|Home | About | Journals | Submit | Contact Us | Français|
Alcohol expectancies, defined as a person’s beliefs about the effects of drinking, can influence alcohol consumption and help predict problem drinking in college students. However, there are concerns that current expectancy measures do not adequately capture mandated student expectations about alcohol use. This study examined the correspondence of 412 self-generated expectancies from mandated students (n = 64) to items on the Brief Comprehensive Effects of Alcohol (B-CEOA; Ham, Stewart, Norton, & Hope, 2005). Self-generated expectancies were reviewed by raters who attempted to match each expectancy with a single B-CEOA item based on the qualitative essence of each statement. Most mandated student expectancies were not represented by the B-CEOA. All expectancies were then classified into 6 categories based on themes and categories from the alcohol expectancy literature. Mandated student expectancies emphasized the physiological aspects of drinking, whereas the B-CEOA assesses expectancies about intrapersonal factors. The findings suggest the B-CEOA may exclude alcohol expectancies that are important and relevant to this population. Self-generated alcohol expectancies from the target population should be considered when developing or administering expectancy questionnaires.
Alcohol expectancies are defined as an individual’s beliefs regarding the effects of drinking (Fromme, Stroot, & Kaplan, 1993; Leigh & Stacy, 1993). Expectancies have been shown to both shape and predict individuals’ drinking behaviors (Burden & Maisto, 2000; Christiansen, Smith, Roehling, & Goldman, 1989; Ham, Zamboanga, Olthuis, Casner, & Bui, 2010). However, there is debate about whether to capture personal expectancies through standardized or self-generated measures (Burden & Maisto, 2000; Noar, LaForge, Maddock, & Wood, 2003; Zamboanga, 2006). In addition, measures were created with volunteer students who tend to drink less heavily than mandated students (i.e. students required to receive an intervention after violating campus alcohol policies, see Barnett, et al., 2004).
The current project compared self-generated expectancies produced by mandated college students to items on the Brief Comprehensive Effects of Alcohol (B-CEOA; Ham, et al., 2005). The B-CEOA is commonly used in brief alcohol interventions for heavy drinking college students (e.g., Brief Alcohol Screening and Intervention for College Students; Dimeff, Baer, Kivlahan, & Marlatt, 1999) to assess and discuss mandated students’ alcohol beliefs (e.g., Barnett, Murphy, Colby, & Monti, 2007; Borsari, et al., in press). The B-CEOA consists of 15 items representing both positive and negative expectancies (see Table 1) and was developed by extracting items from the original CEOA with the highest loadings (Fromme, et al., 1993; Ham, et al., 2005).
We conducted two distinct content comparisons. First, to gauge the overlap between the two lists, raters reviewed each self-generated expectancy to determine if it matched qualitatively to any item found on the B-CEOA. Second, the self-generated expectancy list was examined for themes and grouped into explicit factors to evaluate whether the overall focal points of the students’ and B-CEOA expectancies diverge. We also explored the summation of the students’ positive and negative expectancies.
This project is a secondary data analysis from a multi-site study implementing brief motivational interventions with mandated students (see Borsari & Carey, 2005). Eligibility criteria consisted of scoring 10 or more on the Alcohol Use Disorders Identification Test (AUDIT; WHO Brief Intervention Study Group, 1996) and/or 2 or more binge drinking episodes in the past 30 days. All students who participated in the study reported 2 or more binge drinking episodes, and 48% reported a score of 10 or higher on the AUDIT (M = 7.5, SD = 4.3).
Sixty four participants (82% male, 92% Caucasian, 78% freshmen, mean age 19 years old) completed paper and pencil baseline, 3- and 6-month expectancy assessments at the research offices. Participants could write additional expectancies at each follow-up. Expectancies from the students’ final assessment were reviewed in this study. All study procedures were approved by the University IRB.
This open-ended measure required students to produce a personalized list of beliefs about the effects of alcohol (Leigh & Stacy, 1994). The directions specified, “On the following lines, we would like you to please list as many outcomes or effects (positive and negative) as you can think of that you associate with alcohol use.” Participants could provide up to 10 expectancies and rate them on 6-point scales: (a) the likelihood of its occurrence, (b) its valence (on a scale from −3 to +3, with no neutral point), and (c) its personal importance.
Three independent raters reviewed the students’ self-generated expectancies to determine if the statements matched an item on the B-CEOA. Because the students’ statements rarely aligned perfectly with the B-CEOA items, the raters were given a priori instructions. First, if the general meaning of the expectancy was deemed to be the same by the rater, the statements were counted as having been covered by the B-CEOA (e.g., to a rater “get into fights” would mean the same as “I would act aggressively”). Second, the raters were instructed not to read into the meaning of a participant’s statement. That is, raters could not be certain whether the student’s perception of how they ‘get crazy’ would align with the B-CEOA items (e.g., “I would be loud, boisterous, or noisy”).
Following the instructions, raters coded whether statements were similar to an item on the B-CEOA. Raters picked the one B-CEOA item the expectancy best represented and assigned it that particular B-CEOA code. If the expectancy could not justly fit under any of the items, the raters coded it as ‘other’. When codes were in disagreement, the statement was categorized by majority consensus of the 3 raters.
The self-generated expectancies were compared to the B-CEOA items by classifying all expectancies into 6 factor categories. To do so, the authors used card sorting techniques in combination with knowledge of expectancy factors typically identified in the literature to create the factor categories. Many of the self-generated expectancies corresponded to factors previously found in the expectancy literature (e.g., physiological and interpersonal outcomes). However, since students often reported expectations unique to the college environment (e.g. missing class, getting in trouble with the school), an academic and legal category was created to be more inclusive of their unique experiences. Rather than generate several smaller categories for the low frequency statements, the ‘other’ category was created to capture the few responses that could not fit into the rest of the factors. Formal definitions of the factors were then created and refined by consensus of the authors, with the final six factors identified as intrapersonal, physiological, sexual, interpersonal, academic and legal, and other outcomes of drinking alcohol. Next, the self-generated expectancies were categorized into those 6 factors by three independent raters who were not part of the factor definition process. Finally, the fifteen items of the B-CEOA were categorized using those same 6 factors.
First, we established rater reliability of our expectancy codes and category groupings by calculating percent agreement and Kappa reliability. Second, we compared how the self-generated items correspond to the B-CEOA items by reporting the degree of overlap between the two lists. Third, we compared the percentage of each list (self-generated and the B-CEOA) that pertained to each factor category. Lastly, in exploratory analyses, we determined the valence of the self-generated expectancies by reviewing participants’ own evaluations as recorded in the personal expectancy measure.
Because of the complexity involved in the coding scheme, the level of simple agreement between raters was calculated (number of agreements divided by the number of coding decisions). Overall agreement on the coding of the self-generated items was good, and ranged from 76% to 86%.
Three hundred and twenty-eight (79.6%) expectancy categorizations were unanimous and 79 (19.1%) had 2 of 3 raters agree. For the remaining 5 (1.3%) for which there was no agreement, further discussion ensued among the co-authors, and final coding was determined by majority consensus. Using Kappa Reliability analyses, the average agreement on the factor was .82 (range of .80–.83, indicative of outstanding agreement; Landis & Koch, 1977). The B-CEOA items were categorized with 100% agreement between raters.
In all, 64 participants generated 412 individual expectancy statements. As identified in Table 1, 74% of the participants’ self-generated statements had no corresponding item on the B-CEOA. Of the statements that did match with a B-CEOA item, 9% of the self-generated statements fell within the parameters of “I would act sociable” with expectancies like “meet new people” and “more friendly”. The next highest B-CEOA item that accounted for the self-generated statements was “I would act aggressively”, which matched approximately 7% (e.g., “fight with girlfriend”, “fight or argue”). No one said alcohol would make them a “better lover”.
Nearly three-quarters of the self-generated expectancies were marked as ‘other’ because they did not qualitatively match any of the statements in the B-CEOA. These statements varied considerably, ranging from vague sentiments such as “get into trouble” to very specific outcomes such as “be unable to rely on designated driver’s condition” or “[fail] out of school.” Many reported sex as an outcome with statements such as “[hooking] up with someone.” Another common theme was the negative physical effects of drinking alcohol (e.g., “throwing up”, “alcohol poisoning”, and “liver damage”).
We summarized the distribution of the self-generated expectancies and the B-CEOA items across the six categories, and then calculated whether the proportion of each factor type on the B-CEOA was similar to the self-generated expectancies of our sample (see Table 2). The participants’ self-generated expectancies were primarily concerned with the physiological effects of alcohol consumption, yet only two B-CEOA items represent the physiological aspects of drinking. About half of the B-CEOA items target intrapersonal factors (47%), compared to approximately 19% of self-generated expectancies. Both of the measures reported equivalent amounts of interpersonal expectancies (20% of B-CEOA items, 24% self-generated expectancies).
Analysis of valence evaluations (positive vs. negative) showed two-thirds of the self-generated expectancies were perceived as negative (268 negative [65%], 144 positive [35%]). The trend persisted at the individual level of data as well. The modal number of positive expectancies for each participant was one at baseline and the 3-month follow-up, and two at the 6-month follow-up. The modal number of negative expectancies was three at baseline and 3-month follow-up; and four at the 6-month follow-up.
This study compared self-generated expectancies about alcohol use to the items on the Brief CEOA in a sample of mandated college students. One of the more striking results was how the content of self-generated expectancies diverged from the items that comprise the B-CEOA. Specifically, three-quarters of the mandated students’ spontaneous expectations of their alcohol use differed from the items represented on the B-CEOA. Evidently, the mandated student population expects more physiological severe outcomes of alcohol use than are captured by the B-CEOA. In contrast, the B-CEOA focuses primarily on intrapersonal items, or the internal state of the participant. Furthermore, approximately two-thirds of the self-generated expectancies were perceived as negative by the students. In previous research, volunteer students participating for class credit had nearly a 1:1 ratio (Leigh & Stacy, 1994). This discrepancy may be indicative of the heavier drinking commonly found in mandated students (Barnett, et al., 2004).
The findings of this study should be considered in the context of some limitations. First, raters conservatively coded the expectancies of the B-CEOA. Second, some items could have been coded into multiple B-CEOA items, but our approach only allowed for one match to a category. Third, the B-CEOA was not given to the participants in this study, precluding a direct comparison of measures. Finally, due to the small sample size, these results may not be generalizable to all mandated students.
In sum, the expectancies listed in the B-CEOA do not align well with what mandated students list extemporaneously as their alcohol beliefs. That said, since the B-CEOA is predictive of drinking behaviors (Ham, et al., 2005), it remains to be determined what additional value, if any, is derived from the self-generated expectancies. It may be that explicit lists of alcohol expectancies are tapping into similar enough representations that precise replicas of what mandated students consider their expectations are not necessary. Indeed, Read and colleagues (2009) suggest it may not be the specific item or expectancy that matters, but its location within a particular dimension (positive/negative valence and sedation/arousal activation) which determines its activation pathway. So, while the items on the B-CEOA are not exactly what the student would say spontaneously, they may map onto predictive dimensions. The limited scope of this qualitative analysis leaves this issue to future research efforts.
This work was supported by National Institute on Alcohol Abuse and Alcoholism Grants F31 AA057, and R01 AA017874 to B. Borsari and T32-AA07459 to Nadine Mastroleo. The contents of this manuscript do not represent the views of the Department of Veterans Affairs or the United States Government.
Role of Funding Sources
This work was supported by National Institute on Alcohol Abuse and Alcoholism Grants F31 AA057, and R01 AA017874 to B. Borsari and T32-AA07459 to Nadine Mastroleo. NIAAA had no role in the study design, collection, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication. The contents of this manuscript do not represent the views of the Department of Veterans Affairs or the United States Government.
ContributorsAuthors Colleen Peterson and Brian Borsari designed the study and wrote the protocol. Author Colleen Peterson conducted literature searches and provided summaries of previous research studies. Author Colleen Peterson conducted the statistical analysis. Author Colleen Peterson wrote the first draft of the manuscript and all authors contributed to and have approved the final manuscript.
Conflict of Interest
All authors declare that they have no conflicts of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.