Whose Approval Matters Most? Examining Discrepancies in Self- and Other- Perceptions of Drinking

Abstract Background Research demonstrates that perceptions of others’ attitudes toward drinking behaviors (injunctive norms) are strong predictors of alcohol consumption and problems. Personalized normative feedback (PNF) aims to reduce the discrepancy between one’s perception of others’ attitudes toward drinking and others’ actual attitudes toward drinking. An implicit assumption of PNF is that self and (perceived) other attitudes toward drinking are aligned (thus, shifting one’s perceptions of others’ attitudes shifts one’s own attitudes). However, there is minimal research on the extent to which alignment (or discrepancy) in self-other attitudes toward drinking is associated with alcohol-related outcomes. Methods College students (N = 1,494; M age = 20.11, 61.0% female, 66.4% White) who endorsed past-month heavy episodic drinking reported injunctive norms toward drinking on weekends, drinking daily, drinking to black out, and drinking and driving. Participants reported their perceptions of attitudes toward these drinking behaviors for three reference groups: close friends, typical university-affiliated peers, and parents. Outcomes included weekly drinking, alcohol problems, and alcohol-related risk. Results Response surface analyses indicated that alignment in approval (versus alignment in disapproval) of drinking demonstrated a linear association with alcohol-related outcomes. Discrepancies in self-peer and self-parent attitudes were associated with alcohol-related outcomes and one’s own attitudes (versus one’s ratings of others’ attitudes) of drinking were more strongly associated with outcomes. Conclusions Results provide evidence of how self-other discrepancies in attitudes toward drinking are associated with alcohol-related outcomes. Future work is needed to test whether self-other discrepancies in attitudes toward drinking impacts response to norms-based interventions.

Perceptions of others' approval of behaviors, such as alcohol use, are among the strongest and most consistent antecedents of young adults' engagement in a given behavior (Graupensperger et al., 2020;Lac & Donaldson, 2018;Neighbors et al., 2008;Turrisi et al., 2007).Furthermore, emerging adulthood is a highly impressionable period in which others' behaviors and attitudes have a strong impact on one's own behaviors and attitudes (Burnett et al., 2011;Graupensperger et al., 2018;Webber et al., 2017).This transition period ranging from late teens into early to late 20s is also associated with increased risk for engagement in substance use (Arnett, 2005;Merrill & Carey, 2016;Patrick et al., 2020) and young adults are at high risk for alcohol-related consequences (Hingson et al., 2017;Merrill & Carey, 2016;Schulenberg et al., 2021).Accordingly, student alcohol misuse remains a salient public health concern on college campuses (Merrill et al., 2023).
The relationship between perceptions of others' behavior and one's own behavior has been described by the focus theory of normative conduct (Cialdini et al., 1991), which distinguishes between two types of social norms that influence individuals' behaviors and attitudes.The first type of social norm, descriptive norms, refers to perceptions of others' behavioral patterns (e.g., perceptions about how much or how often one's friends drink).The second type of social norm, injunctive norms, refers to perceptions of others' approval or disapproval of a given behavior (e.g., beliefs about how approving one's friends are of alcohol use).Whereas descriptive norms provide a blueprint for how to fit in, injunctive norms provide a moral compass for one's behavior (Cialdini & Goldstein, 2004).Both descriptive and injunctive norms have been shown to influence young adults' alcohol use behaviors (Borsari & Carey, 2003;Graupensperger, Jaffe, et al., 2021;Graupensperger, Turrisi, et al., 2021;Neighbors et al., 2007).Furthermore, meta-analytic evidence demonstrates that young adults consistently perceive typical peers to drink more and to be more approving toward drinking as compared to themselves; these perceptions often reflect an overestimation of typical peers' drinking behaviors or approval toward drinking (Borsari & Carey, 2003; see Figure 1).
Both descriptive and injunctive norms can be targeted in alcohol-focused personalized normative feedback (PNF) interventions, which aim to correct normative misperceptions in alcohol-related behaviors (e.g., showing how one's own drinking compares to that of their peers).Evidence demonstrates that correcting descriptive, injunctive, or descriptive and injunctive norms all reduce drinking and that providing feedback around injunctive norms is comparably efficacious as providing feedback around descriptive norms (Larimer et al., 2023).However, there is considerably less research on how injunctive norms (and how discrepancies in injunctive norms) impact alcohol-related behaviors (Miller et al., 2013).This is in part due to differences in the measurement of injunctive (versus descriptive) norms.These measurement differences may also contribute to evidence suggesting that injunctive norms are more directly tied to alcohol attitudes than behavior (Lac & Donaldson, 2018).These factors, in addition to challenges in identifying key reference groups, make it more difficult to establish intervention targets for injunctive norms and to understand the mechanisms underlying the efficacy of PNF interventions.
Regardless, there is increasing focus on and recognition of the importance of injunctive norms that describe one's attitudes toward drinking (e.g., Merrill et al., 2016Merrill et al., , 2018)), particularly within emerging adults who may be particularly susceptible to peer approval (e.g., those with higher symptoms of social anxiety).Indeed, PNF interventions focused solely on correcting misperceptions in injunctive norms not only reduce alcohol use and related consequences and produce changes in perceived descriptive norms but also are associated with increased user interest and satisfaction (Prince et al., 2015;Prince & Carey, 2010).Therefore, there are particular advantages to studying injunctive norms and their impact on alcohol-related outcomes.
There are also advantages to studying injunctive norms within a young adult population.College settings represent a "peer-intensive" setting, where students are surrounded by fellow young adults and alcohol-related social influences are salient (Perkins, 2002).According to seminal social psychology theories, perceptions of others' approval can influence behavior through two primary pathways: (a) anticipating benefits of engaging in socially-approved behavior and (b) prioritizing others' approval as the "correct" way to behave (i.e., social proof; Cialdini, 1984;Deutsch and Gerard, 1955).More contemporary social influence literature highlights the potential importance of the normative referent group (i.e., whose attitudes matter?)Among the various sources of peer influence, studies have shown that perceived norms for more proximal normative referents (e.g., close friends) are more strongly associated with alcohol consumption, relative to norms for more distal references (e.g., "typical" college students; Larimer et al., 2011Larimer et al., , 2009;;Lewis and Neighbors, 2004;Stevens et al., 2021).Furthermore, perceptions of parental approval remain a strong correlate/antecedent to college students' substance use behaviors (Mallett et al., 2019;Turrisi et al., 2001).Students who believe their parents are more permissive of alcohol use are at greater risk for alcohol-related outcomes and consequences (Varvil-Weld et al., 2014).Furthermore, the perceptions of others' attitudes may be quite disparate between referent groups.For example, students' perceptions of peer and parent attitudes are often quite different, and greater disparities in perceived approval between referent groups is associated with increased alcohol use and negative consequences (Cail & LaBrie, 2010).
Research describes an overall discrepancy in injunctive norms, such that young adults are more likely to perceive themselves as being less approving of drinking as compared to a typical peer and to overestimate the extent to which their peers approve of drinking (Borsari & Carey, 2003; see Figure 1).However, beyond work describing this initial discrepancy, it is unclear how young adults' own attitudes toward drinking aligns with, or deviates from, their perceptions of others' attitudes toward drinking, and how this alignment (or discrepancy) relates to behavioral outcomes.For example, when considering the impact of alignment between self and parent attitudes toward drinking, previous research has demonstrated that young adults who perceive that their parents approve of drinking are at greater risk for negative alcohol-related outcomes (e.g., Varvil-Weld et al., 2014); implying that high alignment in self-parent perceptions is associated with alcohol-related outcomes.Alternatively, young adults raised by parents who strongly approved of drinking (and perhaps as a result, who were exposed to harmful outcomes from alcohol use), may develop more disapproving attitudes toward their own alcohol use; thus, lower self-parent alignment may buffer against alcohol-related outcomes.Still too, the impact of alignment in self-parent perceptions may be limited, suggesting that perceptions that one's parents disapprove of drinking behaviors may not impact a young adult's own approval of drinking in a peer-intensive setting.Overall, further research is needed to characterize the presence, level, shape, and direction of alignment in self and (perceived) other perceptions of attitudes toward drinking and how alignment in self-other attitudes is associated with alcohol use, problems, and consequences.

Present study
Previous literature has largely focused on characterizing the magnitude in two types of norm discrepancies: (1) self-other discrepancies in rated attitudes toward drinking and (2) perceived and actual attitudes toward drinking (see Borsari & Carey, 2003; see Figure 1).PNF interventions primarily target the second discrepancy described above and rely on the assumption that one's own attitudes toward drinking are aligned in valence and strength with the reference group.As emerging adults are increasingly likely to report reduced alcohol use (or abstinence) compared to previous generations (McCabe et al., 2021), there is likely diversity in self-and (perceived) other-perceptions of drinking behaviors within the college population.Thus, characterizing the level, shape, and direction of alignment in attitudes toward drinking between self and different reference groups (e.g., close friends, parents)-beyond whether or not a discrepancy exists-advances social norms theory by providing evidence regarding how individuals perceive attitudes toward drinking within their broader social network.Examining the impact of alignment between one's own attitudes toward drinking and one's perceptions of others' attitudes toward drinking on alcohol-related outcomes may inform future adaptations of PNF interventions.
This secondary analysis examined alignment between self-other attitudes toward drinking across three different referent pairs: self and typical peers, self, and close friends, and self and parents.We examined the association between the presence, level, shape, and direction of alignment (or discrepancies) and weekly alcohol consumption (Model 1), alcohol-related problems (Model 2), and alcohol-related risk (Model 3) (see Figure 1).We hypothesized that alignment in perceptions of approval (and particularly strong levels of approval) of drinking between self and the three referent pairs would have a significant and positive association with alcohol-related outcomes.Finally, we hypothesized that one's own attitudes toward drinking would be more closely associated with outcomes as compared with one's perceptions of attitudes for other referent groups.

Participants
This study uses data collected from a screening survey included in a large, randomized control trial testing the efficacy of brief alcohol reduction strategies for college students (Larimer et al., 2023).Participants (N = 1,494) were recruited at one large public university and one mid-sized private university by random selection from the Registrar's list.Potential participants received an email invitation to complete the online screening survey.Participants who completed this screening survey were between 18 and 25 years old (M age = 20.11,SD age = 1.37) and a majority (n = 912, 61.0%) identified as female.Most identified as White (n = 992, 66.4%), 12.1% (n = 181) identified as Asian, 10.4% (n = 156) identified as multi-racial, 2.2% (n = 33) identified as Black or African-American, 1.9% (n = 28) identified as Native Hawaiian or Pacific Islander, 0.33% (n = 5) identified as American Indian/Alaskan Native, and 3.3% (n = 49) identified with a race that was not listed.There were 12.3% (n = 184) of participants who identified as Hispanic/Latino(a).We selected data from participants who reported at least one occasion of heavy episodic drinking in the past month (i.e., 4+/5+ drinks during a single occasion for females/males).

Injunctive norms and personal attitudes
The measure of injunctive norms assessed attitudes toward drinking behaviors, including drinking on the weekends, drinking daily, drinking to black out, and drinking and driving (Baer, 1994).Participants were asked to report their perceived attitudes for three referent groups: close friends, a typical peer at their respective university, and their parents.Additionally, participants rated their own attitudes toward the four drinking behaviors.All subscales were assessed using a 7-point Likert scale ranging from 1 (Strongly disapprove) to 7 (Strongly approve) and midpoint of 4 (Neither approve nor disapprove).This measure of injunctive norms and personal approval has been used extensively in previous research; however, this measure demonstrated relatively low internal reliability in the present sample: close friends (α = 0.68), typical peers (α = 0.74), parents (α = 0.59), and personal approval (α = 0.64).

Weekly number of drinks
The Daily Drinking Questionnaire (Collins et al., 1985) was used to assess the average estimated quantity of alcohol consumed in a given week during the past month.Individuals reported the number of drinks consumed each day on a typical week during the past month.Values were summed to create an average number of drinks per week.This measure has demonstrated strong psychometric properties (Borsari & Carey, 2001;Miller et al., 2002).

Alcohol-related problems
Alcohol problems were assessed using the 25-item Rutgers Alcohol Problem Index (White & Labouvie, 1989).This measure used a 5-point scale ranging from 0 (Never) to 4 (10 or more times) to assess the frequency of the alcohol-related problems during the past month (e.g., "Got into fights, acted bad, or did mean things?").We also included two additional items that have been frequently used in recent administrations of this measure: "Drove shortly after having two or more drinks" and "Drove shortly after having 4 or more drinks" (Geisner et al., 2018).Responses on the RAPI were summed to create an index of alcohol-related problems.

Alcohol-related risk
We used the Alcohol Use Disorders Identification Test (AUDIT) to assess alcohol-related risk (Babor et al., 2001).The AUDIT is a 10-item measure that assesses frequency and quantity of alcohol use and the presence of problems associated with alcohol use.The 10 items use a 5-point scale ranging from 0 to 4 to assess the extent of hazardous drinking with higher values reflecting more severe levels of that item.Responses on AUDIT items were summed to create an index of alcohol-related risk.The AUDIT has demonstrated strong psychometric properties among college student sample (see Reinert & Allen, 2007).

Data analytic strategy
We used response surface analyses (RSA), a type of polynomial regression, to analyze the impact of alignment in self-(perceived) other attitudes, as well as the level, direction, and shape (i.e., linear vs. curvilinear) of discrepancies in self-other attitudes toward drinking behaviors on alcohol use, consequences, and alcohol-related risk.RSA provides a strong alternative analytic method to the use of difference scores in examining alignment (or discrepancies) in perceptions of behavior (see Edwards and Parry, 1993).Prior to constructing our models, we standardized our predictor variables: self-attitudes toward drinking behaviors (X) and perceptions of others' (i.e., close friends, typical peers, parents) attitudes toward drinking behaviors (Y).We constructed a second-order polynomial model that tested main effects (X, Y), interaction (XY), and quadratic effects (X 2 , Y 2 ) on alcohol-related outcomes (Z) using the equation: 1 Full information maximum likelihood was used to account for missingness for each model; however, rates of missingness overall were very low (lowest observation count was 1,490, 99.7%).R 2 values described the amount of variance in the given outcome explained by the two predictors and their interactions.When R 2 was significant for a given model, we used guidelines from previous research to examine parameters describing the surface response curve (Edwards & Parry, 1993;Rodrigues, 2021;Shanock et al., 2010).The response curve is primarily defined by two axes which form a surface.These two lines are referred to as the line of congruence (X = Y), which characterizes the impact of alignment in the two predictors, and the line of incongruence (X = −Y), which characterizes the impact of discrepancies in the two predictors.The shape of the response surface is further characterized by using the beta estimates from the model to calculate four coefficients (a1-a4).Equations and interpretations of these coefficients characterizing the response surface are described below.
The two parameters describing the slope (a1) and curvature (a2) of the line of congruence (i.e., alignment) characterize the impact of aligned attitudes on alcohol-related outcomes.The a1 coefficient (b 1 + b 2 ) describes how the level of perception (i.e., approval vs. disapproval) is associated with outcomes.For example, in Model 1 which tested the impact of self-other attitudes on alcohol use quantity, a significant and positive a1 coefficient suggests that alignment in approval of drinking is associated with heavier drinking; whereas a negative a1 coefficient suggests that alignment in disapproval is associated with heavier drinking.The a2 coefficient (b 3 + b 4 + b 5 ) describes the shape of aligned perceptions (i.e., linear vs. curvilinear).A significant positive a2 coefficient suggests that individuals are more likely to drink heavily when their perceptions toward drinking behaviors are aligned and particularly strong (i.e., strong approval or strong disapproval).A significant negative a2 coefficient suggests that individuals are more likely to drink heavily when self and other attitudes are aligned, but more neutral in level.
The two parameters describing the slope (a3) and curvature (a4) of the line of incongruence (i.e., discrepancies) characterize the impact of discrepant perceptions on alcohol-related outcomes.The a3 coefficient (b 1 − b 2 ) describes the direction of discrepancies (e.g., whose perceptions are more strongly associated with heavy drinking).For example, a significant positive a3 coefficient suggests that self-perceptions of drinking (versus ratings of others' perceptions) are associated with heavy drinking.A negative a3 coefficient suggests that ratings of others' perceptions of drinking (versus self-perceptions) is associated with heavy drinking.The a4 coefficient (b 3 − b 4 + b 5 ) describes the relationship between discrepancies in perceptions and outcomes.A significant and positive a4 coefficient suggests that discrepancies in perceptions of drinking (e.g., strong self-approval and ratings of others' perceptions that reflect strong disapproval) are associated with heavier drinking.A negative a4 coefficient suggests that less discrepant perceptions are associated with heavy drinking.

Descriptive statistics
On average, participants reported that they consumed 11 alcoholic beverages per week (M DDQ = 11.04;SD DDQ = 9.81), frequency of alcohol-related problems that were consistent with problematic drinking (M RAPI = 5.03; SD RAPI = 4.69; Hagman, 2017) and reported scores on the AUDIT that were slightly below levels of hazardous use on average (M AUDIT = 6.93;SD AUDIT = 3.60; Saunders et al., 1993).Participants expressed minimal approval overall toward drinking behaviors and perceived that their close friends shared similar levels of approval.Participants perceived typical university-affiliated peers to be the most approving toward drinking behaviors (though still disapproving overall), and their parents to be the least approving toward drinking behavior (see Table 1).On average, there was adequate range in the level of alignment between self-and other-attitudes toward drinking behaviors (defined by |Δz| > 0.5, see Table 1).For example, some individuals reported low alignment in self-other attitudes, such that they approved of drinking more than the referent group.Other individuals reported low alignment in self-other perceptions, such that they perceived the referent group to approve of drinking more than themselves.Finally, some individuals reported that they perceived similar attitudes toward drinking behaviors.
Bivariate correlations indicated self-approval was most strongly correlated with perceived approval of close friends (r = 0.80, p < 0.001), followed by parent (r = 0.51, p < 0.001), and typical, university-affiliated peers (r = 0.38, p < 0.001).Participants perceived that their close friends and typical peers (r = 0.47, p < 0.001) and close friends and parents (r = 0.43, p < 0.001) shared somewhat similar levels of approval; whereas they perceived that typical peers and parents shared minimal levels of approval toward drinking behaviors (r = 0.25, p < 0.001).

Model 1: How are self-other perceptions of drinking associated with quantity of weekly alcohol use?
The R 2 values for Model 1 were significant across all reference groups suggesting that a significant amount of the variance in weekly alcohol use was accounted for by self-, other-, or the interactions in self-other perceptions (see Tables 2  and 3; Figure 2).

Self and close friends
On average, alignment in approval of drinking (versus alignment in disapproval; a1 > 0, p < 0.001) demonstrated a linear (versus curvilinear; a2 > 0, p > 0.05) association with heavier alcohol use in a typical week.Although on average, discrepancies in perceptions of drinking were not significantly associated with weekly alcohol use (a4 > 0, p > 0.05), when discrepancies did occur, one's own perceptions of drinking were more often associated with the number of drinks/week than their ratings of their close friends' perceptions (a3 > 0, p < 0.001).

Self and typical peers
Alignment in approval of drinking (a1 > 0, p < 0.05) demonstrated a linear (a2 < 0, p > 0.05) association with heavier alcohol use in a typical week.Additionally, self-peer discrepancies in perceptions of drinking were associated with suggesting that overall, participants reported negative drinking-related attitudes and perceived others to have negative drinking-related attitudes as well.Self > Other refers to discrepancies in which one's own drinking-related attitudes are more positive (or less negative) than one's perceptions of others' drinking-related attitudes.Self < Other refers to discrepancies in which one's own drinking-related attitudes are less positive (or more negative) than one's perceptions of others' drinking-related attitudes.Note.coefficients refer to second-order polynomial regression coefficients, including the intercept (β 0 ), the main effects for self-(β 1 ) and (perceived) other-(β 2 ) attitudes, the quadrative effect for self-attitudes (β 3 ), the interaction effect for self-other attitudes (β 4 ), and the quadratic effect for other-attitudes (β 5 ).all coefficients are standardized (β).*refers to p < 0.05, ** refers to p < 0.001.
heavier alcohol use in a typical week (a4 > 0, p < 0.001), such that one's own perceptions of drinking were more closely associated with quantity of weekly alcohol use than their ratings of a typical peer's perceptions (a3 > 0, p < 0.001).

Self and parents
Alignment in approval of drinking (a1 > 0, p < 0.001) demonstrated a linear (a2 < 0, p > 0.05) association with heavier alcohol use in a typical week.Discrepancies in perceptions of drinking were not significantly associated with weekly alcohol use on average (a4 > 0, p > 0.05), although when these discrepancies did occur, one's own perceptions of drinking were more often associated with the number of drinks/week than their ratings of their parents' perceptions (a3 > 0, p < 0.001).

Model 2: How are self-other perceptions of drinking associated with alcohol related problems?
The R 2 values for Model 2 models were significant across all reference groups (although lower than the R 2 values for Model 1), suggesting that a significant amount of the variance in alcohol-related problems was accounted for by self-, other-, or the interactions in self-other perceptions (see Tables 2 and 3; Figure 3).

Self and close friends
Alignment in approval of drinking (versus alignment in disapproval; a1 > 0, p < 0.001) demonstrated a linear (versus curvilinear; a2 > 0, p > 0.05) association with alcohol-related problems.Discrepancies in perceptions of drinking were not significantly associated with alcohol-related problems on average (a4 > 0, p > 0.05) and both one's own perceptions of drinking, as well as their ratings of their close friends' perceptions were similarly associated with alcohol-related problems (a3 > 0, p > 0.05).

Self and peers
Alignment in approval of drinking (a1 > 0, p < 0.001) demonstrated a linear (a2 > 0, p > 0.05) association with  alcohol-related problems.On average, discrepancies in perceptions of drinking were not significantly associated with alcohol-related problems (a4 > 0, p > 0.05), although when discrepancies did occur, one's own perceptions of drinking were more often associated with alcohol-related problems than their ratings of their peers' perceptions (a3 > 0, p < 0.05).

Self and parents
Alignment in approval of drinking (a1 > 0, p < 0.001) demonstrated a linear (a2 > 0, p > 0.05) association with alcoholrelated problems.Furthermore, self-parent discrepancies in perceptions of drinking were associated with more alcoholrelated problems (a4 > 0, p < 0.05), such that one's own perceptions of drinking were more closely associated with alcohol-related problems than their ratings of their parents' perceptions (a3 > 0, p < 0.001).

Model 3: How are self-other perceptions of drinking associated with alcohol-related risk?
The R 2 values for Model 3 were significant across all reference groups (and of similar magnitude to the R 2 values for Model 1), suggesting that a significant amount of the variance in alcohol-related risk was accounted for by self-, other-, and the interactions in self-other perceptions (see Tables 2 and 3; Figure 4).

Self and close friends
On average, alignment in approval (versus alignment in disapproval; a1 > 0, p < 0.001) demonstrated a linear (versus curvilinear; a2 < 0, p > 0.05) association with greater alcoholrelated risk.On average, discrepancies in perceptions of drinking were not significantly associated with alcohol-related risk (a4 > 0, p > 0.05), although when discrepancies did occur, one's own perceptions of drinking were more often associated with alcohol-related risk than their ratings of their friends' perceptions (a3 > 0, p < 0.05).

Self and peers
Alignment in approval of drinking (a1 > 0, p < 0.05) demonstrated a linear (a2 < 0, p > 0.05) association with greater alcohol-related risk.Furthermore, self-peer discrepancies in perceptions toward drinking behaviors were associated with greater alcohol-related risk (a4 > 0, p < 0.05), such that one's own perceptions of drinking were more closely associated with alcohol-related risk than their ratings of their peers' perceptions (a3 > 0, p < 0.001).

Self and parents
On average, alignment in approval (a1 > 0, p < 0.001) demonstrated a linear (a2 < 0, p > 0.05) association with greater alcohol-related risk.Discrepancies in perceptions of drinking were not significantly associated with alcohol-related risk on average (a4 > 0, p > 0.05); however, when discrepancies did occur, one's own perceptions of drinking were more closely associated with alcohol-related risk than their ratings of their parents' perceptions (a3 > 0, p < 0.001).

Discussion
Research on alcohol-related injunctive norms has largely focused on examining the impact of alignment (or discrepancies) between perceived and actual norms for various reference groups on alcohol-related outcomes (or response to a PNF intervention).However, there has been minimal characterization of other forms of alignment in alcohol-related norms.We characterized discrepancy between one's own attitudes toward drinking and one's perceptions of others' attitudes toward drinking and examined the extent to which alignment (or discrepancies) in self and (perceived) other attitudes toward drinking were associated with drinking quantity, problems, and alcohol-related risk.Overall, we found that greater alignment between self-other approval (versus alignment in disapproval) of drinking-regardless of reference group-consistently demonstrated a linear association with greater indices of alcohol use and related problems.These results suggest that individuals are at increased risk for alcohol-related outcomes when they perceive alignment in approval toward drinking between themselves and others.Furthermore, results suggest that when there were discrepancies in perceptions of drinking, one's own perceptions of drinking were more strongly associated with alcohol-related outcomes than an individual's ratings of others' perceptions.This finding supports previous research demonstrating that one's own positive attitudes toward heavy drinking were associated with alcohol-related outcomes (DiBello et al., 2018), even after accounting for other social-cognitive factors.Additionally, we found that discrepancies in self-peer and self-parent (but not self-friend) perceptions of drinking were significantly associated with alcohol-related outcomes.Therefore, when evaluating risk for alcohol-related outcomes, it may be useful to consider the level to which one approves of their own drinking and then consider the level to which there is shared approval toward drinking.
These findings may help provide future direction regarding adaptation of existing PNF interventions by targeting salient patterns in injunctive norms.The majority of PNF interventions have targeted descriptive norms over injunctive norms (Lewis & Neighbors, 2006;Miller et al., 2013).However, targeting injunctive norms in PNF interventions has demonstrated increased effectiveness in that intervention effects generalize to descriptive norms (in addition to injunctive norms) and have the added benefit of increased user interest and satisfaction (Larimer et al., 2023;Prince et al., 2015;Prince & Carey, 2010).Meta-analytic evidence describes an overall discrepancy in injunctive norms, such that young adults are more likely to perceive themselves as being less approving of drinking as compared to a typical peer and to overestimate the extent to which their peers approve of drinking (Borsari & Carey, 2003).Our results expand on this work and suggest those who perceive greater alignment in attitudes toward drinking may be the most likely to benefit from PNF interventions as there may be more leverage to correct misperceived approval.Furthermore, those who exhibit greater discrepancies between their own perceptions of drinking (e.g., higher approval) and that of other reference groups (e.g., lower approval or disapproval) may be less likely to benefit from PNF interventions as there may be less discrepancy between perceived and actual perceptions (e.g., less room for an intervention effect).However, future work is needed to test whether discrepancies in self-other perceptions are related to PNF intervention response.
Furthermore, the magnitude of the a1 coefficients describing self-friend and self-parent alignment in approval of drinking were consistently larger than the magnitude of the a1 coefficient describing self-peer alignment in approval of drinking, suggesting that aligned approval demonstrated a stronger association with alcohol-related outcomes, particularly weekly alcohol consumption.This finding may help describe why PNF interventions have demonstrated limited response when norms from more proximal groups are targeted (e.g., close friends, parents; Larimer et al., 2009Larimer et al., , 2011;;Lewis & Neighbors, 2004;Stevens et al., 2021).Although PNF interventions are brief, cost-effective, and relatively simple to implement, effect sizes from PNF interventions range from small to medium (Saxton et al., 2021) and tend to decline over time (Carey et al., 2007;Foxcroft et al., 2015); thus, additional focused interventions may be beneficial.Taken together, dyadic prevention and intervention efforts may also be helpful in addressing the relationship between alcohol-related attitudes and alcohol-related risk.Individuals who perceive aligned approval with close friends and parents may benefit from skills-or education-based interventions that promote effective conversations around alcohol use (LaBrie et al., 2018;Mallett et al., 2019).
Previous norms-based work has largely focused on studying the impact of discrepancy in perceived and actual norms within the same reference groups (often one's peers).This study extends the literature by characterizing discrepancies in self-other attitudes toward drinking and the impact of discrepancies in self-other attitudes toward drinking on alcohol-related outcomes.Limitations of this study include a relatively narrow range of four injunctive norms that demonstrated lower alpha reliability.The level of internal consistency was impacted by the relatively few numbers of behaviors assessed and may also reflect systematic differences in drinking-related attitudes toward some behaviors (e.g., drinking alcohol every weekend) vs. other more risky behaviors (e.g., driving a car after drinking).Future work should use measures that assess more drinking-related behaviors (i.e., to increase reliability) as well as a wider range of drinking-related behaviors (i.e., to increase construct validity).Such work could test whether these findings replicate with drink-based measures of injunctive norms (Krieger et al., 2016).Additionally, these data were obtained from a cross-sectional, secondary data collection.Given changing trends in alcohol and substance use (e.g., as cannabis is increasingly legalized for recreational use), ongoing work is needed to understand how discrepancies in alcohol-related norms change over time, particularly in response to changing alcohol use patterns or how discrepancies in alcohol-related norms differ from marijuana (or co-use) norms.
Overall, this study characterized alignment in self-other attitudes toward drinking and examined the impact of alignment (or discrepancies) in self-other attitudes toward drinking on alcohol quantity, problems, and related risk.Perceived alignment in approval of drinking (regardless of reference group), as well as discrepancies in self-peer and self-parent perceptions of drinking were associated with alcohol-related outcomes.Results extend our understanding of the relationship between injunctive norms and alcoholrelated risk and call for future work to test the impact of aligned or discrepant attitudes toward drinking on PNF intervention response.

Notes
1. We conducted a series of sensitivity analyses that included demographic covariates (i.e., age, race, ethnicity, gender identity, and Greek status) to examine whether our results were robust when accounting for covariates.Accounting for covariates had a very minimal impact on the parameter estimates in the model; thus, we report and interpret the findings from the main models in the main text and include results from our sensitivity analyses in supplementary material (see Table S1).

Figure 1 .
Figure 1.illustrating facets of social norms theory that underlie mechanisms of personalized normative feedback.

Figure 2 .
Figure 2. the association between self-(perceived) other attitudes and weekly alcohol consumption.

Figure 3 .
Figure 3. the association between self-(perceived) other attitudes and alcohol problems.

Figure 4 .
Figure 4. the association between self-(perceived) other attitudes and alcohol-related risk.

Table 1 .
Descriptive analyses of approval ratings of drinking (injunctive norms) and discrepancies in approval ratings for each model.

Table 2 .
Standardized coefficients from the second-order polynomial regression model.

Table 3 .
Parameters describing response surface curves from the second-order polynomial regression.Note.a1 describes slope of the line of congruence, a2 describes curvature on the line of congruence, a3 describes slope of the line of incongruence, a4 describes curvature on the line of incongruence.R 2 describes the amount of variance in the outcome variable explained by the second-order polynomial regression equation.*Refers to p < 0.05, ** refers to p < 0.001.