Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Sep 1.
Published in final edited form as: Addict Behav. 2017 Mar 23;72:41–44. doi: 10.1016/j.addbeh.2017.03.014

Resolving an Identity Crisis: Implicit Drinking Identity and Implicit Alcohol Identity Are Related but Not the Same

Jason J Ramirez 1, Cecilia C Olin 1, Kristen P Lindgren 1
PMCID: PMC5490364  NIHMSID: NIHMS864558  PMID: 28366723

Abstract

Two variations of the Implicit Association Test (IAT), the Drinking Identity IAT and the Alcohol Identity IAT, assess implicit associations held in memory between one’s identity and alcohol-related constructs. Both have been shown to predict numerous drinking outcomes, but these IATs have never been directly compared to one another. The purpose of this study was to compare these IATs and evaluate their incremental predictive validity. US undergraduate students (N = 64, 50% female, mean age = 21.98 years) completed the Drinking Identity IAT, the Alcohol Identity IAT, an explicit measure of drinking identity, as well as measures of typical alcohol consumption and hazardous drinking. When evaluated in separate regression models that controlled for explicit drinking identity, results indicated that the Drinking Identity IAT and the Alcohol Identity IAT were significant, positive predictors of typical alcohol consumption, and that the Drinking Identity IAT, but not the Alcohol Identity IAT, was a significant predictor of hazardous drinking. When evaluated in the same regression models, the Drinking Identity IAT, but not the Alcohol Identity IAT, was significantly associated with typical and hazardous drinking. These results suggest that the Drinking Identity IAT and Alcohol Identity IAT are related but not redundant. Moreover, given that the Drinking Identity IAT, but not the Alcohol Identity IAT, incrementally predicted variance in drinking outcomes, identification with drinking behavior and social groups, as opposed to identification with alcohol itself, may be an especially strong predictor of drinking outcomes.

Keywords: Alcohol use, Hazardous drinking, Drinking identity, Alcohol identity, Implicit association test

1. Introduction

Two variants of the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998) measure implicit associations between one’s identity (the self) and alcohol-related constructs. IATs are computer-administered reaction time tasks thought to assess the relative strength of associations between constructs held in memory and have demonstrated utility for predicting hazardous drinking, often predicting above and beyond self-report questionnaires (for reviews, see Lindgren, Neighbors, Gasser, Ramirez, & Cvencek, 2016a; Reich, Below, & Goldman, 2010). The Drinking Identity IAT (DI-IAT; Lindgren et al., 2013b) assesses the relative associations between the constructs me (stimuli: me, my, mine, and self) and not me (stimuli: they, them, theirs, other) with the constructs drinker (stimuli: drinker, drink, drunk, and partier) and non-drinker (stimuli: non-drinker, abstainer, sober, and abstain). The Alcohol-Identity IAT (AI-IAT; Gray, LaPlante, Bannon, Ambady, & Shaffer, 2011), also includes the constructs (and identical stimuli) for me and not me, but assesses their associations with alcohol and water (stimuli: images of alcohol or water, respectively). Both IATs have good psychometric properties and are positively associated with a range of drinking outcomes (Caudwell & Hagger, 2014; Gray et al., 2011; Lindgren et al., 2016a). The DI-IAT has been more widely implemented. It predicts unique variance in multiple drinking outcomes after controlling for other well-validated cognitive factors (e.g., alcohol expectancies, drinking motives, drinking norms; Lindgren, Ramirez, Olin, & Neighbors, 2016b), as well as other alcohol-related IATs (Lindgren et al., 2013a; 2013b), thereby highlighting the unique role of alcohol-related identities as predictors of hazardous drinking.

Despite their similarities, the DI-IAT and the AI-IAT have not been evaluated in the same study. Therefore, it is unknown whether these IATs assess redundant information or whether one is more positively linked to drinking outcomes than the other. Thus, the primary aims of this study were to investigate their relation to one another, the incremental validity of each IAT separately after controlling for explicit drinking identity (which, to date, is unknown for the AI-IAT), and the incremental validity of each IAT when both were included in the same model. We hypothesized that the IATs would be moderately correlated, but not fully redundant. Competing hypotheses were offered and tested regarding the incremental validity of each IAT. First, it is possible that the DI-IAT may be a stronger predictor of drinking outcomes than the AI-IAT. Associations with behaviors and a social group are theorized to be important for substance-related identities (Frings & Albery, 2015), and the category labels and stimuli for the DI-IAT include words that involve drinking behavior and social groups (e.g., drink, partier), whereas the AI-IAT contains only pictures of alcohol itself (e.g., beer, wine). Conversely, AI-IAT may be a stronger predictor than the DI-IAT. Alcohol itself is more proximal to actual drinking outcomes, and the AI-IAT focuses only on alcohol.

2. Method

2.1. Participants

Participants were 64 undergraduate students (50% women) from a large public Pacific Northwestern university (age: M = 21.98, SD = 0.88). Forty-two percent identified as White, 39% as Asian, 8% as multiracial, 6% as African American and 5% as Native Hawaiian, Pacific Islander, or unknown. Five percent identified as Hispanic or Latino.

2.2. Procedure

Procedures were approved by the university’s Institutional Review Board. Students were invited via email to take part in a study on cognitive factors in drinking. Eligible students (at least 21 years old, a full-time student) came to the lab to complete a computer-based assessment. Participants were paid $15.

2.3. Measures

2.3.1. Implicit Association Tests (IATs)

Two IATs were included in the study. The DI-IAT (Lindgren et al., 2013b) assessed the associations between “me” (vs. “not me”) and “drinker” (vs. “non-drinker”). The AI-IAT (adapted from Gray et al., 2011) assessed the associations between “me” (vs. “not me”) and alcohol (vs. water). Both IATs used the traditional seven-block structure (Greenwald et al., 1998) and were computer-administered. Each block contains multiple trials in which participants are presented with a single stimulus and are asked to sort it according to the categories listed on the left or right side of the screen (using the d key for left and the k key for right) as quickly and accurately as possible. Blocks 1, 2, and 5 are practice blocks for learning the sorting rules. Blocks 3, 4, 6, and 7 are critical blocks, in which two categories are sorted on the same side using the same key. For example, in Blocks 3 and 4, one sorts “drinker” and “me” stimuli on the left and “non-drinker” and “not me” stimuli on the right. The order is reversed for Blocks 6 and 7: one sorts “drinker” and “not me” stimuli on the left and “non-drinker” and “me” stimuli on the right. The sorting speed (reaction time) for the first pairing (“drinker” and “me” vs. “non-drinker” and “not me”) is compared to the sorting speed for the second pairing (“non-drinker” and “me” vs. “drinker” and “not me”). The difference in sorting speed is a proxy for the relative strength of the implicit associations (i.e., shorter sorting speed for the first pairing compared to the second would indicate a relatively stronger implicit drinking identity). The order in which IATs were presented was randomized across participants. IATs were interspersed among self-report measures.

IAT scores were calculated using the D-score algorithm (Greenwald, Nosek, & Banaji, 2003). Data were screened following practices recommended by Nosek and colleagues (2007). D-scores were calculated such that higher scores indicated stronger associations with drinker and me (DI-IAT) or alcohol and me (AI-IAT). Internal consistencies (calculated by creating D-scores for Blocks 3 & 6 and Blocks 4 & 7 and correlating them; see Greenwald et al., 2003) were .49 (DI-IAT) and .46 (AI-IAT).

2.3.2. Explicit Drinking Identity

The Alcohol Self-Concept Scale (ASCS) assessed explicit drinking identity (Lindgren et al., 2013b). It is a 5-item measure examining the extent to which drinking alcohol plays a role in an individual’s life and personality (e.g., “Drinking is part of who I am”). Participants rated their agreement using a 7-point scale (−3 = strongly disagree and 3 = strongly agree). Cronbach’s Alpha = 0.89.

2.3.3. Drinking Outcomes

The Daily Drinking Questionnaire (DDQ; Collins, Parks & Martlatt, 1985) examined participants’ typical alcohol consumption in the last three months. Participants reported the number of standard drinks they consumed on each day of a typical week and were summed to represent total drinks per week. The Alcohol Use Disorder Identification Test (AUDIT; Babor, Higgins-Biddle, Saunders, & Monteiro, 2001), a 10-item measure evaluating consumption, consequences, and symptoms of dependence, was used as an index for hazardous drinking. Items were summed. Cronbach’s Alpha = 0.98.

3. Results

3.1. Descriptive Statistics and Correlations

See Table 1 for means and correlations between the study measures1. The IATs were significantly, but moderately correlated with each other (r = .36), and each was significantly associated with explicit drinking identity and both drinking outcomes (r’s between .29 and .52). Mean D-scores were significantly different from zero for both IATs, but were positive for the DI-IAT (M = .17, t58 = 3.53, p = .001) and negative for the AI-IAT (M = −.38, t59 = −9.03, p < .001), indicating that participants were more likely to associate “me” with “drinker” than with “non-drinker” but more likely to associate “me” with water than alcohol.

Table 1.

Correlation Matrix for Study Variables

Measure M (SD) 1 2 3 4 5
1. Drinking Identity IAT 0.17 (0.38) ——
2. Alcohol Identity IAT − 0.38 (0.33) .36** ——
3. Explicit Drinking Identitya − 1.93 (1.16) .34** .39** ——
4. Typical Alcohol Consumptionb 9.64 (11.14) .43** .29* .63** ——
5. Hazardous Drinkingc 7.45 (5.87) .52* .34** .38** .52** ——
Open in a new tab

Note. IAT = Implicit Association Test; higher scores on IATs indicate stronger drinking and alcohol identities;

a

derived from the Alcohol Self-Concept Scale; higher scores indicate greater identification with drinking alcohol;

b

number of standard drinks derived from the Daily Drinking Questionnaire;

c

summary score derived from the Alcohol Use Disorders Identification Test;

*

p < .05

**

p < .01.

3.2. Regression Analyses

The drinking outcome variables were positively skewed. Thus, count regression models with a negative binomial log link were used to test the primary aims. First, models evaluated whether each IAT significantly predicted the outcomes after controlling for explicit identity. Then models evaluated the IATs simultaneously. All models included sex as a covariate.

When evaluated separately, the DI-IAT and the AI-IAT were both significantly and positively associated with alcohol consumption even after controlling for explicit identity (see Table 2). When evaluated simultaneously, the DI-IAT, but not the AI-IAT, was significantly and positively associated with alcohol consumption.

Table 2.

Negative Binomial Regression Models Predicting Drinking Outcomes

Predictors Typical Alcohol Consumptiona
Hazardous Drinkingb
Exp (b) SE b p Exp (b) SE b p
Drinking Identity IAT Model
 Sex 0.56 0.36 .101 0.60 0.21 .017
 Drinking Identity IAT 3.46 0.50 .014 2.06 0.24 .002
 Explicit Drinking Identityc 1.23 0.13 .123 1.28 0.07 .001
Alcohol Identity IAT Model
 Sex 0.44 0.29 .004 0.53 0.19 .001
 Alcohol Identity IAT 3.00 0.53 .037 1.78 0.30 .056
 Explicit Drinking Identity 1.29 0.12 .043 1.31 0.07 <.001
Combined IAT Model
 Sex 0.48 0.34 .032 0.53 0.21 .003
 Drinking Identity IAT 3.23 0.49 .018 2.28 0.28 .003
 Alcohol Identity IAT 2.39 0.62 .158 1.63 0.34 .153
Open in a new tab

Note. Exp (b) represents exponentiated coefficients. Sex was dummy-coded (0 = men, 1 = women); IAT = Implicit Association Test; higher scores on IATs indicate stronger drinking and alcohol identities;

a

number of standard drinks derived from the Daily Drinking Questionnaire;

b

summary score derived from the Alcohol Use Disorders Identification Test;

c

derived from the Alcohol Self-Concept Scale; higher scores indicate greater identification with drinking alcohol.

When evaluated separately, the DI-IAT, but not the AI-IAT, was significantly and positively associated with hazardous drinking. The same pattern of findings was observed when evaluating the IATs simultaneously.

4. Discussion

This is the first study we know of to compare the DI-IAT and AI-IATs. They were related to one another, but not redundant, suggesting functional differences between identification as a drinker, which involves alcohol, a behavior, and a social group, and identification with alcohol, which involves the substance only. The DI-IAT was the more robust predictor of typical and hazardous drinking relative to the AI-IAT, a finding consistent with the hypothesis that identification with a drinker social group (as well as the behavior and alcohol) might make the DI-IAT a stronger predictor of drinking outcomes. Identification with drinking social groups may be particularly important in college given that students predominantly drink in social settings (O’Hara, Armeli, & Tennen, 2015), endorse social drinking motives more than other types of motives (Kuntsche, Knibbe, Gmel & Engels, 2005) and that drinking for social reasons is positively associated with problematic drinking (Vaughan, Corbin, & Fromme, 2009). The importance of identification with social groups has also been previously demonstrated among members of AA and NA: they showed improved treatment outcomes if identification with recovery groups was strong (Buckingham, Frings, & Albery, 2013). Given these findings, novel intervention strategies that seek to reduce identification with drinking social groups may be important to consider on college campuses.

Second, this study is also the first we know of to show that the AI-IAT is a significant positive predictor of typical consumption (but not hazardous drinking) after controlling for explicit identity, providing further support that implicit measures assess a meaningful aspect of drinking identity that may be unavailable for introspection or that participants may be unwilling to report. The DI-IAT was also positively associated with both typical consumption and hazardous drinking after controlling for explicit drinking identity replicating previous studies (Lindgren et al., 2016a), which may add some confidence that the study’s novel findings will replicate.

We also note differences in the overall mean D-scores for the DI-IAT and AI-IAT. Both scores were significantly different from zero but in opposite directions—the mean DI-IAT D-score was positive whereas the mean D-score on the AI-IAT was negative. Typically, DI-IAT scores are positive for individuals who drink and negative for those do not (Lindgren et al., 2013b), which was the case in this study. In contrast, for the AI-IAT, which has been less studied, it appears that scores are simply less negative for drinkers compared to non-drinkers. While that pattern might indicate that participants identified more strongly with the construct of drinker (and the social identity and behavior) versus with alcohol, it might simply reflect the IAT’s construction. It is a relative measure: each IAT has different contrast categories (e.g., AI-IAT: alcohol vs. water; DI-IAT: drinker vs. non-drinker), and those categories are integral to the overall D-score [e.g., the AI-IAT assesses the strength of associations between alcohol + me (and water + not-me) vs. alcohol + not-me (vs. water + me)]. Thus, the negative overall mean for AI-IAT may be a reflection of participants having strong associations with water.

With regard to limitations, the sample consisted of undergraduates and whether results generalize to other populations of drinkers warrants future research. It is possible that the DI-IAT is a robust predictor of drinking for college students, who predominantly drink in social settings (O’Hara, Armeli, & Tennen, 2015), but this may not be the case for populations of drinkers who are less likely to drink in social contexts. Also, the study may have been underpowered to detect small to medium effects (e.g., AI-IAT as a unique predictor of hazardous drinking) due to the small sample size. Replication with larger samples and additional alcohol-related outcomes will be important.

Ultimately, both variants of alcohol-related IATs appear related to one another and associated with typical drinking above and beyond explicit drinking identity. However, only the DI-IAT was associated with hazardous drinking after controlling for explicit drinking identity, and it was a stronger predictor of drinking outcomes. These findings potentially highlight the importance of identification with drinking as a behavior and drinking social groups.

Highlights.

  • We compared two alcohol-related identity implicit association tests (IATs).

  • The two IATs were the Drinking Identity IAT and the Alcohol Identity IAT.

  • We tested if each predicted incremental variance in typical and hazardous drinking.

  • Both IATs were positively correlated to each other and both drinking outcomes.

  • Only the Drinking Identity IAT predicted unique variance in drinking outcomes.

Footnotes

1

Of the 64 participants, 53 reported drinking alcohol in the previous three months, and had stronger mean D-scores than 11 non-drinkers for the DI-IAT (drinkers: M = 0.23, SD = 0.38; non-drinkers: M = −0.08, SD = 0.28; t57 = 2.87, p = .011) and the AI-IAT (drinkers: M = −0.31, SD = 0.29; non-drinkers: M = −0.71, SD = 0.28; t58 = 3.85, p = .002).

Author Disclosure Statements

This work was supported by National Institute on Alcohol Abuse and Alcoholism grants T32AA007455 (PI: Larimer) and R01s AA021763 and AA024732 (PI: Lindgren). NIAAA had no role in the study design, collection, analysis or interpretation of the data, writing the manuscript, and the decision to submit the manuscript for publication.

JJR wrote the first draft of the manuscript excluding the Methods section. CCO wrote the first draft of the Methods section and oversaw data collection. KPL designed the study, wrote the protocol, conducted most statistical analyses, and edited the manuscript. All authors contributed to and have approved the final manuscript.

The authors have no conflicts of interest to disclose.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Babor TF, Higgins-Biddle JC, Saunders JB, Monteiro MG. AUDIT: The Alcohol Use Disorders Identification Test: Guidelines for use in primary care. 2. Geneva: World Health Organisation; 2001. [Google Scholar]
  2. Buckingham SA, Frings D, Albery IP. Group membership and social identity in addiction recovery. Psychology of Addictive Behaviors. 2013;27(4):1132–1140. doi: 10.1037/a0032480. [DOI] [PubMed] [Google Scholar]
  3. Caudwell KM, Hagger MS. Pre-drinking and alcohol-related harm in undergraduates: the influence of explicit motives and implicit alcohol identity. Journal of Behavioral Medicine. 2014;37(6):1252–1262. doi: 10.1007/s10865-014-9573-6. [DOI] [PubMed] [Google Scholar]
  4. Collins RL, Parks GA, Marlatt GA. Social determinants of alcohol consumption: The effects of social interaction and model status on the self-administration of alcohol. Journal of Consulting and Clinical Psychology. 1985;53(2):189–200. doi: 10.1037//0022-006x.53.2.189. [DOI] [PubMed] [Google Scholar]
  5. Frings D, Albery IP. The Social Identity Model of Cessation Maintenance: Formulation and initial evidence. Addictive Behaviors. 2015;44:35–42. doi: 10.1016/j.addbeh.2014.10.023. [DOI] [PubMed] [Google Scholar]
  6. Gray HM, LaPlante DA, Bannon BL, Ambady N, Shaffer HJ. Development and validation of the Alcohol Identity Implicit Associations Test (AI-IAT) Addictive Behaviors. 2011;36(9):919–926. doi: 10.1016/j.addbeh.2011.05.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Greenwald AG, McGhee DE, Schwartz JK. Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology. 1998;74(6):1464–1480. doi: 10.1037/0022-3514.74.6.1464. [DOI] [PubMed] [Google Scholar]
  8. Greenwald AG, Nosek BA, Banaji MR. Understanding and using the implicit association test: I. An improved scoring algorithm. Journal of Personality and Social Psychology. 2003;85(2):197–216. doi: 10.1037/0022-3514.85.2.197. [DOI] [PubMed] [Google Scholar]
  9. Kuntsche E, Knibbe R, Gmel G, Engels R. Why do young people drink? A review of drinking motives. Clinical Psychology Review. 2005;25(7):841–861. doi: 10.1016/j.cpr.2005.06.002. [DOI] [PubMed] [Google Scholar]
  10. Lindgren KP, Foster DW, Westgate EC, Neighbors C. Implicit drinking identity: drinker + me associations predict college student drinking consistently. Addictive Behaviors. 2013a;38(5):2163–2166. doi: 10.1016/j.addbeh.2013.01.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Lindgren KP, Neighbors C, Gasser ML, Ramirez JJ, Cvencek D. The American Journal of Drug and Alcohol Abuse. 2016a. A review of implicit and explicit substance self-concept as a predictor of alcohol and tobacco use and misuse; pp. 1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Lindgren KP, Neighbors C, Teachman BA, Wiers RW, Westgate E, Greenwald AG. I drink therefore I am: Validating alcohol-related implicit association tests. Psychology of Addictive Behaviors. 2013b;27(1):1–13. doi: 10.1037/a0027640. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Lindgren KP, Ramirez JJ, Olin CC, Neighbors C. Not the same old thing: Establishing the unique contribution of drinking identity as a predictor of alcohol consumption and problems over time. Psychology of Addictive Behaviors. 2016b;30(6):659–671. doi: 10.1037/adb0000195. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. O’Hara RE, Armeli S, Tennen H. College students’ drinking motives and social-contextual factors: Comparing associations across levels of analysis. Psychology of Addictive Behaviors. 2015;29(2):420–429. doi: 10.1037/adb0000046. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Nosek BA, Greenwald AG, Banaji MR. The implicit association test at age 7: A methodological and conceptual review. In: Bargh JA, editor. Automatic processes in social thinking and behavior. Psychology Press; 2007. pp. 265–292. [Google Scholar]
  16. Reich RR, Below MC, Goldman MS. Explicit and implicit measures of expectancy and related alcohol cognitions: a meta-analytic comparison. Psychology of Addictive Behaviors. 2010;24(1):13–25. doi: 10.1037/a0016556. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Vaughan EL, Corbin WR, Fromme K. Academic and social motives and drinking behavior. Psychology of Addictive Behaviors. 2009;23(4):564–576. doi: 10.1037/a0017331. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES