The Link Between Planning and Doing: Daily-level Associations between College Students’ Plans for and Use of Alcohol-related Protective Behavioral Strategies

Abstract

Objective:

The current study expands the literature on alcohol-related protective behavioral strategies (PBS) that individuals may use to reduce risk of intoxication and/or consequences. This study used daily data collected prospectively to test (1) the extent to which college students’ plans for using different types of PBS on a given day were associated with actual PBS use and (2) whether drinking intentions moderated the strength of the association between PBS plans and use.

Method:

College students ages 18–24 (N = 189; mean (SD) = 20.16 (1.54) years; 48.68% female; 67.20% White/Caucasian) completed eight consecutive weekends of online daily surveys (2x/day; 83.72% completed) and reported on PBS plans/use and also drinking intentions/use. Eligibility included drinking two days/week in the past month and heavy episodic drinking in the past two weeks. Three PBS subscales were tested in separate multilevel models: limiting/stopping, manner of drinking, and serious harm reduction.

Results:

As hypothesized, for each PBS subscale, afternoon PBS plans were positively associated with use of that type of strategy later that night. Moderation results showed a larger positive association between daily limiting/stopping plans and use of limiting/stopping strategies on days when drinking intentions were elevated compared to days with lower drinking intentions.

Conclusions:

Findings indicated that college students do plan to use PBS ahead of drinking occasions, and when students had stronger than usual plans for PBS, they tended to follow through on their plans. It may be beneficial to enhance students’ PBS plans in interventions by addressing potential barriers to PBS.

Alcohol use remains prevalent among college students, despite recent modest declines in its prevalence. Recent national estimates indicate that about 74.6% of college students drank in the past year and 28.4% engaged in heavy episodic drinking in the past two weeks (Schulenberg et al., 2019). High-intensity alcohol use (8+/10+ drinks for women and men, respectively) is prevalent among young adults (Patrick & Terry-McElrath, 2017, 2019; Patrick et al., 2016). Alcohol use is associated with myriad negative consequences including blackouts, injury, and impaired driving (Hingson et al., 2017a, 2017b). Given relatively modest effect sizes for college student and young adult brief alcohol interventions (Huh et al., 2015), it is imperative that we further enhance interventions aimed to reduce heavy alcohol use and consequences. One strategy for determining how to best improve interventions is to consider how to leverage the types of behaviors that young adults may implement on their own to reduce negative consequences, namely protective behavioral strategies (PBS).

Protective Behavioral Strategies for Alcohol

PBS are cognitive behavioral strategies that may be used to reduce risk of intoxication and/or consequences (Bravo et al., 2018; Pearson, 2013). Three types of PBS that have often been examined in the literature are strategies for limiting or stopping drinking (e.g., drink water while drinking alcohol), strategies that alter the manner of drinking (e.g., avoid drinking games, drink slowly rather than gulp or chug), and strategies for serious harm reduction (e.g., use a designated driver) (Martens et al., 2005). An extensive literature exists on the use of PBS among young adults, especially college students; however, less is known about the extent to which college students may have intentions to use PBS or make plans to use PBS. The goal of the current study was to elucidate the extent to which college students’ plans to use PBS on a given day were associated with actual use of PBS that day, after accounting for amount of alcohol consumed, thus suggesting which types of PBS college students are likely to follow through with on a given occasion. Findings can inform brief alcohol interventions by identifying whether students tend to follow through on their PBS plans or whether they may be better at following through on certain types of PBS, such as serious harm reduction.

Associations Between PBS Use and Alcohol Use/Consequences

An abundance of cross-sectional and longitudinal research has demonstrated the global-level associations between PBS and drinking behavior, such that greater use of PBS is negatively associated with quantity and frequency of drinking as well as negative consequences (Grazioli et al., 2015; Pearson, 2013; Prince et al., 2013). Emerging event-level research is beginning to demonstrate that use of PBS varies across days as does the type of PBS used (Lewis et al., 2012, 2015; Pearson et al., 2013; Sell et al., 2018). Moreover, these studies show that across days different types of PBS have differing associations with alcohol use and consequences. Similar to cross-sectional research, greater use of manner of drinking strategies have been associated with less alcohol use when examined at the event-level (occasion-level) (Lewis et al., 2012, 2015; Pearson et al., 2013). In contrast, greater use of limiting/stopping and serious harm reduction strategies have been associated with greater alcohol use and consequences at the event-level. These event-level findings are in contrast to cross-sectional research that focus on between-person differences, which show that people who, in general, use limiting/stopping PBS and serious harm reduction PBS report less alcohol use overall. Other work has compared days with high-intensity drinking (8+/10+ drinks for women/men) versus heavy episodic drinking and has found that college students used fewer limiting/stopping PBS on high-intensity drinking days compared to heavy episodic drinking days. Students were as likely to use serious harm reduction PBS on heavy episodic and high-intensity drinking days (Linden-Carmichael et al., 2019). Further, when examining PBS use and high-intensity drinking in relation to the experience of negative consequences, research has shown that college students’ use of manner of drinking strategies and serious harm reduction strategies on given day have been associated with a weaker association between high-intensity drinking and specific negative alcohol-related consequences (e.g., passing out, having no one sober enough to drive), relative to days with heavy episodic drinking (Linden-Carmichael et al., 2018), thus suggesting that when such strategies were used the risk of consequences was reduced. Taken together, the associations among PBS use, alcohol use, and related consequences are complex, especially when considering how behaviors relate to one another at the daily level.

Intentions-to-Behavior Link

To further shed light in the area of PBS research, it can be beneficial to examine the extent to which intentions or plans to engage in a behavior predict the behavior itself. A key component of the theory of planned behavior is that behavioral intentions directly predict behavior (Ajzen, 1991). Although the theory of planned behavior describes constructs leading to intentions (i.e., attitudes, subjective norms, and perceived behavioral control), a key hypothesis focuses on the intentions-to-behavior link. The theory of planned behavior has been effectively applied to college student alcohol use (Collins et al., 2011; Fekadu & Kraft, 2001; Huchting et al., 2008; Norman, 2011) and to heavy episodic drinking (Glassman et al., 2010; Johnston & White, 2003). However, behavioral intention does not fully predict drinking behavior, having an average correlation of .54 across 40 studies in a meta-analysis (Cooke et al., 2016). Drinking intention may be even less predictive of behavior among extreme heavy drinkers (Collins & Carey, 2007) and during specific events such as college football games (Glassman et al., 2010). Thus, there is variability in the degree to which drinking intentions predict behavior. One area of research that needs greater attention is the extent to which college students plan or intend to use each type of PBS and then end up using PBS as planned, which has direct implications for informing brief interventions as well as in-the-moment interventions.

Advantages of Daily Surveys

In accordance with a key component of the theory of planned behavior, the current study examines intention-behavior links at the occasion level by focusing on college students’ plans for PBS (reported in the afternoon) and actual PBS use later that night (reported the next morning). The current study used daily surveys to collect students’ self-reports on alcohol use, consequences, plans for PBS, and use of PBS across eight consecutive weekends. Daily survey methodology offers several advantages over traditional data collection methods (Shiffman, 2009). First, daily surveys can be used to ask about current or recent (e.g., previous day) behaviors and psychological states, thus minimizing recall bias. Second, daily surveys can account for temporal ordering whereby plans are measured prior to engagement in the behavior. Third, daily surveys enable researchers to assess the strength of the daily-level association between constructs. For example, as done in the current study, we can ascertain whether plans for PBS use (as reported in the afternoon) are linked to PBS use later that night (as reported the next morning in order to capture drinking-related experiences from across the previous night). Fourth, daily survey methods allow us to separate between-person effects from within-person effects, and we can isolate (1) whether college students with higher PBS plans on average across days tend to report greater PBS use (between-person effects) and (2) whether days students report higher PBS plans than usual are linked to greater PBS use on that day (focusing on how behavior may change from day to day, within-person effects).

The Current Study

The current study uses daily-level data to investigate the extent to which afternoon plans for PBS were associated with use of those PBS strategies later that night with separate tests for each of three types of PBS: limiting/stopping, manner of drinking, and serious harm reduction strategies (Aim 1). We hypothesized that, for each type of PBS, on days college students have elevated plans for PBS (more plans than they typically have on average), they will report greater use of PBS later that night, controlling for drinking intentions and number of drinks consumed. We also examined whether drinking intentions moderated the link between plans for PBS and PBS use (Aim 2). We hypothesized that elevated drinking intentions on a given day would strengthen the link between PBS plans and PBS use, because students would be more inclined to follow through on their plans for using manner of drinking strategies and serious harm reduction strategies on days they intend to drink more alcohol. In contrast, students’ inclination to follow through on their PBS plans for limiting/stopping strategies may not vary systematically with respect to drinking intentions; students may plan to use fewer limiting and stopping strategies if they expect that they won’t be drinking heavily as well as when they plan to drink more than they usually consume. This conceptualization of the moderating effect of drinking intentions may help shed light on previous event-level research demonstrating that the use of limiting/stopping and serious harm reduction strategies were linked to greater alcohol use and consequences (Lewis et al., 2012, 2015; Pearson et al., 2013).

Method

Participants and Stratification

College students (N = 230) were recruited for a study examining daily-level associations among PBS and alcohol-related behaviors. Eligibility criteria were: (1) be 18 to 24 years old, (2) consume any kind of drink containing alcohol at least two days a week in the past month, (3) consume 4+/5+ drinks (females/males) in one sitting at least once in the past two weeks, (4) provide a cell phone number and agree to receive study text messages, (5) not be studying abroad currently or planning to study abroad in the next 2–3 months, (6) have regular access to the internet, (7) be willing to come to the lab for an in-person session, (8) have a schedule that allows for participating in a daily study, (9) not be participating in another research study, (10) provide first and last name, (11) provide an email address, (12) provide birth sex, and (13) report not being pregnant or trying to get pregnant (females only). All study procedures were approved by the University’s Institutional Review Board, and no adverse events were reported. A federal Certificate of Confidentiality was obtained to help ensure privacy of research participants.

Following baseline completion, participants were randomized by stratified random assignment, which was based on birth sex and number of drinks per week (i.e., “0–10 drinks per week” versus “more than 10 drinks per week”). Participants were randomized into one of two conditions: (1) the main sample (n = 189; morning and afternoon surveys), and (2) the reactivity sample (n = 41; only morning surveys). Preliminary reactivity analyses showed that daily reports of PBS use, alcohol use, and consequences did not significantly differ between participants in the main sample and participants in the reactivity sample.

This study used the 189 participants in the main sample, who completed morning and afternoon surveys, because of our focus on PBS plans, which were reported in the afternoon. The demographic characteristics of the participants randomized to the main sample are as follows. About half (48.68%) of the sample reported their biological sex as female, and the average age at baseline was 20.16 years (SD = 1.54, range 18–24). The majority (67.20%) of participants identified as White, followed by 18.52% who identified as Asian, 8.99% who identified as multiracial, and the remaining 5.29% who identified as another race. About 9.84% of the sample identified as Hispanic or Latinx. Almost half (47.62%) of the sample were fraternity/sorority members.

Procedures

A random sample was recruited through the university’s registrar’s list that included enrolled undergraduate students at a large public university in the northwestern United States. Recruitment and study participation occurred in three academic quarters (Spring 2017, Autumn 2017, Winter 2018) with daily data collection being completed in April 2018. Students were emailed invitations to complete a brief online screening survey to determine if they were eligible for the study. Students were first presented with an online information statement that contained all elements of informed consent, including a description of all study components and their rights as a participant. Thirty-three individuals declined to participate in the study. Individuals who consented were presented with the online screening survey to determine eligibility with 2,007 (14.7% of those invited) students completing the survey (for discussion on low bias from low response rates see Fosnacht et al., 2017 and Trespalacios & Perkins, 2016). No compensation was provided for completing the screening survey. During each academic quarter that recruitment occurred, individuals who completed the screening survey (regardless of study eligibility) were entered into a drawing to win one of two $50 Amazon gift cards.

Of the 2,007 who completed the online screening survey, 342 (17.04%) met eligibility criteria and were invited to the online baseline survey, which was completed by 301 participants (88.01% of those eligible). The most common reasons that participants were ineligible for the study were: (1) reporting drinking at levels below the alcohol criteria (52%), (2) currently studying abroad or planning to study abroad in the next 2–3 months (10%), (3) answering “no” to allowing text messages to be sent to their cell phone (4.8%), and (4) not providing a cell phone number (4.8%).

Participants who completed the online baseline survey were automatically invited to schedule appointments online for the 30-minute in-person training session to review the daily survey protocol. Participants could also call the study office to find a convenient appointment timeslot. In-person sessions for the 30-minute training were completed by 230 participants (76.41% of those who completed baseline and were therefore randomized to the main sample or reactivity sample). The 230 participants who were fully enrolled in the study (i.e., completed the in-person training) did not significantly differ on biological sex, age, typical drinks per week, problematic drinking, or alcohol consequences compared to the 71 individuals who only completed the baseline survey. During the in-person training session, participants met with a member of the research team who provided information about the schedule for receiving the daily online surveys, review of survey questions (e.g., standard drink definition, protective behavior strategies), how to access and complete the daily surveys, and compensation for completing the daily surveys. All participants received a study pamphlet with the project information and also a calendar with the survey schedule that they could use as a reference.

The daily portion of the study used ecological momentary assessment via online surveys that occurred across eight consecutive weekends (Thursday to Sunday) and four random weekdays. Daily data collection focused on weekends, including Thursday night, given that college students tend to use alcohol on weekends and the study aims focused on drinking days (Hoeppner et al., 2012; Tremblay et al., 2010). Participants in the main sample began the daily survey protocol on the first Thursday following completion of the in-person training session. Afternoon surveys always occurred on Thursdays, Fridays, and Saturdays. Morning surveys always occurred on Fridays, Saturdays, and Sundays. A random weekday occurred in four of the eight weeks, and random surveys could occur anytime between Sunday afternoon and Thursday morning. The random days always had an afternoon survey, which was followed the next day by a morning survey. Morning surveys were administered up to four times per week as were the afternoon surveys, depending on which weeks the random day occurred. Across the entire study, each participant received a total of 56 survey invitations (28 morning, 28 afternoon). A series of reminder and check-in calls (e.g., up to 3 phone calls for each check-in or reminder) occurred throughout the study (e.g., midway through the 8-week daily survey protocol).

Participants always had three hours to complete a survey (9am-12pm in the morning; 1pm-4pm in the afternoon). Participants received both a text and an email invitation with a link to the survey at the beginning of each survey window (i.e., 9am and 1pm). Surveys could be accessed on any device with internet access (e.g., cell phone, tablet, computer). If participants had not yet completed their survey 30 minutes prior to the close of the survey window, they received both a text and email reminder (i.e., 11:30am and 3:30pm). If a participant missed a morning survey, items were administered in the afternoon survey later that day to obtain responses on key alcohol-related questions (e.g., alcohol use on the previous day). We tested for differences between participants who completed the afternoon survey instead of the morning survey at least once during the study (n = 118) versus zero times (n = 71). No significant differences were found for biological sex, fraternity/sorority status, baseline drinks per week, and baseline AUDIT scores. Differences were found for age and baseline alcohol consequences. Participants who completed at least one afternoon survey instead of a morning survey were younger (Mean = 19.91 years, SD = 1.43) than those who did not do this during the study (Mean = 20.59 years, SD = 1.64). Participants who completed at least one afternoon survey instead of a morning survey reported more alcohol consequences at baseline (Mean = 8.97, SD = 4.13) than those who did not do this during the study (Mean = 7.23, SD = 4.08). Of note, 37.6% of the sample never completed an afternoon survey instead of a morning survey, 46.6% did so 1–3 times, and 15.8% did so 4 or more times during the eight weekends of daily data collection.

Participants received all compensation in the form of Amazon.com gift cards sent via email. They were emailed a $25 gift card for completing the online baseline survey. Participants earned $2 for each morning survey and $2 for each afternoon survey they completed, and they were paid twice during the daily survey protocol (after 4 and 8 weeks). If they completed 80% or more of their surveys (morning and afternoon), they received a $15 bonus for the first 4 weeks and a $20 bonus for the last 4 weeks. Participants could receive up to $172 in compensation for baseline and daily survey completion. Each week that students completed all of the daily surveys, they were entered into a drawing to win a $25 gift card. A winner was selected each week of daily data collection.

Participants randomized to the main sample completed 83.72% (including partial surveys) of the 10,584 possible surveys with 81.90% of the morning surveys and 85.54% of the afternoon surveys completed. Four of the 189 participants opted out after starting the daily surveys; data were retained for analysis. Almost three-quarters (73.0%, 138/189) of the participants completed 80% or more of the possible surveys, and almost half (48.1%, 91/189) of the participants completed 90% or more of the possible surveys. Morning surveys took an average of 7.89 minutes to complete (SD = 15.90; median = 4.10 minutes); afternoon surveys took an average of 9.39 minutes to complete (SD = 21.59; median = 4.00 minutes). On days participants reported not using alcohol, they received additional items (e.g., reasons for not drinking, time allocation) to offset differences in the length of the morning survey on drinking and non-drinking days.

Measures

Drinking Behaviors Yesterday

In the morning survey, participants were asked if they drank any alcohol yesterday coded as 0 (No, I did not drink alcohol) or 1 (Yes, I drank alcohol). If they answered yes, participants were asked how many drinks they had in total from 1 (1 drink) to 25 (25 or more drinks). Participants who indicated that they did not drink alcohol were recoded as having zero (0) drinks.

PBS Use Yesterday

In the morning survey, participants who reported drinking yesterday were asked whether or not they used different types of PBS when using alcohol the previous day coded as 0 (no) and 1 (yes). From a total of 20 items, seven items (e.g., set a maximum number of drinks) were summed for the limiting or stopping subscale. Five items (e.g., drink slowly rather than gulp or chug) were summed for the manner of drinking subscale. Eight items (e.g., use a designated driver) were summed for the serious harm reduction subscale. The 20 items were based on Treloar et al. (2015), with certain items rephrased to remove terms like “avoid” and then responses were reverse scored. For example, items asking whether participants played drinking games or mixed different types of alcohol were reverse scored, so that higher subscale scores reflected protective behaviors.

PBS Plans

In the afternoon survey, participants indicated, if they were to drink tonight, how likely they would engage in each of the 20 PBS from 0 (Not at all likely) to 4 (Very likely). Parallel to the assessment of PBS use described above, three separate subscale scores were created that averaged the items in the subscale (rather than summed the items as with PBS use).

Drinking Intentions

In the afternoon survey, nine questions asked participants about their planned activities tonight based on how they currently feel. Participants were asked if they planned to drink alcohol tonight, coded as 0 (no) and 1 (yes), and how many alcoholic drinks in total they planned to drink from 0 (0 drinks) to 25 (25 or more drinks). Participants who indicated they did not plan to drink were recoded as having zero (0) planned alcoholic drinks.

Baseline Demographic Information

Demographic information was collected at baseline and used as covariates: age, biological sex coded 0 (female) and 1 (male), and fraternity/sorority membership coded 0 (No) and 1 (Yes).

Data Analysis Plan

Preliminary Analyses

Of the 189 participants, three participants did not report any drinking days and one did not provided data on PBS plans, leaving 186 for analyses. Descriptive analyses for each of the outcomes, predictors, and covariates were assessed. Unconditional means models (i.e., models without any predictors) were conducted for drinking intentions, the three PBS plans subscales, and the three PBS use subscales to assess the variability accounted for by between-person and within-person differences by examining the interclass correlation coefficient (ICC). All analyses were conducted in SAS/IML 14.1 Software (SAS Institute Inc., 2015).

Multilevel Analyses

Examination of the distribution of the outcomes showed that the data were negatively skewed, indicating that regression models typically utilized for count variables, such as Poisson or Negative Binomial, were not a good fit. Thus, both generalized estimating equation (GEE) and multilevel linear regression were assessed as possible analytical strategies. Estimates from GEE and from linear regression without random slopes were almost identical. Therefore, to allow for possible random slopes of plans for PBS, linear regression models were conducted utilizing SAS PROC MIXED and are reported here. Separate models were conducted for each of the three outcomes (i.e., reported use of PBS by subscale). Due to the number of analyses, we chose to use a conservative p-value of < .01 to indicate significance. To test Aim 1, each model included Sex coded 0 (female) and 1 (male), age, fraternity/sorority status coded 0 (no) and 1 (yes), and person-means of number of drinks intend to drink, number of drinks consumed, and plans to use PBS as Level 2 (between-person) variables; Level 1 (within-person) variables included week of the study, weekend status where weekdays were coded 0 (Sunday-Wednesday) and weekends were coded 1 (Thursday-Saturday), and daily-level reports of number of drinks intend to drink, number of drinks consumed, and plans to use PBS. Plans to use PBS were specific to the PBS subscale being tested as the outcome. For example, in the model assessing limiting/stopping strategies, plans to use limiting/stopping strategies were utilized as the predictor. Continuous Level 2 predictors were grand-mean centered, and continuous Level 1 predictors were person-mean centered to improve interpretability. To test Aim 2, we included the interaction between daily-level intentions to drink and daily-level plans to use PBS.

Results

Descriptive Analyses

Means and standard deviations for the outcomes, number of PBS used for each subscale, are provided in the top of Table 1. On average participants reported using at least one of the PBS from each subscale during drinking days. Participants reported using serious harm reduction strategies the most, averaging more than 5 for each drinking day. Descriptive information for the predictors and covariates is shown in the bottom of Table 1. As expected given that the study design focused on weekend drinking, more than 90% of the survey days were reported for the weekend (i.e., Thursday, Friday, or Saturday). On average, participants reported plans to use strategies from each of the PBS subscales with plans for serious harm reduction strategies being the highest. On average, participants reported intending to drink between 3 and 4 drinks, but they drank between 5 and 6 drinks on each drinking day. Unconditional means models indicated significant variation in use of PBS at both the between-person (Level 2) and within-person (Level 1) levels of analysis for drinking intentions (ICC = 0.26), the three PBS plans subscales (ICCs: plans to use limiting/stopping = 0.64; manner of drinking = 0.51; serious harm reduction = 0.73) and the three PBS use subscales (ICCs: use of limiting/stopping = 0.45; manner of drinking = 0.24; serious harm reduction = 0.44).

Table 1.

Descriptive Statistics on Key Constructs

Variables	N	Mean (SD) or %	Range
Outcomes
Level 2 (Person-level)
Use of Limiting/stopping	185	1.32(1.10)	0.00–5.33
Use of Manner of drinking	185	3.35(0.77)	0.67–5.00
Use of Serious harm reduction	185	5.59(1.24)	1.90–7.80
Level 1 (Daily-level)
Use of Limiting/stopping	1914	1.24(1.44)	0–7
Use of Manner of drinking	1919	3.40(1.31)	0–5
Use of Serious harm reduction	1910	5.60(1.71)	1–8
Predictors
Level 2 (Person-level)
Male sex	186	97(52.15%)	0–1
Age at baseline	186	20.19(1.54)	18–24
Fraternity/sorority	186	89(47.85%)	0–1
Drinking Intentions	186	3.60(2.08)	0–10.67
Number of Drinks Consumed	186	5.40(2.29)	1.20–12.25
Plans for Limiting/stopping	186	1.70(0.862)	0.06–3.81
Plans for Manner of drinking	186	1.78(0.848)	0.00–3.80
Plans for Serious harm reduction	186	3.01(0.738)	0.03–4.00
Level 1 (Daily-level)
Week in study	1952	2.01(1.59)	0–5
Weekend (Thurs/Fri/Sat or not)	1953	1801(92.26%)	0–1
Drinking Intentions	1757	3.73(3.37)	0–25
Number of Drinks Consumed	1944	5.49(3.52)	1–20
Plans for Limiting/stopping	1738	1.63(1.04)	0.00–4.00
Plans for Manner of drinking	1737	1.82(1.11)	0.00–4.00
Plans for Serious harm reduction	1732	3.02(0.85)	0.00–4.00

Note. N refers to people at Level 2 and to drinking days at Level 1. Descriptive statistics were calculated using variables in their original metric (i.e., prior to centering). The analytic sample consists of drinking days only.

Multilevel Analyses Testing Aim 1 (Link between PBS Plans – PBS Use) and Aim 2 (Moderation by Drinking Intentions)

Limiting/Stopping Strategies

Table 2 shows Aim 1 results for examining plans of limiting/stopping strategies to predict the use of limiting/stopping strategies later that night. Level 1 (within-person) findings showed that, as expected, afternoon plans for limiting/stopping strategies were significantly and positively associated with use of limiting/stopping strategies later that night. On days when plans for limiting/stopping strategies were elevated (i.e., higher than a participant’s own average level of plans), college students reported increased use of limiting/stopping strategies.

Table 2.

Results of Multilevel Analyses Predicting Use of Limiting/Stopping Strategies

Predictor	Estimate	Standard error	df	t-value
Level 2 (Between-person)
Sex	0.21	0.12	178	1.73
Age	-0.06	0.04	178	-1.52
Fraternity/Sorority	0.07	0.12	178	0.59
Drinking Intentions (person-mean)	0.03	0.05	178	0.56
Number of Drinks Consumed (person-mean)	-0.05	0.04	178	-1.06
Plans for Limiting/Stopping (person-mean)	0.85	0.07	178	12.37**
Level 1 (Within-person)
Week in Study	-0.02	0.02	1505	-1.23
Weekend (Thurs/Fri/Sat or Not)	0.03	0.10	1505	0.32
Drinking Intentions	0.04	0.01	1505	3.21*
Number of Drinks Consumed	-0.02	0.01	1505	-1.40
Plans for Limiting/Stopping	0.36	0.06	1505	5.78**
Plans for Limiting/Stopping * Drinking Intentions	0.05	0.02	1505	3.31*

Note. Sex (0 = Female, 1 = Male), Fraternity/sorority status (0 = No, 1 = Yes).

^*p < .01.

^**p < .001.

Drinking intentions and the interaction between plans for limiting/stopping strategies and drinking intentions were also significant at Level 1. On days when drinking intentions were elevated (i.e., higher than a participant’s own average), college students reported increased use of limiting/stopping strategies. Figure 1 shows the interaction effect (Aim 2) between drinking intentions and plans for limiting/stopping on use of limiting/stopping strategies. There was a larger positive association between daily plans to use limiting/stopping strategies and use of these strategies on days when drinking intentions were elevated compared to days with lower drinking intentions.

Figure 1.

Interaction Between Daily Plans for Limiting and Stopping PBS and Drinking Intentions on the Number of Limiting/Stopping PBS Used during Drinking Events

Note. PBS = protective behavioral strategies.

The only significant predictor at Level 2 was plans for limiting/stopping strategies, showing that participants who reported more plans for limiting/stopping strategies across the sampled days also reported using more of these strategies on a given drinking day than participants who had fewer plans to use them.

Manner of Drinking Strategies

Results for manner of drinking strategies are shown in Table 3. Level 1 findings showed that, as expected, afternoon plans for manner of drinking strategies were significantly and positively associated with use of manner of drinking strategies later that night. On days when plans for manner of drinking strategies were elevated (i.e., higher than a participant’s own average level of plans), college students reported increased use of manner of drinking strategies. Additional Level 1 findings showed that on days when the number of drinks consumed was elevated (i.e., higher than participants’ own average), fewer manner of drinking strategies were used. Level 1 drinking intentions was not significantly associated with manner of drinking strategies nor did it moderate the association between plans to use manner of drinking strategies and their use.

Table 3.

Results of Multilevel Analyses Predicting Use of Manner of Drinking Strategies

Predictor	Estimate	Standard error	df	t-value
Level 2 (Between-person)
Sex	0.22	0.08	178	2.92*
Age	0.06	0.03	178	2.56
Fraternity/Sorority	-0.19	0.08	178	-2.45
Drinking Intentions (person-mean)	-0.10	0.03	178	-3.36*
Number of Drinks Consumed (person-mean)	-0.06	0.03	178	-2.16
Plans for Manner of Drinking (person-mean)	0.30	0.05	178	6.10**
Level 1 (Within-person)
Week in Study	0.01	0.01	1508	0.55
Weekend (Thurs/Fri/Sat or Not)	-0.06	0.03	1508	-0.73
Drinking Intentions	-0.01	0.01	1508	1.36
Number of Drinks Consumed	-0.19	0.01	1508	-19.46**
Plans for Manner of Drinking	0.30	0.04	1508	7.97**
Plans for Manner of Drinking * Drinking Intentions	0.02	0.01	1508	2.17

Note. Sex (0 = Female, 1= Male), Fraternity/sorority status (0 = No, 1 = Yes).

^*p < .01.

^**p < .001.

Level 2 findings showed that participants who had more plans for manner of drinking strategies across the sampled days reported using more manner of drinking strategies. In contrast, participants with higher drinking intentions across the sampled days reported using fewer manner of drinking strategies. Finally, being male was associated with using more manner of drinking strategies.

Serious Harm Reduction Strategies

Aim 1 results for serious harm reduction strategies are displayed in Table 4. Level 1 findings showed that, as expected, afternoon plans for serious harm reduction strategies were significantly and positively associated with use of serious harm reduction strategies later that night. On days when plans for serious harm reduction strategies were elevated (i.e., higher than a participant’s own average level of plans), college students reported increased use of serious harm reduction strategies. Further, drinking intentions were significantly associated with use of serious harm reduction strategies, such that days with elevated drinking intentions were associated with increased use of serious harm reduction strategies. Additional Level 1 findings showed that on days when the number of drinks consumed was elevated (i.e., higher than participants’ own average), more serious harm reduction strategies were used. The hypothesized interaction between plans for serious harm reduction strategies and drinking intentions was not significant.

Table 4.

Results of Multilevel Analyses Predicting Use of Serious Harm Reduction Strategies

Predictor	Estimate	Standard error	df	t-value
Level 2 (Between-person)
Sex	-0.34	0.11	178	-2.98*
Age	-0.02	0.04	178	-0.45
Fraternity/Sorority	0.18	0.12	178	1.54
Drinking Intentions (person-mean)	0.09	0.04	178	2.11
Number of Drinks Consumed (person-mean)	-0.01	0.04	178	-0.26
Plans for Serious Harm Reduction (person-mean)	1.19	0.08	178	15.56**
Level 1 (Within-person)
Week in Study	-0.02	0.02	1498	-1.09
Weekend (Thurs/Fri/Sat or Not)	0.11	0.11	1498	1.04
Drinking Intentions	0.06	0.01	1498	5.03**
Number of Drinks Consumed	0.09	0.01	1498	7.39**
Plans for Serious Harm Reduction	0.58	0.10	1498	5.65**
Plans for Serious Harm Reduction* Drinking Intentions	0.04	0.03	1498	1.43

Note. Sex (0 = Female, 1 = Male), Fraternity/sorority status (0 = No, 1 = Yes).

^**p < .01.

^***p < .001.

Level 2 findings showed participants who had more plans for serious harm reduction strategies across the sampled days reported using more serious harm reduction strategies. Finally, compared to females, males reported using fewer serious harm reduction strategies on average.

Discussion

The theory of planned behavior underscores the role of behavioral intentions in predicting actual behavior (Ajzen, 1991). Drawing from this central tenet of the theory of planned behavior, the current study examined alcohol-related PBS plans and PBS use as reported in morning and afternoon surveys from a sample of college students who drank at least twice a week. This study tested the extent to which plans for using different types of PBS (limiting/stopping, manner of drinking, and serious harm reduction strategies) were associated with actual PBS use. A key finding is that, for each type of PBS, days with elevated plans for PBS were associated with using more of those strategies. Notably, these associations held even after controlling for the amount of alcohol consumed that day. A benefit of collecting daily-level surveys here is that we found that more than half of the variation in PBS plans was attributed to between-person differences (ICCs ranging from .51 to .73 for the PBS plans subscales). Moreover, more than half of the variation in PBS use was within-person, such that individuals substantially varied in their own use of PBS from day to day (based on ICCs reflecting the proportion of between-person variability ranging from .24 to .44 across the PBS use subscales). As such, generally speaking, college students may tend to set out having good intentions of using PBS, but then are more variable in their likelihood of following through on these intentions by actually implementing the PBS on specific drinking occasions.

Overall, higher drinking intentions were associated with using more limiting/stopping strategies and more serious harm reduction strategies, but not the use of manner of drinking strategies. Individuals may be more likely to use certain PBS, like serious harm reduction, on days they intend to consume more alcohol. In contrast, individuals may use manner of drinking strategies on days they intend to consume fewer drinks as well as days with heavier alcohol use, because these strategies tend to reflect how one drinks (e.g., drink slowly) rather than the amount one drinks. Interestingly, the association between daily drinking intentions and PBS use differed between PBS subtypes. Drinking intentions was shown to significantly moderate the association between plans for limiting/stopping PBS and actual limiting/stopping PBS used, in that the positive association between plans for limiting/stopping strategies and use of these strategies was stronger on days participants had higher drinking intentions. This finding may indicate that college students may be especially good at following through on their plans for limiting/stopping strategies when they intend to drink at higher levels.

The actual number of drinks consumed in a night had differential associations with various types of PBS. Consuming more drinks on a given day was associated with using fewer manner of drinking strategies but more serious harm reduction strategies; number of drinks consumed was not associated with use of limiting/stopping strategies. These findings are consistent with other research that showed a negative daily-level association between manner of drinking PBS and alcohol use and a positive association between use of serious harm reduction PBS and alcohol use and (Lewis et al., 2012, 2015; Pearson et al., 2013). Unlike previous research, we did not observe a significant association between use of limiting/stopping PBS and alcohol use, which indicates that days on which college students drank more than usual were not associated with the number of limiting/stopping strategies used. Limiting/stopping were the least reported PBS used, although the extent to which students planned to use these PBS were similar to that for manner of drinking PBS. Together, increasing use of limiting/stopping strategies (even on nights when students drank more than usual) may be a particular area where improvements could be made to increase intervention efficacy.

Study Strengths and Limitations

The current study provides a novel examination of how plans for PBS use are associated with actual use of PBS among college students. The study spanned eight weekends of daily reports that enabled us to capture fluctuations in behavior (e.g., drinking behavior and PBS use) as well as fluctuations in intentions (drinking intentions and PBS plans). Notably, daily-level reports of PBS allowed for a more detailed assessment of how often these types of strategies were used and which type of PBS was used more (or less) often. No evidence of reactivity was found, and the completion rate for the daily surveys was high (above 80%), even with the surveys spanning eight consecutive weekends with a college student sample. Despite these strengths, the following limitations should be noted. First, the sample was recruited from a single university and included a high-risk sample of college students, thus results may not generalize across the college student population or across non-college attending young adults in the community or lighter and infrequent drinkers. Further, we cannot determine how PBS plans may be associated with PBS use among less regular or lighter drinkers. Second, PBS plans and drinking intentions as well as PBS use were based on self-reports. Third, daily data collection focused on weekend nights and PBS plans and PBS use may differ on weekdays compared to weekends (e.g., less PBS use due to lower alcohol use). Finally, the majority (67.20%) of participants identified as White followed by 18.52% who identified as Asian, and we were not able to examine race or ethnicity as a moderator of the effects. Future research is needed to determine how findings may vary across diverse populations.

Clinical Implications and Future Research

Findings from the present study have implications for brief interventions for young adults. PBS tips are often provided as a part of college student alcohol interventions (Larimer et al., 2007; Reid & Carey, 2015). However, studies testing PBS use as a mediator have been inconclusive about whether or not PBS use is an important mechanism of intervention effects (Reid & Carey, 2015). Current findings can inform our knowledge of the types of PBS for which students may need more focused intervention messages to increase students’ motivations for planning to use PBS and also to provide guidance on how to best implement the strategy and follow through on the implementation plan. The current study found that daily plans to use PBS were positively associated with the number of PBS used in a night, controlling for both intentions to drink and actual number of drinks consumed. This finding suggests that interventions may be able to increase students’ PBS use by increasing their plans for PBS, even if students do not change their drinking intentions or the amount of alcohol they consume. Future research could examine whether the link between PBS plans and PBS use is stronger for certain specific strategies (e.g., use a designated driver) as well as young adults’ motivations for using PBS, or not using PBS, as a means to better understand when individuals may plan, or not plan, to use certain PBS strategies. Further, it may be valuable to distinguish between occasions when young adults have the intention to use PBS (and are considering engaging in PBS use) versus are actively making PBS plans (e.g., talking with friends and deciding when to leave a party or deciding the maximum number of drinks they want to consume). Future work in this area could be used to inform interventions by building motivations to use PBS in an effective manner or working on reducing perceived barriers to PBS use (i.e., reasons individuals are not using PBS). Just-in-time interventions could focus on high-risk occasions, such as when drinking intentions are high and plans for manner of drinking PBS are low, and also on reducing drinking intentions or motivations for becoming intoxicated.

Another area for future research is further understanding how students are using PBS through the drinking occasion, moving beyond the number of different strategies in which they engaged. It could be that engagement in one particular strategy is equally, if not more, useful in reducing alcohol use or consequences than engaging in more PBS overall. Future research should identify which strategies may be particularly useful and for whom or in what types of contexts (Braitman et al., 2017; Dekker et al., 2020). Further, understanding when students engage in any given PBS would be helpful for understanding if the strategies are used in a way as intended for reducing harm. For example, an individual could have reported that they drank water while drinking, but perhaps they only remembered later on in the night, or as they drank more alcohol they stopped drinking water. Future studies should assess how well participants engaged in PBS across an entire night (e.g., quality and consistency) and its association with daily-level alcohol use and consequences. Research elucidating the ways in which young adults use PBS as well as the extent to which the manner of use is effective and/or protective can inform brief interventions.

Conclusions

The current findings showed that college students do in fact plan to use PBS ahead of drinking occasions, and days on which students had elevated plans for PBS were linked to greater PBS use for all three types of PBS examined here (limiting/stopping, manner of drinking, and serious harm reduction strategies). Future work is needed to investigate factors associated with when young adults are more likely to make PBS plans as well as factors that predict when young adults are more (or less) likely to follow through with their PBS plans. This area of research holds promise for informing future intervention research that targets PBS as a key mechanism in the process leading to reduced drinking.

Public Health Significance.

This study underscores the need to focus on how and why young adults may plan (or not plan) to use protective behavioral strategies (e.g., spacing drinks, having a designated driver) in order to reduce their alcohol intoxication and consequences. By enhancing our understanding of young adults’ decisions to implement protective behavioral strategies, we may be able to reduce the level of intoxication and alcohol-related negative consequences experienced among young adults through more informed alcohol interventions.