analysis to identify two groups of college students, each with differing attitudes regarding underage alcohol consumption and prescription stimulant misuse.
Our study was not without limitations, however. The participants in our study were students enrolled at a small midwestern college; caution must be taken, therefore, about generalizing our data to the larger population of college students. Future research might utilize scenarios that vary the gender of the actor, the particular drug being misused and reason for use to assess attitudes and look for relationships between attitudes and actual behavior of participants as related to underage drinking and misuse of prescription stimulants. In addition, future work might examine other demographic variables (race, for example) with regard to the misuse of alcohol and prescription stimulants.
Arria, A.M., Wilcox, H.C., Caldeira, K.M, Vincent, K.B., Garnier-Dykstra, L.M. & O'Grady, K.E. (2013). Dispelling the myth of "smart drugs": Cannabis and alcohol use problems predict nonmedical use of prescription stimulants for studying. Addictive Behaviors, 38, 1643-1650.
Benson, K., Flory, K., Humphreys, K.L.& Lee, S.S. (2015). Misuse of stimulant medication among college students: A comprehensive review and metaanalysis. Clinical Child and Family Psychology Review, 18, 50-76.
Dussault, C.L. & Weyandt, L.L. (2013). An examination of prescription stimulant misuse and psychological variables among sorority and fraternity college populations. Journal of Attention Disorders, 27, 87-97.
Kelley, K. (2017). MBESS: The MBESS R Package. R package version 4.4.0. https://CRAN.R-project.org/package=MBESS
Kelley, K., & Pornprasertmanit, S. (2016). Confidence intervals for population reliability coefficients: Evaluation of methods, recommendations, and software for composite measures. Psychological Methods, 21 (1), 69 - 92. doi: 10.1037/a0040086
Marshall, B.L., Roberts, K.J., Donnelly, J.W. & Rutledge, I.N. (2011). College student perceptions on campus alcohol policies and consumption patterns. Journal of Drug Education, 41, 345-358.
McCabe, S.E. (2008). Misperceptions of non-medical prescription drug use: A web survey of college students. Addictive Behaviors, 33, 713-724.
Morton, W.A. and Stockton, G.G. (2000). Methylphenidate abuse and psychiatric side effects. Primary care companion: Journal of Clinical Psychiatry, 2-5.
National Survey on Drug Use and Health (2017). https://www.samhsa.gov
NIAAA (2019). National Institute of Alcohol Abuse and Alcoholism. U.S. Department of Health and Human Services.
Nylund, K. L., Asparouhov, T., & Muthen, B. O. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14(4), 535 569. doi: 10.1080/10705510701575396
Pastor, D. A., & Barron, K. E., Miller, B. J., & Davis, S. L. (2007). A latent profile analysis of college students' achievement goal orientation. Contemporary Educational Psychology, 32, 8 - 47. doi: 10.1016/j.cedpsych. 2006.10.003
Pastor, D. A., & Gagne, P. (2013). Mean and covariance structure mixture models. In G. R. Hancock & R. O. Mueller (Eds.), Structural equation modeling: A second course (2nd Ed.), (pp. 343 - 394). Charlotte, NC: Information Age Publishing.
Rabiner, D.L., Anastopoulos, A.D., Costello, E., Hoyle, R.H., McCabe, S. & Swartzwelder, H. (2009b). Motives and perceived consequences of nonmedical ADHD medication use by college students: Are students treating themselves for attention problems? Journal of Attention Disorders, 13, 259-270.
R Core Team (2019). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www. Rproject.org/.
Revelle, W. (2018) Psych: Procedures for Personality and Psychological Research, Northwestern University, Evanston, Illinois, USA, https:// CRAN.R- project. org/package=psych Version = 1.8.12.
Schwartz, G. (1978). Estimating the dimension of a model. The Annals of Statistics, 6, 461-464.
Sample scenario *
Sarah is a first-year student in college who is having a difficult time fitting in socially. She is rather shy and has always found it hard to make friends. During the second week of the semester, Sarah's roommate mentions a party being held on campus by a fraternity over the weekend and suggests that they go together to try to meet new people. Although Sarah has never consumed alcohol, she drinks a few beers at the party to fit in with the other students who are also drinking. She finds that she is less shy and more able to talk with other students after consuming a few beers. Although she sometimes gets sick after drinking and doesn't always enjoy the taste of alcohol, Sarah continues to attend campus parties throughout her first year of college to feel "socially connected" and to meet new friends. Because of occasional hangovers on Sunday, Sarah finds she is not able to study as much as she would like to; in fact, twice, she has earned "D"s on Monday exams due to her lack of studying on Sunday.
Use the following scale to answer each of the following:
1: strongly disagree
3: neither agree nor disagree
5: strongly agree
--1. Since alcohol helps Sarah meet new people and feel more connected in college, it's okay that she drinks on the weekends
--2. Sarah has an alcohol problem
--3. The benefits of meeting people and feeling more comfortable in social settings outweigh the periodic negative effects Sarah suffers due to her weekend drinking
--4. Sarah is breaking the law
--5. The fraternity members providing alcohol are breaking the law
Note: All scenarios are available upon request from the first author.
Susan Kennedy &, Michaela Allen
Gary J. Kennedy
The Ohio State University
Author info: Correspondence should be sent to: Susan Kennedy, Ph.D., Psychology Dept., Denison University, Granville, OH 43023 Kennedys@Denison.edu
Caption: Figure 1: Means of Profile Indicator by Latent Profile
Table 1: Scenario Items and Factor Loadings by Latent Profile Indictor Derived Latent Scenario Factor Factor Factor Factor Profile Indicators Items 1 2 3 4 1. 0.51 0.43 0.03 -0.13 Justification for 2. 0.37 0.25 0.08 -0.31 taking stimulants 3. 0.53 0.19 -0.07 0.06 4. 0.73 0.00 -0.06 -0.01 5. 0.94 -0.15 0.03 -0.01 6. 0.03 0.50 0.16 -0.10 Justification for 7. -0.02 0.77 0.09 -0.07 drinking 8. 0.24 0.48 -0.17 0.12 9. -0.06 0.71 -0.07 0.15 10. 0.00 0.04 0.80 0.13 Breaking the law 11. -0.06 0.00 0.87 0.07 (alcohol) 12. 0.05 -0.04 0.77 0.01 13. 0.05 -0.01 0.55 0.15 14. -0.06 0.03 0.08 0.79 Breaking the law 15. -0.09 -0.02 0.06 0.82 (stimulants) 16. 0.03 -0.02 0.16 0.81 17. 0.04 0.01 0.04 0.82 Problem with 18. -0.41 -0.02 0.20 -0.28 Alcohol 19. -0.05 -0.40 0.23 -0.13 Consumption 20. -0.20 -0.21 0.08 0.10 Cheating Derived Latent Scenario [h.sup.2] Categorical Profile Indicators Items Omega 1. 0.62 Justification for 2. 0.38 0.854 taking stimulants 3. 0.38 (.732, .935) 4. 0.55 5. 0.82 6. 0.27 Justification for 7. 0.57 .785 drinking 8. 0.41 (.670, .924) 9. 0.51 10. 0.75 Breaking the law 11. 0.83 .877 (alcohol) 12. 0.60 (.751, .924) 13. 0.39 14. 0.71 Breaking the law 15. 0.76 .932 (stimulants) 16. 0.78 (.871, .964) 17. 0.69 Problem with 18. 0.21 N/A Alcohol 19. 0.24 N/A Consumption 20. 0.15 N/A Cheating Table 2: Legend for Scenario Items 1. The student is justified in taking Adderall to help him/her study for the important exam. 2. The student is justified in taking Ritalin to help his/her weight loss goal. 3. Because the student is only using Adderall one time, there is no need for concern. 4. Because the student uses Ritalin occasionally, there is no need to concern. 5. Ritalin is a relatively safe drug for the student to use, even though he/she does not have a prescription for it. 6. The benefits of meeting people and feeling more comfortable in social settings outweigh the periodic negative effects the student suffers due to his/her weekend drinking. 7. Since alcohol helps the student to relax, it's okay that he drinks twice a week to cope with the stresses of his busy academic schedule. 8. Since alcohol helps the student meet new people and feel more connected in college, it's okay that she drinks on the weekends. 9. The benefits of relaxing and reducing stress by drinking outweigh the occasional negative effects the student. 10. The student is breaking the law (in an effort to meet new people and feel socially comfortable). 11. The student is breaking the law (in an attempt to cope with stress). 12. The student's friends who drink with the student are breaking the law (in an attempt to fit in socially). 13. The student's friends who drink with the student are breaking the law (in an attempt to cope with stress). 14. The student is breaking the law (in an attempt to better his/her grade). 15. Others who provided Adderall to the student are breaking the law. 16. The student is breaking the law (in an attempt to lose weight). 17. The student's sister who provided the Ritalin is breaking the law. 18. The student has an alcohol problem (in an attempt to fit in socially). 19. The student has an alcohol problem (in an attempt to cope with stress). 20. Because the student is using Adderall to better her exam grade, he/she is cheating. Table 3 Fit Statistics across Classes Fit Statistics 1 Class 2 Classes 3 Classes 4 Classes Log-likelihood -629.51 -551.849 -539.827 -528.518 (number of (200/200) (200/200) (126/200) (8/200) replications) AIC 1287.023 1143.698 1135.655 1129.035 BIC 1320.545 1191.587 1202.700 1215.236 SABIC 1276.394 1128.514 1114.397 1101.704 LMR N/A 52.979 23.378 10.389 LMR p-value N/A 0.024 0.4154 0.2398 BLRT N/A 54.486 24.043 10.685 BLRT p-value N/A <0.0001 0.060 1.0000 Entropy N/A 0.860 0.779 0.879 Table 4 Means of Profile Indicator of the Two Latent Profile Classes Alcohol or Alcohol or Latent Profile Indicators Stimulant Stimulant Concern Use May be Use is Not a Concern Necessarily a concern (C1) (C2) Justification for Taking Stimulants 2.668 1.577 Justification for Drinking 2.959 2.410 Problem with Alcohol Consumption 2.308 2.961 Cheating 1.885 3.058 Breaking the Law--Alcohol 3.151 3.929 Breaking the Law--Stimulants 3.125 4.144
Please Note: Illustration(s) are not available due to copyright restrictions.
|Printer friendly Cite/link Email Feedback|
|Author:||Kennedy, Susan; Allen, Michaela; Kennedy, Gary J.|
|Publication:||North American Journal of Psychology|
|Date:||Dec 1, 2019|
|Previous Article:||Changes in College Students' Awareness, Empathy, and Openness After Multiple Diversity Courses.|
|Next Article:||Evaluating Time Delay and Exposure to Songs with Prosocial Lyrics and Their Effects on Prosocial Behavior.|