Identifying Substitute Activities for Alcohol Consumption: A Preliminary Analysis

Abstract

Individuals with alcohol use disorder may excessively value alcohol reinforcement over other types of rewards and may seek out environments supportive of alcohol consumption despite negative consequences. Therefore, examining ways to increase engagement in substance-free activities may be useful in treating alcohol use disorder. Past research has focused on preference and frequency of engagement in alcohol-related versus alcohol-free activities. However, no study to-date has examine the incompatibility of such activities with alcohol consumption, an important step in preventing possible adverse consequences during treatment for alcohol use disorder and for ensuring that activities do not function in a complementary fashion with alcohol consumption. The present study was a preliminary analysis comparing a modified activity reinforcement survey with the inclusion of a suitability question to determine the incompatibility of common survey activities with alcohol consumption. Participants recruited from Amazon’s Mechanical Turk (N=146) were administered an established activity reinforcement survey, questions regarding the incompatibility of the activities with alcohol consumption, and measures of alcohol-related problems. We found that activity surveys may identify activities that are enjoyable without alcohol, but that some of these activities were still compatible with alcohol. For many of the activities examined, participants who rated those activities as suitable with alcohol also reported higher alcohol severity, with the largest effect size differences for physical activity, school or work, and religious activities. The results of this study are an important preliminary analysis for determining how activities may function as substitutes, and may hold implications for harm reduction interventions and public policy.

A behavioral economic approach to alcohol use disorder (AUD) posits a molar perspective on alcohol use, suggesting it is maintained by myriad contextual variables and reinforcing consequences such as the availability of reinforcing alternative activities. Moreover, individuals with AUD tend to excessively value alcohol reinforcement and prefer immediate access to reinforcement despite delayed consequences (Bickel et al., 2014). Additionally, environments in which regular alcohol use occurs may begin to exert stimulus control over alcohol use, thus eliciting acute craving for or motivation to consume alcohol in those contexts (DeGrandpadre & Bickel, 1994). Because individuals with AUD may continue to seek out environments that support alcohol use despite negative consequences, identifying reinforcing alternative environments and activities that do not support alcohol consumption can be an important component of a behavior-based intervention.

Research on clinical interventions for AUD has begun to focus on assessing the reinforcing efficacy of and increasing engagement in substance-free alternative activities (Mckay, 2017; Murphy et al., 2006, etc.). For example, the Adapted Adolescent Reinforcement Survey Schedule – Substance Use Version (ARSS-SUV) assesses the reinforcing properties of activities with and without alcohol or substance use among youth and college-aged students (Murphy et al., 2005). Activities derived from the Pleasant Events Schedule have also been adapted into a similar survey targeting the general population of adults (Correia et al., 1998; Correia et al., 2003).

These survey instruments allow individuals to rate their subjective “frequency” and “enjoyment” of each activity both with and without drug or alcohol use. From these ratings, a reinforcement ratio is derived, intended to identify the global amount of reinforcement derived from activities in which alcohol consumption or drug use occurs and may be used to predict the amount of time that will be allocated to activities that involve drug or alcohol use (Murphy et al., 2006). Indeed, reliance on substance-related activities may indicate increased severity of substance-related problems (Correia et al., 2003; Murphy et al., 2006). Behavior occurs in the context of multiple response options; therefore, increasing engagement in substance-free activities can result in decreases in substance-related activity (Higgins et al., 2003).

Substance-free activity surveys hold clinical utility in treatment of AUD for several reasons. First, they help inform treatment by predicting potential problems based on relative alcohol-related reinforcement. They are also straightforward and easy to administer. Other questionnaires and standardized methods of identifying alcohol use, such as the Timeline Follow-Back method (Sobell & Sobell, 1992), are useful in identifying activities an individual may engage in without alcohol consumption (e.g. Murphy et al., 2006) but are more intensive and time-consuming. The reinforcement ratio derived from instruments like the ARSS is often used conjunction with motivational interviewing or substance-free activity sessions and as a mediator of treatment outcomes (Murphy et al., 2019).

Isolated frequency and enjoyment rankings for each activity can also be used as metrics of an individual’s molar behavioral patterns; individuals may prefer and/or engage in some activities more than others. It may therefore be useful to identify activities with high rankings of frequency and enjoyment without alcohol as possible alternatives to promote during treatment for AUD. However, such rankings may not always indicate incompatibility with alcohol consumption. Because individuals who engage in heavy alcohol use are likely to seek out contexts that support alcohol consumption (Correia et al., 2003), it is important to identify reinforcing activities in which alcohol use cannot occur at all or those activities in which heavy alcohol use may not be compatible. This may include both situations in which any alcohol consumption is not appropriate or may have a negative impact, and other situations in which heavy drinking specifically is incompatible or would have adverse impacts to some degree. Failing to ensure that such activities are also incompatible with alcohol consumption may have adverse consequences when promoting engagement in those activities as part of a treatment plan for AUD. These activity survey instruments also have yet to be analyzed at a group level to determine whether certain activities have a greater tendency to be incompatible with alcohol consumption, and therefore serve as better candidate activities to promote.

Behavioral economic theory offers a way to conceptualize how incompatible alternative activities decrease alcohol consumption and the boundary conditions under which this behavior change may occur. First, the activity must be enjoyable and available at a lower unit price (though unit price does not solely indicate monetary costs; effort, reinforcing value, and the absence of negative outcomes associated with the activity may all be considered cost/benefit factors that contribute to demand). Second, activities that appear from surveys like the ARSS to be enjoyable without alcohol may nevertheless operate in a complementary fashion with alcohol consumption such that the increase in the given activity is associated with a concomitant increase in alcohol consumption (Hursh et al., 2013). Effective alternative reinforcement, therefore, depends on both the amount of substitution that occurs between commodities and whether engagement in one activity prevents extraction of reinforcement from another activity (i.e., incompatibility; Hursh & Roma, 2013). In behavioral economic terms, when consumption of an initial commodity of interest decreases as a function of an increase in unit price and consumption of a secondary commodity, available at a lower unit price, increases concomitantly, that secondary commodity is deemed a functional substitute to the initial commodity. While not synonymous with substitutability, incompatibility may be an important pre-condition to identify prior to assessing the substitutability of a given activity for alcohol consumption because drinking occurs during alcohol-related activities.

Within the behavioral economic perspective, the presence of and preference for low-cost or low-effort alternative substance-free activities may also help maintain molar levels of overall reinforcement when alcohol-related activities are eliminated from an individual’s repertoire (Murphy et al., 2007). Offering reinforcing activities that are incompatible with drinking may moderate the opportunity cost that treatment seekers may experience when engaging in alcohol-free activities. While incompatibility of activities may not necessarily indicate that those activities are substitutable, and vice versa, it is important to determine the kinds of activities that are most likely to function as incompatible and thus, offer little to no opportunity for alcohol consumption. Analyzing the incompatibility of established activities in common substance-free activity surveys may therefore be useful; however, this has not yet been examined in a systematic way.

To-date, no study has conducted an individual activity analysis of activity surveys and assessed whether and which activities can function as economic substitutes to alcohol consumption. However, the feasibility of employing activity surveys to identify activities that are incompatible with alcohol must be explored as a precursor to behavioral economic decision-making tasks. Therefore, the primary purpose of this study was to conduct a preliminary analysis using crowdsourced sample to identify whether the questions from the ARSS may identify incompatible activities, and if can we use these questionnaires to identify these kinds of activities with the help of additional questions. We also investigated whether incompatibility was related to severity of alcohol-related risk (as indexed by total score on the Alcohol Use Disorder Identification Test [AUDIT]; Saunders et al., 1993), and whether likelihood of finding a way to covertly consume alcohol during incompatible activities was also related to AUDIT scores.

Method

Participants

Participants were crowdsourced from Amazon’s Mechanical Turk (MTurk) where Workers (individuals who complete tasks via Mturk) were administered a Qualtrics survey (Provo, UT) via a human intelligence task (HIT). Participants were eligible for the HIT if they were over the age of 18, had a previous HIT approval rating of 99%, had more than 1000 previous HITs approved, and were located in the United States. Unique Turker (https://uniqueturker.myleott.com/) code was implemented in the HIT to ensure only one response per IP address. Participants were paid $2.00 for successful completion of the HIT, were allotted one hour to complete the HIT, and were required to enter a unique response completion code into MTurk to ensure that they completed the task in its entirety. To increase confidence in legitimate responses from human participants (i.e., to screen for potential computerized bot responses), participants responded to three attention checks (two multiple-choice, one open-ended question) dispersed throughout the survey. Two graduate-level researchers determined the acceptability of responses on the open-ended question by independently scoring the responses and arriving on a resolution together when conflicts arose.

Measures and Survey Administration

Alcohol misuse was assessed using the AUDIT questionnaire (Saunders et al., 1993). Measures of cannabis misuse and tobacco and e-cigarette use were also administered but not analyzed in the present study. Participants were presented with a modified version of the ARSS-SUV (Murphy et al., 2005), which comprised 36 activities (for the full list of activities and the corresponding abbreviated activity labels used for data analysis, see Supplemental Materials Table 1). Participants were first asked to rate, from 0 to 4, their frequency and their enjoyment of each activity without alcohol. The ARSS was then presented again, but this time participants were asked to rate, from 0 to 4, their frequency and their enjoyment of each activity with alcohol. This order of questionnaire administration was kept consistent across participants. In an attempt to reduce demand characteristics and carryover effects between the ARSS survey and the subsequent suitability survey, we administered a demographic questionnaire and self-report questionnaires assessing substance use and related problems in between these two series of surveys.

We calculated reinforcement ratios for each activity by dividing the cross-product of frequency and enjoyment ratings for a given activity with alcohol by the added cross-products both with and without alcohol. In a diversion from traditional reinforcement ratio calculations, in which the cross-products across all activities are aggregated and one single reinforcement ratio is calculated for each participant, we calculated the reinforcement ratios across participants for individual activities. Thus, if a participant marked both their reported frequency of engagement in and enjoyment of an activity as “0” both with and without alcohol, the reinforcement ratio for that activity resulted in an undefined value. The undefined reinforcement ratios were removed prior to analysis, which caused a difference in number of participant data for each activity. The number of undefined reinforcement ratios derived for each activity, and each individual subjects’ reinforcement ratio across activities, are located in the Supplemental Materials.

Next, participants were asked a targeted question about the compatibility or incompatibility with alcohol consumption for each of the 36 activities, taken directly from the previously administered ARSS measure. Specifically, participants were instructed to select whether each activity was “suitable” or “unsuitable” with alcohol. “Suitable” and “unsuitable” were defined for the participants and the definitions were present above each question (see Supplemental Materials for the full instructional set provided to participants, including follow-up questions). Rather than “compatible” or “incompatible”, we used the terms “suitable” and “unsuitable” for the measure due to the likelihood that participants have a history of using the term “compatibility” in other ways (e.g., romantic compatibility). Follow-up questions were displayed based on the choice selected. If participants selected that the activity was “suitable” with alcohol consumption, they were then asked to rate their likelihood of consuming alcohol during the activity on a 5-pt Likert scale that ranged from “very unlikely” to “very likely”. If participants selected that the activity was “unsuitable” with alcohol consumption, they were then asked the likelihood that they would “find a way to” consume alcohol during the activity on a 5-pt Likert scale that ranged from “very unlikely” to “very likely”. Though compatibility of alcohol consumption across activities likely exists on a continuum, we chose to provide participants with a binary choice in this preliminary analysis for the purposes of a simplified data analysis, and to prevent tempered outcomes using a Likert scale (e.g., a selection of 2, 3, or 4 on a Likert scale of 1–5) by providing a forced-choice question.

All participants were paid after completion of the survey, but participant data were excluded if they did not pass all three attention checks, if they selected “never” when asked how often they drink alcohol (defined as an AUDIT total of 0), or if they did not complete the ARSS in its entirety.

Results

Participant characteristics

A total of 201 participants were recruited through MTurk. After removing participants who failed any number of the attention checks (n = 23), who scored a 0 on the AUDIT (n = 24), and who did not fully complete the ARSS (n = 23), the final number of participants whose data were included was 146. These exclusion criteria were not mutually exclusive. Participants completed the task, on average, in 19.4 minutes. Mean participant age was 37.02 (SD = 10.29). Most participants reported obtaining a bachelor’s degree as their highest level of education (57.5%), followed by some college (17.8%). Participants reported their gender identity as male (52.7%), female (46.6%), or transgender (0.7%). Most participants reported being white, non-Hispanic (87.0%). The remaining participants reporting their race as Black/African American (6.2%); Hispanic/Latino (3.5%); Asian (2.1%); Alaska Native, Native American, Métis, First Nations, or Inuit (<1%); or more than one race (<1%).

Suitability and reinforcement ratios

To examine the proportion of participants who indicated that a given activity was suitable with alcohol consumption, we calculated a suitability ratio for each activity. These ratios were calculated by dividing the total number of accounts in which the activity was indicated as “suitable” by the total N (146). Thus, suitability ratios ranged from 0 to 1. The reinforcement ratios and suitability ratios are presented/depicted in Figure 1. Some overlap existed between the activities with the highest reinforcement ratios and the highest suitability ratios (going to a club or bar, going to a party) and the lowest ratios (participating in aerobic exercise), indicating good construct validity; however, some differences were also present. In general, we observed many high suitability ratios across activities. This indicates that many of the activities included in the survey would not serve as good candidate substitutes because they would frequently be perceived as compatible with alcohol consumption even though they may be enjoyable without alcohol. Further comparison between the two ratios would be untenable because of the differences in the method in which they were calculated.

Figure 1.

Reinforcement and Suitability Ratios Across Activities

Because the reinforcement ratios across activities did not include all participants due to the presence of undefined values, we examined the average frequency and enjoyment ratings across each activity with and without alcohol. Figure 2 shows the aggregated responses of frequency (top panel) and enjoyment (bottom panel). In general, participants’ frequency and enjoyment rankings for each activity were higher without alcohol than with alcohol, save for the frequency rankings for going to a club or bar, going to a party, and gambling, and enjoyment rankings for going to a club or bar, and going to a party. However, participants still reported moderate to high levels of frequency and enjoyment in activities with alcohol. When compared to the suitability ratio data shown in Figure 1, these results indicate that despite enjoyment of an activity without alcohol, many of these activities may still be compatible with alcohol consumption and thus may not be useful candidate substitute activities.

Figure 2.

Frequency and Enjoyment Ratings Across Activities

Associations between suitability and alcohol problem severity

To elucidate the relationship between reported alcohol problem severity and suitability scores across activities, we compared AUDIT total scores between participants who rated each activity as suitable and those who rated the activity as not suitable (see Table 1 for complete results). Independent samples t-tests revealed significant differences in AUDIT total score for 18 out of 36 of the activities. Rank ordering the Cohen’s d values for these comparisons indicates that the largest effect size differences were observed for engaging in non-aerobic (d = 2.16) or aerobic (d = 1.96) exercise, attending religious services (d = 1.68), and attending work or school (1.49). Of note, participating in sports or being physically active were also among the largest effect sizes. For all but one activity (going to a party), participants in the suitable group reported higher AUDIT total scores relative to the not suitable group. The opposite pattern observed for going to a party should be interpreted with caution as only 19 participants viewed this activity as not suitable.

Table 1.

AUDIT Total by Suitability Rating

	Suitable			Not Suitable
Activity	N	M	SD	N	M	SD	t	p	d	d Rank
Having a conversation	121	9.79	8.06	25	10.28	8.20	−0.27	0.784	−0.06	29
Cultural event	100	9.79	8.04	46	10.07	8.17	−0.19	0.849	−0.03	27
Dancing	108	9.47	7.94	38	11.03	8.36	−1.02	0.308	−0.19	32
Gardening	51	14.82	8.49	95	7.22	6.43	6.07	<0.001	1.05	7
Shopping	46	14.96	8.69	100	7.54	6.57	5.70	<0.001	1.02	8
Hobby	86	10.40	7.99	60	9.13	8.15	0.93	0.353	0.16	23
Reading	81	10.49	8.66	65	9.11	7.22	1.03	0.303	0.17	22
Sexual activity	98	9.91	8.31	48	9.81	7.59	0.07	0.947	0.01	25
Watching TV	117	9.80	8.34	29	10.17	6.89	−0.22	0.826	−0.05	28
Sporting event	106	9.61	8.35	40	10.58	7.26	−0.64	0.522	−0.12	30
Club or bar	134	9.57	8.10	12	13.33	6.96	−1.56	0.121	−0.47	36
Aerobic exercise	33	19.33	6.32	113	7.12	6.21	9.91	<0.001	1.96	2
Work or school	28	18.25	6.50	118	7.89	7.06	7.08	<0.001	1.49	4
Walking	57	14.19	8.84	89	7.11	6.11	5.72	<0.001	0.97	9
Religious service	35	18.26	6.98	111	7.23	6.41	8.68	<0.001	1.68	3
Napping	64	12.92	8.63	82	7.50	6.72	4.27	<0.001	0.71	12
Write letters	62	12.85	8.70	84	7.68	6.80	4.04	<0.001	0.68	15
Time with friends	126	9.38	8.17	20	13.00	6.66	−1.88	0.062	−0.45	35
Playing games	116	9.61	8.04	30	10.90	8.16	−0.78	0.437	−0.16	31
Listening to music	115	9.85	8.44	31	9.97	6.55	−0.07	0.944	−0.01	26
Going to movies	70	12.13	8.48	76	7.80	7.09	3.35	0.001	0.56	16
Non-aerobic exercise	31	20.16	5.56	115	7.10	6.15	10.69	<0.001	2.16	1
Going to a party	127	9.09	7.80	19	15.11	7.96	−3.12	0.002	−0.77	10
Time in outdoors	90	11.21	8.53	56	7.73	6.76	2.59	0.011	0.44	17
Restaurant	115	9.25	8.18	31	12.19	7.24	−1.82	0.071	−0.37	34
Politically active	40	16.80	8.42	106	7.26	6.17	7.50	<0.001	1.39	5
Being alone	116	10.19	8.17	30	8.67	7.59	0.92	0.358	0.19	21
Participating in sports	43	15.44	8.59	103	7.55	6.58	6.01	<0.001	1.09	6
Participating in civics	71	12.63	9.05	75	7.27	5.96	4.25	<0.001	0.70	13
Singing	63	12.83	8.65	83	7.64	6.81	4.05	<0.001	0.68	14
Cooking	85	11.22	8.62	61	8.00	6.83	2.42	0.017	0.41	18
Bathing	78	10.85	8.53	68	8.76	7.38	1.57	0.120	0.26	20
Gambling	104	10.08	8.11	42	9.38	8.00	0.47	0.638	0.09	24
Bowling	92	10.82	8.88	54	8.28	6.15	1.85	0.066	0.32	19
Playing with pets	63	12.98	8.88	83	7.52	6.49	4.30	<0.001	0.72	11
Relaxing	121	9.45	8.12	25	11.92	7.53	−1.40	0.164	−0.31	33

Note. N = number of participants who rated activity as suitable or not suitable; M = mean AUDIT score; SD = standard deviation.

Follow-up questions to suitability score

We next analyzed responses to the follow-up questions based on whether a participant marked an activity as “suitable” or “unsuitable”. The general distribution of responses across all participants and all activities is shown in Figure 3. If “suitable” was selected, the most common response to the follow-up question was “somewhat likely” (38.3%). If “unsuitable” was selected for that activity, the most common response to the follow-up question was “very unlikely” (52.4%). However, participants reported that, despite an activity being unsuitable with alcohol consumption, they would be neither likely nor unlikely (14.5%), somewhat likely (8.9%), or very likely (2.9%) to consume alcohol during that activity.

Figure 3.

Distribution of Follow-Up Responses to Question of Suitability

We also analyzed the proportion of responses to the follow-up questions from the suitability task across individual activities based on whether the activity was marked as suitable or unsuitable. Heat maps of the response distributions for the follow-up question for suitable and unsuitable activities are located in Supplemental Materials Figures 1 and 2, respectively. At the individual activity level, some activities which were marked as suitable for alcohol consumption had a much higher proportion of participants reporting that they would be “very likely” to drink during the activity (e.g., attending a party). Thus, these activities may hold increased likelihood of alcohol consumption. When activities were marked as unsuitable, the distribution of responses to the follow-up question indicated that some activities may be better candidate substitutes than others. For example, activities such as attending religious services or groups, aerobic exercise, or caring for pets had some of the largest proportion of responses that the participant would be very unlikely to consume alcohol during the activity. However, for some activities, like going to a restaurant or a sporting event, this was not the case.

Based on these response patterns, we hypothesized that individuals who reported a greater likelihood of finding a way to drink during incompatible activities would report higher severity of alcohol-related problems. Figure 4 shows Pearson correlations of reported likelihood of drinking, or of finding a way to drink, for marked suitable and unsuitable activities, respectively. For marked suitable activities, we observed generally positive correlations between likelihood of drinking and AUDIT scores, with the highest positive correlations observed for the following categories: performing a hobby (r = .54), writing letters or emails (r = .48), and taking a walk (r = .45). For reported unsuitable activities we found that, for all activities except one (going to a party), likelihood of finding a way to drink during an activity was positively correlated with AUDIT score severity. The highest positive correlations observed for the following categories: being alone (r = .65), cooking (r = .61), and dancing (r = .61).

Figure 4.

Pearson Correlations of Reported Likelihood of Drinking and AUDIT Score

Discussion

The results of this preliminary analysis show that activity surveys may be useful in identifying activities that participants find rewarding without alcohol; however, these activities were not always incompatible (unsuitable) with alcohol consumption. The suitability addition to traditional activity surveys allowed us to identify differences across activities. Additionally, the follow-up questions showed variance in likelihood of finding a way to consume alcohol during an activity, even among activities denoted as unsuitable for alcohol consumption. In general, exercise, attending religious services or events, being politically and civically active, and gardening were activities with the lowest aggregate suitability ratios. While continued assessment of various activities and modified activities may be useful and allow researchers and clinicians to identify a list of activities that tend to be more unsuitable than suitable, the present data show that individualized assessment of personally relevant activities continue to be important for AUD.

AUDIT severity differed based on incompatibility ratings for many of the activities and was positively correlated with reported likelihood to “find a way“ to consume alcohol during incompatible activities. We acknowledge that the way this question was worded inadvertently implies a compulsive tendency to drink even when it is not permissible or advisable, and the validity of the notion of compulsivity in addictions has been subject of recent debate (e.g., George et al., 2022). Nevertheless, these correlations support previous findings that severity of AUD and reinforcement derived from substance-related activities are related. Indeed, animal-models have demonstrated genetically mediated differences in sensitivity to alternative reinforcement (Augier et al., 2018). Though the present data came from a non-clinical sample, these results may hold important clinical implications. Identifying activities incompatible with alcohol consumption may be more challenging for individuals with greater alcohol problem severity. Individuals with AUD may be more likely to experience reward deprivation when alcohol is removed from their daily activities (Joyner et al., 2016). Indeed, an inverse relationship exists between alcohol consumption and engagement in substance-free activities (Acuff et al., 2019; Murphy et al., 2007). Alcohol consumption and alcohol-related activities are oftentimes social in nature, promoting social relationships and bonding (Kelly et al., 2011). It is of considerable importance that social reinforcement be maintained because nonsocial passive outdoor and indoor activities tend to be the substance-free activities least likely to be retained (Correia et al., 2003), and increasing social engagement in non-drinking activities while decreasing ties to pro-drinking networks is associated with successful treatment outcomes for AUD (Kelly, et al., 2011). Because the prevailing activity surveys may not be useful in identifying a large number of potentially substance-free activities, continued adaptation of these surveys is needed.

From a harm reduction standpoint, we also consider this incompatibility assessment useful in identifying risky activities or activities that may be compatible with alcohol but that an individual may not prefer to remove from their day-to-day life. This assessment can arm both a clinician and their client with the knowledge that certain activities may, from the client’s perspective, continue to be conducive to alcohol use. The clinician can then be prompted to 1: direct the therapeutic intervention to managing craving for alcohol and building self-efficacy in those settings, and 2: provide the client with mechanisms to cope and manage their alcohol use if they choose to consume alcohol in those contexts. Reducing heavy drinking across contexts of an individual’s environment can also serve to improve several other psychological states; that is, reducing the extent to which one’s identity is tied with alcohol use, enhancing mood, etc. Future research should involve administering these survey questions among a clinical sample to identify whether response patterns would be similar in a target population.

Frequency and enjoyment ratings on the ARSS, suitability scores, and response patterns on the follow-up questions may reflect participants’ social circles. For example, social norms may be predictive of college student drinking and may indicate more frequent acceptability of drinking in this population compared to other populations (Neighbors et al., 2007). Social network analyses have been useful in identifying substance use behaviors among social circles (e.g. Knox et al., 2019; Maisel et al., 2015). Future research should include exploring individual social circles, for these may be indicators of one’s molar drinking environment and may predict risk of alcohol misuse or AUD.

The results of this study and continued research in this area may also hold important implications for public policy. The tendency for individuals with AUD to limit their behavioral repertoires as alcohol use severity increases may pose a challenge to treatment when identifying candidate substitutes, as we emphasized above. Solutions may be most efficient at the policy level, in which funding can be allocated to increase ease of access to activities that are likely incompatible with alcohol consumption at low or no cost. For example, the proportion of green space and parks is much lower in traditionally lower-income urban areas, becoming a target issue of environmental justice work (e.g. Byrne et al., 2009; Dai, 2011). Continued research on the importance of incompatible activities in identifying substitutes for alcohol consumption may promote policy changes and help in treatment of AUD and in preventing heavy or risky drinking.

Behavioral economics offers a useful way to conceptualize and quantify the degree of substitutability of activities with alcohol consumption by helping us understand how manipulating contextual factors may shift behavior in one direction or another (Hursh et al., 2013). While providing access to alternative, substance-free reinforcement is a useful treatment component (Murphy et al., 2006) and reinforcement-based interventions have demonstrated efficacy (Fazzino et al., 2019), according to behavioral economic theory, availability of alternatives alone may not be sufficient in promoting meaningful and durable shifts in behavior (Green & Fisher, 2000). Behavioral economic assessments of cross-commodity demand can help researchers and clinicians alike to understand the nuanced relationships between alternative activities and engagement in alcohol consumption, as well as the boundary conditions under which certain activities may be substitutes for alcohol use and conditions under which this would not be the case (e.g. changes in effort, costs, or acute motivation). While not necessarily synonymous with the substitutability/complementarity continuum, the self-reported compatibility of an activity with alcohol consumption is an important component in identifying activities most appropriate to test as substitutes. Experimental assessments of substitutability as a method of identifying substance-free activities remains an uncharted area and warrants research.

There are several limitations worth noting. First, to our knowledge, this was the first study to measure the incompatibility of activities with alcohol consumption by administering a suitability score. Additional research is needed to validate whether this assessment will indicate activities that are both enjoyable and truly incompatible with alcohol use. Furthermore, we do not yet know whether the activities scored as incompatible by some participants would have functioned as a substitute for alcohol-related activities. We also acknowledge limitations to the study sample. Races other than white were not well-represented, nor were gender identities other than male and female. Crowdsourced research has historically been questioned due to the limited diversity of the participant pools. Future research should prioritize diversity in sampling methods. The non-clinical sample also poses a concern for the limited generality of findings. Generally, frequency and enjoyment of activities were ranked higher without alcohol consumption than with alcohol consumption. It is possible that the participants recruited for the current study have a larger and more frequent repertoire of activity engagement without alcohol consumption than we would observe in a sample meeting criteria for an AUD.

Despite the rationale for employing a forced-choice, binary outcome when assessing suitability of each activity with alcohol consumption, we acknowledge that the natural environment is complex and that the variables that may evoke drinking behavior are multifaceted. For example, the presence of next-day responsibilities impacts alcohol consumption (e.g. Gilbert et al., 2014; Skidmore & Murphy, 2011). Alcohol consumption in a given context also involves choices following every beverage finished: whether to continue drinking or to stop drinking. To that end, we did not ask participants to report the number of drinks they would be likely to consume during activities, preventing us from differentiating between consuming a few drinks and engaging in a heavy drinking episode. This is important because many of the activities examined in our study may be viewed as compatible with consuming 1 or 2 drinks, but less compatible when consuming 5 or more drinks. Our choice of a binary outcome was, we felt, necessary for to simplify the present preliminary analysis, but we acknowledge that the choice naturally oversimplifies the complexities of real-world choice situations. Future studies would benefit from assessing compatibility in different situations, using a more nuanced definition, and involving varying levels of alcohol consumption. Finally, it may be important to use a more molar lens targeting overall levels of alcohol consumption when implementing a harm reduction-based clinical treatment program, rather than identifying a smaller number of incompatible activities. In the natural environment, few activities are fully incompatible with alcohol consumption. Future researchers may therefore consider adapting the current suitability score for the purposes of targeting a molar reduction in alcohol consumption.

In sum, the present analysis was a crucial first step in finding ways to identify incompatible activities and was generally successful. We found that activity surveys alone may not be sufficient in identifying adequately incompatible activities, but administering targeted suitability and likelihood-to-drink follow-up questionnaires may be a more useful method in doing so. The positive relationship between AUDIT score severity with suitability and likelihood of drinking during unsuitable activities underscores the importance of exploring novel methods of activity list administration. Finally, the relationships discovered in the present analysis indicate that necessary next steps include devising methodology to directly assess substitutability of incompatible activities with alcohol use and alcohol-related activities and a more nuanced consideration of heavier vs. lighter drinking episodes.

Biographies

Dr. Sarah Weinsztok is a postdoctoral researcher with the Cofrin Logan Center for Addiction Research and Treatment and is a member of the Behavioral Economics and Addictions Neuroscience lab at the University of Kansas. Her research focuses on employing the science of behavior analysis and behavioral economics to understand the factors underlying maladaptive behavior patterns among individuals with substance use disorders. Dr. Weinsztok received a M.S. and Ph.D. in Psychology from the University of Florida. She has obtained grants from institutions such as the Society for the Advancement of Behavior Analysis and the Society for the Experimental Analysis of Behavior to fund her research.

Dr. Derek Reed is a Professor in the Department of Applied Behavioral Science at the University of Kansas, a Scientist in the Cofrin Logan Center for Addiction Research and Treatment, and director of the Applied Behavioral Economics Laboratory. He is a Board Certified Behavior Analyst at the Doctoral Level, and a Licensed Behavior Analyst in the State of Kansas. Derek’s research investigates quantitative models of choice and reinforcer efficacy, as well as the role of reinforcement pathologies of health and addictive behaviors. He specializes in the development of behavioral economic measures of substances of abuse and risky health decisions, with the aim of using these concepts and measures to inform treatment and public policy. Derek received his M.S. and Ph.D. in School Psychology from Syracuse University. Derek has over 150 publications, coauthored three edited books and one textbook, and has won several awards for his scholarship, including the American Psychological Association Division 25 B. F. Skinner Foundation New Applied Researcher Award and the Federation of Associations in Behavioral and Brain Sciences Early Career Award.

Dr. Michael Amlung is the Associate Director for Training of the Cofrin Logan Center for Addiction Research and Treatment, an Associate Professor in the Department of Applied Behavioral Science, and the Director of the Behavioral Economics and Addictions Neuroscience Lab @ KU. His research examines factors that contribute to pathological decision-making in individuals with substance use disorders, the behavioral and brain basis of motivation to use alcohol and other drugs, and the effects of environmental contexts and physiological states on addictive behaviors. Dr. Amlung received a M.S. and Ph.D. in Psychology from the University of Georgia. He completed a NIAAA-funded postdoctoral fellowship in addictions research at the University of Missouri. Dr. Amlung’s research is funded by grants by the National Institute on Alcohol Abuse and Alcoholism (NIAAA), the Canadian Institutes of Health Research, and other institutional grants and awards.

Data availability statement

The data presented in the paper are available by request from the corresponding author.