Association of the Firewater Myth with Drinking Behavior Among American Indian and Alaska Native College Students

Abstract

The firewater myth (FM) is the notion that American Indians and Alaska Natives (AI/ANs) are more susceptible to the effects of alcohol and vulnerable to alcohol problems due to biological or genetic differences. Believing that one is vulnerable to problems with alcohol may have negative effects on expectancies and drinking behavior among AI/ANs who drink; however, the association of belief in the FM with alcohol outcomes has not previously been examined. In this study we examined the factor structure of a revised version of the Firewater Myth Scale (LaMarr, 2003) and the association of belief in the FM with alcohol use, consequences, attitudes, and expectancies with 159 AI/AN college students who drink. On average participants “slightly agreed” with the FM and scores were normally distributed. There were significant small to moderate positive associations between believing that AI/ANs have a biological vulnerability to problems with alcohol (i.e., the FM) and drinks consumed per week, frequency of heavy episodic drinking, and alcohol consequences, as well as belief in a disease model of "alcoholism,” attempts to control drinking, guilt over drinking small amounts of alcohol, both positive and negative alcohol expectancies, temptation to drink heavily, and lack of self-efficacy to drink moderately. Although this is only an initial examination of potential consequences of belief in the FM for AI/AN students who drink, the results suggest that this belief may be harmful and have negative effects on attempts to moderate drinking.

The firewater myth (FM) is the notion that American Indians and Alaska Natives (AI/ANs) are more susceptible to the effects of alcohol and more vulnerable to alcohol problems due to biological or genetic differences (LaMarr, 2003; Mail, 2002). Although genetics clearly play a role in the risk for an alcohol use disorder (Köhnke, 2008), there is little evidence to support the notion that biological differences or genetics play a greater role in alcohol use disorders among AI/ANs compared to other racial groups (Ehlers, Liang, & Gizer, 2012; Garcia-Andrade, Wall, & Ehlers, 1997; Gizer, Edenberg, Gilder, Wilhelmsen, & Ehlers, 2011; Mail & Johnson, 1993). However, the idea that alcohol-related health disparities affecting AI/ANs are driven by biological variables remains pervasive in the general public and among AI/AN peoples. This is problematic for a number of reasons. Not only is this notion unsupported by evidence, but believing that one’s ethnic/racial group has a particular biological vulnerability to alcohol use disorders could have negative effects on alcohol expectancies and drinking behavior among AI/ANs (LaMarr, 2003; Quintero, 2001). Although it has been suggested that the FM may affect AI/ANs’ drinking behavior, to date this has not been examined in the empirical literature. In this study we propose and examine potential theoretical links between the FM and AI/ANs’ alcohol use and outcomes.

Alcohol-related stigma and stereotypes

In the US, AI/ANs have a distinct history with alcohol compared with other ethnic or racial groups. Prior to colonization, alcohol use was very limited among AI/ANs (Abbott, 1996). Subsequently, alcohol use was facilitated by colonists and traders, giving rise to alcohol problems and eventually to the stereotype of the drunken Indian (Mail, 2002). In 1832, the U.S. Congress passed a law prohibiting the sale of liquor to all Native peoples, which was in effect until Congress gave tribes control over tribal alcohol policy in 1953 (Kovas, McFarland, Landen, Lopez, & May, 2008). This law supported the notion that AI/ANs are biologically different from Europeans and are unable to “hold their liquor,” which was thought to make them especially vulnerable to alcohol problems (i.e., the firewater myth), with complete abstinence seen as the only solution (LaMarr, 2003; Mail, 2002; Westermeyer, 1974).

Many tribes have chosen to adopt or to retain stricter alcohol control policies than those adopted by their respective state laws governing the sale and possession of alcohol (Kovas et al., 2008). In 1981, Alaska state law gave community residents the option to vote for local alcohol control policies, called the local option law. Prohibition of the importation and sale of alcohol has been implemented almost exclusively in Alaska Native communities in the state, suggesting that Alaska Natives have a more favorable view of prohibition while few non-Natives support it (Berman, Hull, & May, 2000). The adoption of prohibition on some tribal lands and Alaska Native communities appears to demonstrate a belief in the FM, with restriction or prohibition seen as a potential solution to community alcohol problems (LaMarr, 2003).

Among AI/ANs, there is evidence of negative attitudes toward drinking at any level. It has been suggested that this may stem from negative stereotyping regarding AI/AN peoples and alcohol, concern regarding greater alcohol-related consequences among some tribes, or belief in the FM (Daisy, Thomas, & Worley, 1998). In a study conducted with Navajo participants, even minimal alcohol consumption (i.e., one drink) was seen as “a bad thing to do” by the majority of both non-drinkers (71% of males and 74% of females) and drinkers (57% of males and 62% of females; May & Smith, 1988). Likely related to these findings, 63% of Navajo participants agreed that American Indians have a unique physical weakness to alcohol (i.e., endorsed the FM) and 94% agreed that “Indians have a problem with alcohol.” A more recent study with participants from five different tribes similarly found that most American Indians participants held negative opinions about alcohol (Yuan et al., 2010) and in a study of rural Alaska Natives, participants expressed the belief that it was not possible for Alaska Natives to consume any alcohol without experiencing harm (Mohatt et al., 2004).

Self-fulfilling prophecy

The FM in general has contributed to racial stereotyping, stigma, and discrimination. Further, believing that by virtue of one’s race, one is particularly sensitive to the effects of alcohol and more likely to develop alcohol problems ― and thus should not drink or at least only do so with great caution ― could cognitively set the stage for future fulfillment of that expectation among those who are not lifetime abstainers (LaMarr, 2003). In addition to conceptualizing this as a self-fulling prophecy, this effect also could be likened to stereotype threat, in that stress resulting from concern about fulfilling a negative AI/AN stereotype may impair ability to self-regulate alcohol use (Blume, 2016).

The drinking restraint model provides a theoretical mechanism through which the FM could be associated with excessive drinking and alcohol problems among AI/AN drinkers. In general, this model suggests a cycle of drinking restraint (i.e., efforts to control alcohol use) and lapses in this control in the form of excessive drinking (Bensley, 1991; Southwick Bensley, 1989). Although limiting alcohol intake can be quite beneficial by helping one to avoid alcohol-related problems or health consequences, at times self-imposed limits can paradoxically result in more drinking (Collins, 1993; Muraven, Collins, Morsheimer, Shiffman, & Paty, 2005a). The limit violation effect occurs when self-imposed drinking limits are exceeded, resulting in guilt or other negative affect, which then triggers more drinking (Collins, 1993). Daily process research examining how events or emotions experienced on a given day affect subsequent alcohol consumption that day or the next has shown that the limit violation effect occurs as theorized, with exceeding one’s personal limit triggering subsequent distress (i.e., feeling guilty and bad about the amount consumed) the morning after, regardless of the actual amount consumed. This distress then triggers more drinking (Muraven et al., 2005a).

An important factor in triggering negative affect in response to limit violations is the personal meaning ascribed to the limit violation. If a drinker attributes their failure to exercise restraint to something personal, global, and stable, then it is more likely to result in negative affect and further excessive drinking (Collins, 1993; Marlatt, 1985). Daily process research has found that individuals who make internal attributions when they exceed their drinking limit experience more negative mood, and drink more in response (Muraven, Collins, Morsheimer, Shiffman, & Paty, 2005b). For AI/ANs who drink, belief in having a biological vulnerability to alcohol problems by virtue of their race may have a particularly salient meaning associated with excessive drinking that is global, stable, and internal. Thus the FM may paradoxically lead to excessive drinking because of heightened preoccupation with the need to limit drinking, guilt in response to drinking, and the attributions made when personal limits are violated.

In addition, believing in the FM could be associated with increased alcohol consumption and consequences due to greater temptation to drink and positive alcohol expectancies. AI/ANs who believe in the FM would likely be concerned with controlling their drinking given their belief that they are especially vulnerable to the effects of alcohol and have an increased risk for developing a drinking problem. Among those who choose to drink despite belief in the FM, temptation to drink would by necessity be high in order to override their concerns. Further, while greater belief in the FM is likely associated with negative expectancies regarding alcohol (Mohatt et al., 2004; Yuan et al., 2010), expectancies that alcohol has beneficial effects (i.e., positive expectancies) also would need be high in order to motivate drinking despite belief in the FM. In the absence of a strong temptation to drink, individuals who believe in the FM may be more likely to choose abstinence or to drink lightly in order to avoid the alcohol problems to which they believe they are particularly susceptible. In general, both temptation and positive alcohol expectancies are associated with more drinking and more alcohol problems (Carey, 1995; Connors, Collins, Dermen, & Koutsky, 1998; Cox et al., 2001; Jones, Corbin, & Fromme, 2001).

The FM also could put AI/ANs who drink at particular risk for excessive alcohol use and consequences because it may undermine self-efficacy for moderate drinking and the use of protective behavioral strategies for avoiding alcohol-related consequences. The FM is essentially based on a disease model, which emphasizes biological or genetic underpinnings for alcohol problems and abstinence as a solution or as prevention. Consistent with this notion, belief in the FM among AI/AN college students has been found to be associated with belief in a disease model for alcohol problems (LaMarr, 2003). If AI/ANs who drink believe they are vulnerable to developing alcohol problems due to biological factors that are out of their control, they may lack self-efficacy regarding their ability to moderate their drinking. Self-efficacy to avoid drinking or to moderate drinking is an important factor in actually doing so (Oei, Fergusson, & Lee, 1998; Oei & Morawska, 2004). When people lack self-efficacy that they can do something, they are less likely to try to perform that behavior and are less likely to be successful if they do.

Current study

The FM is a potentially important construct that may affect attitudes toward alcohol and drinking behavior among AI/ANs; however, we could only locate one unpublished scale measuring belief in the FM, which was developed by C. June LaMarr (2003). For this study, we revised LaMarr’s Firewater Myth Scale (FMS) given her conclusion that the relatively low endorsement of items found in her study, which averaged “slightly disagree,” may have been related to participants’ offense regarding the scale’s items. When revising the scale, we primarily focused on items related to genetic or biological vulnerability, although LaMarr’s scale also measured aspects of the FM relating to out of control behavior while drinking and the drunken Indian stereotype. LaMarr (2003) found that items relating to genetic or biological differences had greater endorsement among AI/AN students and concluded that these aspects are more central to the FM, while subscales such as Drunken Indian had relatively low item endorsement and were more overtly offensive.

In addition to creating a revised version of the FMS scale that was meant to elicit less reactance from participants, we also tested hypotheses relating the theoretical effects of belief in the FM to drinking, alcohol problems, and factors that affect these outcomes. Based on the drinking-restraint model and previous research, we hypothesized that among AI/AN college students who drink, greater belief in the FM would be associated with greater alcohol use and problems, belief in a disease model of "alcoholism,” both negative alcohol expectancies and positive alcohol expectancies, guilt associated with drinking even small amounts of alcohol, greater temptation to drink heavily, and lower self-efficacy to resist drinking heavily. We also hypothesized that having previously lived in a village or town that had adopted greater alcohol control policies (e.g., prohibition) would be associated with greater belief in the FM.

Method

Participants

Participants were American Indian or Alaska Native women (69.8%, n = 111) and men (30.2%, n = 48) attending one of two large, open enrollment universities in different regions of Alaska (campus A, n = 121; campus B, n = 38). Participants’ mean age was 27.10 years (SD = 9.12), with a range from 18 to 61 years of age; 72.3% of the sample was under 30 years old. Although the age range in this sample was higher than the typical mean age for a U.S. college student sample, it was representative of the average age for college students in Alaska. The majority of participants were full-time students (73.9%). Approximately 74% of the sample had never been married and 75% did not have children.

Procedures

The study protocol was approved by the Institutional Review Boards of both universities where the study was conducted. Participants were recruited via email solicitations, fliers posted on campus, in-class announcements, and advertisements in the school newspapers. Advertisements noted that the study sought AI/AN college students for a study of attitudes toward alcohol interventions and factors that affect those attitudes. Advertisements directed potential participants to a webpage that screened for eligibility. Inclusion criteria included self-identifying as (a) AI/AN; (b) a college student; (c) age 18 or older; and (d) a current drinker, defined as having consumed one or more standard alcoholic drinks in the 30 days prior to screening. The alcohol inclusion criterion was selected to ensure a wide range of alcohol use behaviors and consequences in the study sample, while excluding those who were abstinent.

Eligible individuals were scheduled for a single in-person data collection session on their respective campus, where study materials were presented in random order on laptop computers using MediaLab software (Jarvis, 2006). Participants were compensated with a $25 Visa gift card for their time.

Measures

Revised Firewater Myth Scale (RFMS)

The authors and an advisory group of AI/AN and psychology graduate students from other ethnic backgrounds revised the original FMS (LaMarr, 2003) by adding, eliminating, and rewording items. Some of these reworded items were phrased to be reverse scored (i.e., indicating lack of endorsement of the FM). Consistent with the original FMS, which included distractor items in an effort to decrease the blatant aim of the scale, we created items that were parallel to those that asked about AI/AN people. These items capitalized on common American stereotypes regarding alcohol and other ethnic groups (for example, “People of Irish descent are more likely to have a genetic vulnerability to problems with alcohol.”). We also included items that were simply accurate based on epidemiological surveys (e.g., higher rates of abstinence among African Americans, lower rates of drinking problems among Asian Americans). In total, 14 items referred to AI/ANs and 21 referred to other ethnic or racial groups (e.g., African American, Hispanic, White/European American). Items on the original FMS and the Revised FMS are rated from strongly agree (1) to strongly disagree (6). Only the 14 items relevant to AI/ANs were used to measure belief in the FM. After appropriate reverse scoring, higher scores indicate greater agreement with the FM.

Alcohol use and consequences

Alcohol use during the past year was measured using six items modified from the National Institute of Alcohol Abuse and Alcoholism’s (NIAAA) alcohol consumption question set (Gonzalez, Reynolds, & Skewes, 2011; National Institute on Alcohol Abuse and Alcoholism, 2003). Items were open response and asked about typical monthly drinking during the past 12 months. Items inquired about (1) drinking days per month, (2) standard drinks consumed on a typical drinking day, and (3) days on which heavy episodic drinking (i.e., ≥4 standard drinks for a female or ≥5 standard drinks for a male on one occasion or sitting) occurred with separate questions inquiring about these drinking variables in social and in solitary contexts. Participants were provided with a handout that defined and depicted a standard drink (e.g., 12 oz. of beer, 5 oz. of wine, 8 to 9 oz. of malt liquor, or 1.5 oz. of 80-proof liquor). We computed quantity × frequency products for both social and solitary drinking, summed the result, and then divided by four to obtain an overall quantity-frequency index of typical drinks consumed per week in the past 12 months, regardless of context. Similarly, we summed typical frequency of heavy episodic drinking in social and solitary contexts per month to represent frequency of heavy drinking episodes during a typical month in the past year.

For descriptive purposes, the Alcohol Use Disorders Test—Consumption (AUDIT-C; Bush et al., 1998) total score cutoff score (≥ 4 for women, ≥ 5 for men) was used to classify participants as potential problem drinkers or non-problem drinkers. These cut-points have been shown to have good specificity and sensitivity when classifying people with alcohol use disorders across a number of studies and populations (Reinert & Allen, 2007).

The Young Adult Alcohol Consequences Questionnaire (YAACQ; Read, Kahler, Strong, & Colder, 2006) is a 48-item self-report inventory of problems associated with alcohol use. Items are rated dichotomously as present (1) or absent (0) in the past year. The YAACQ has demonstrated high test-retest reliability, as well as good convergent, concurrent, and predictive validity with college students (Kahler, Strong, & Read, 2005; Read, Merrill, Kahler, & Strong, 2007). The YAACQ total score was used to quantify alcohol problems in the last year. In this sample, the alpha coefficient for the total scale was .96.

Alcohol expectancies

The Comprehensive Effects of Alcohol Questionnaire (CEOA; Fromme, Stroot, & Kaplan, 1993) is a 38-item measure of positive (20 items) and negative (18 items) alcohol expectancies. Participants rate their level of agreement that a particular effect would likely occur to them if they were under the influence of alcohol from disagree (1) to agree (4). The CEOA has demonstrated good concurrent validity and test-retest reliability (Fromme et al., 1993; Ham, Stewart, Norton, & Hope, 2005). In this sample, the alpha coefficient was .85 for negative expectancies and .88 for positive expectancies.

Efforts to control drinking

The Short Readiness to Change Questionnaire (Rollnick, Heather, Gold, & Hall, 1992) is a 12-item measure of motivation to change one’s drinking. Items are rated from strongly disagree (−2) to strongly agree (+2) The action subscale is made up of four items relating to efforts to change one’s drinking habits (e.g., “I am trying to drink less than I used to.”). This subscale was used in the current study to provide a measure of efforts to control drinking and had an alpha coefficient of .83.

Self-efficacy and temptation

The Alcohol Abstinence Self-Efficacy Scale (DiClemente, Carbonari, Montgomery, & Hughes, 1994) is a measure of temptation to drink in a variety of situations and confidence (i.e., self-efficacy) to resist doing so, with 20 items in each subscale. For this study, a modified version was used in which participants were asked to indicate how tempted they would be to drink heavily in each situation, and how confident they were that they could resist heavy drinking (Heather et al., 2010). Items were rated from not at all (1) to extremely (5). Internal consistency for this study was .94 for self-efficacy and .93 for temptation.

Beliefs about alcoholism

The Short Understanding of Alcoholism Scale (SUAS; Humphreys, Greenbaum, Noke, & Finney, 1996) is a measure of beliefs about the nature of alcoholism. Items are rated from strongly disagree (1) to strongly agree (5). The disease model subscale is comprised of 7 items that relate to the belief that “alcoholism” is a progressive, incurable disease that can only be arrested by abstinence. The psychosocial model subscale is comprised of 5 items that relate to belief that alcoholism is a learned behavior influenced by culture and family environment. The SUAS has demonstrated discriminant validity, showing differences between treatment staff from varied disciplines that ascribe to different models of addiction (Humphreys et al., 1996). In this sample, the alpha coefficients were .79 for disease model and .74 for psychosocial model.

Guilt related to drinking

For this study, six items were developed to measure guilt associated with drinking. Participants were asked, “How bad or guilty would you feel if: (1) you had 1 or 2 beers, glasses of wine, or wine coolers; (2) you had 3 or 4 beers, glasses of wine, or wine coolers; (3) you had 5 or more beers, glasses of wine, or wine coolers.” These items were rated from not at all bad or guilty (0) to extremely bad or guilty (3). Three parallel items asked about guilt associated with drinking these same amounts in “shots of hard liquor.”

Local option law

For this study, participants were asked if they had previously lived in an Alaskan village or town that had adopted a local option law limiting the sale, possession, or importation of alcohol. Participants were asked to indicate if: (a) sale, importation, and possession was prohibited; (b) sale of alcohol was prohibited, but importation for personal use was allowed; (c) sale and importation was prohibited, while possession was not; or (d) sale of alcohol was limited to only one specific, licensed store.

Analyses

The factor structure of the 14 AI/AN specific items on the RFMS was examined using exploratory factor analysis with principal axis factor extraction with oblimin (oblique) rotation. We then used the scale to measure hypotheses regarding the FM. Separate multiple regression analyses were conducted to examine the association of the FM with the following dependent variables: (a) drinks per week, (b) frequency of heavy episodic drinking per month, (c) alcohol consequences, (d) self-efficacy to not drink heavily, (e) temptation to drink heavily, (f) positive expectancies, and (g) negative expectancies. Two separate multiple regressions also were conducted to examine the association of (a) having previously lived in a town or village that chose to adopt additional restrictions on alcohol and (b) belief in disease or psychosocial model of alcoholism (entered in the model simultaneously), with belief in the FM as the dependent variable. Finally, we examined the association of belief in the FM with six items measuring guilt associated with drinking varying amounts of (a) beer, wine, or wine coolers or (b) hard liquor. Partial correlations were used to examine these associations for all items except the item relating to guilt for having one or two beers, glasses of wine, or wine coolers. Relatively few participants (16.7%) reported any guilt for drinking this amount, therefore this item was dichotomized to represent no guilt (0) or any guilt (1). Logistic regression was used to examine the association of the FM with this variable. In all analyses age, gender, and campus were entered as covariates.

Prior to analyses, data were screened following the procedures outlined in Tabachnick and Fidell (2014). In order to improve normality and reduce the influence of outliers, drinks per week and heavy episodic drinking were square-root transformed. One invariant responder and two multivariate outliers were eliminated from the study. Missing data were minimal, ranging from 0% to 1.85% across measures, and were the result of computer or administration errors rather than owing to participants choosing to omit ratings. Given both of these factors, we choose to use pairwise deletion for missing data. Sample size across analyses ranged from 156 to 159.

Results

Factor structure of the Revised Firewater Myth Scale

Based on mean scores for AI/AN items on the RFMS (see Table 1), for most items participants “slightly agreed” or for reverse scored items “slightly disagreed,” indicating a tendency toward agreement with the FM. The scree plot for the exploratory factor analysis suggested a three factor solution, while a parallel analysis suggested a four factor solution. Both a three and four factor solution were examined. A three factor solution provided the most interpretable results and best simple structure; therefore, the three-factor model was retained.

Table 1.

Exploratory Factor Analysis of American Indian/Alaska Native Items of the Revised Firewater Myth Scale^a: Factor Loadings, Communalities, Means, and Standard Deviations

		Factor Loadings
Item		1	2	3	h2	M	SD
31.	Alaska Natives and American Indians are more likely to have a genetic vulnerability to problems with alcohol.	.86	−.17	−.05	.68	4.28	1.54
23.	Alaska Natives and American Indians metabolize alcohol differently than non-Native people.	.74	.07	−.17	.53	3.84	1.55
18.	Alaska Natives and American Indians feel the effects of alcohol, or feel intoxicated, more easily than people of European descent.	.63	.22	.10	.55	3.81	1.55
27.	Because of biological or genetic differences, it would be best for Alaska Natives and American Indians to never drink or stop drinking completely.	.61	−.06	.11	.41	3.69	1.69
6.	Drinking even small amounts of alcohol is harmful for many Alaska Natives and American Indians.	.56	−.12	.13	.37	3.96	1.61
8.	Alaska Natives and American Indians are more likely to be born with a tendency to develop problems with alcohol.	.54	−.16	.12	.35	4.35	1.40
35.	Alaska Natives and American Indians have a shorter history of exposure to alcohol compared to people of European descent, therefore are more susceptible to the effects of alcohol.	.53	.20	−.17	.34	4.59	1.49
16.	Alaska Natives and American Indians are as biologically capable of handling alcohol as non- Native people.	.48	.36	.12	.47	3.77	1.65
20.	Alaska Natives and American Indians have as much control over their drinking as non-Native people.1	.41	.09	.42	.47	3.69	1.65
3.	Alaska Native and American Indians can hold their liquor better than non-Natives.	−.06	.73	.03	.52	4.72	1.52
33.	Alaska Native and American Indians are proud of how well they can hold their liquor.	.00	.57	−.01	.33	4.14	1.44
21.	Most Alaska Natives and American Indians who drink don’t have a problem with alcohol.	.02	.16	.68	.48	4.12	1.42
13.	Alaska Natives and American Indians don’t drink any more than any other ethnic or racial group.	.03	.02	.50	.26	4.12	1.34
10.	Many Alaska Natives and American Indians don’t drink at all.	−.03	−.13	.35	.15	3.50	1.63
% of variance explained		27.70%	8.54%	5.89%

Note. Items in bold are reverse scored, with higher mean scores indicating greater agreement with the item. Higher scores for items that are not in bold indicate greater disagreement. Shading indicates which items were retained on a given factor.

^aThe scale in its entirety includes distractor items that refer to other ethnic/racial groups.

An examination of the pattern coefficients for the rotated three-factor solution revealed that nine items loaded on the first factor. Consistent with the aim of the scale, the first factor related to a biologically-based (e.g., genetics, alcohol metabolism) vulnerability to alcohol problems (see Table 1). Two items (items 16 and 20) cross-loaded on other factors but were retained on Factor 1 given the primary purpose of the measure. This factor had a Cronbach’s alpha of .85 and was named Biological Vulnerability. Factor 2 was comprised of only two items, both alternative wordings relating to AI/AN people’s ability to “hold their liquor.” The alpha coefficient for these items was .64. Factor 3 had three items without cross-loadings relating to the belief that AI/ANs drink more or that AI/ANs who drink have more alcohol problems than non-Natives. Item 10 loaded onto Factor 3 relatively weakly, with a factor loading < .40, and displayed a low communality; therefore, this item was dropped. The alpha for the remaining two items on Factor 3 was .51; this factor was named Alcohol Problems.

There was a moderate zero-order correlation between the Biological Vulnerability and Alcohol Problems factors (r = .32, p < .001). Factor 2 relating to AI/AN’s lack of ability to “hold their liquor” was not significantly associated with the Alcohol Problems (r = −.02, p = .833) or Biological Vulnerability factors (r = .15, p = .064).

The distributions of scores on the Biological Vulnerability and Alcohol Problems factors were normal, with neither skew nor kurtosis significantly different than zero for either factor. Overall, mean scores on both factors suggest the sample tended to “slightly agree” with both (see Table 2). The mean score for the two items relating to “holding their liquor” suggest general disagreement with the notion that AI/ANs have greater alcohol tolerance than non-Natives. The pattern of loadings for these items (see Table 1), as well as the non-significant correlations with the Biological Vulnerability and Alcohol Problems factors, suggest that these items are not associated with the FM; therefore this subscale was not included in subsequent analyses. The third factor relating to AI/AN people having more problems with alcohol was comprised of only two items and their content was in part included in the scale to provide positively worded items regarding AI/AN alcohol problems. These items loaded together to form a factor that, while not ideal given the few items and relatively low alpha level, provided an index of belief that AI/AN people drink more and have greater alcohol problems than non-Natives. In the current study the planned analyses were conducted with both the Biological Vulnerability and Alcohol Problems factors, allowing for an examination of the convergence or divergence in their associations with alcohol-related variables.

Table 2.

Means, Standard Deviations, and Intercorrelations of Study Variables

	Variables	M	SD	1	2	3	4	5	6	7	8	9	10	11	12	13
1.	RFMS Biological Vulnerability	4.00	1.06	--
2.	RFMS Alcohol Problems	4.12	1.13	.32 ***	--
3.	Drinks per week a	9.20	12.33	.17*	.06	--
4.	Heavy episodic drinking (days/month) a	4.35	5.95	.18*	.06	.80***	--
5.	Alcohol consequences	15.93	12.12	.30***	.09	.51***	.68***	--
6.	Efforts to control drinking	1.18	4.19	.28***	.09	.33***	.28***	.43***	--
7.	Self-efficacy	3.67	.80	−.15	−.04	−.32***	−.36***	−.52***	−.26**	--
8.	Temptation	2.18	.73	.20*	.15	.34***	.43***	.64***	.27***	−.73***	--
9.	Positive expectancies	2.86	.50	.18*	−.04	.16*	.26**	.29***	.22**	−.40***	.38***	--
10.	Negative expectancies	2.57	.51	.35***	.00	.19*	.22**	.45***	.28***	−.34***	.46***	.39***	--
11.	Disease model	1.99	.88	.41***	.12	.04	.07	.14	.20*	−.04	.07	.02	.36***	--
12.	Psychosocial model	2.85	.62	.14	.22**	.05	−.04	.14	.11	−.12	.24**	.09	.21**	.08	--
13.	Age	27.10	9.12	.08	−.09	−.10	−.17*	−.13	−.00	.30***	−.25**	−.25**	.01	.19*	−.11	--
14.	Genderb	--	--	−.02	−.08	−.13	−.07	−.07	−.03	.01	.01	.07	−.02	.03	−.01	.00

Note. RFMS = Revised Firewater Myth Scale.

^aCorrelations were calculated using the square-root transformed variable, while the mean and standard deviation shown represent the untransformed variables.

^bGender was coded such that men = 0 and women = 1.

^*p < .05.

^**p < .01.

^***p < .001.

Alcohol use and problems

The inclusion criterion of having at least one standard drink in the prior month was effective in recruiting a diverse range of drinkers, including a significant proportion of potential problem drinkers (57.9% based on the AUDIT-C). Participants’ mean frequency of heavy drinking episodes was 4.35 (SD = 5.95) times during a typical month in the past year, with 77.6% reporting at least one such episode a month. Participants’ mean drinks per week was 9.20 (SD = 12.33). Means, standard deviations, and correlations are presented in Table 2.

Separate multiple regression analyses examining the association of the RFMS Biological Vulnerability subscale with alcohol use and consequences revealed small positive associations with drinks consumed per week (B = .29 [.12], β = .19, p = .016, f²= .04), frequency of heavy episodic drinking (B = .25 [.10], β = .21, p = .011, f²= .04), and a small-moderate positive association with alcohol problems (B = 3.59 [.88], β = .32, p < .001, f²= .11). In contrast, separate multiple regression analyses with the RFMS Alcohol Problems subscale revealed that believing that AI/AN people have greater alcohol problems was not significantly related to drinks per week (B = .05 [.12], β = .04, p = .642), frequency of heavy episodic drinking (B = .04 [.09], β = .04, p = .646), or alcohol problems (B = .79 [.86], β = .07, p = .363), with trivial effect sizes for each of these associations (f² ≤ .005).

Efforts to control drinking, self-efficacy, temptation, and expectancies

The Biological Vulnerability subscale was negatively associated with self-efficacy to resist drinking heavily (B = −.12 [.06], β = −.16, p = .046, f²= .03) and was positively associated with attempts to control drinking (B = 1.15 [.31], β = .29, p < .001, f²= .09) and temptation to drink heavily (B = .14 [.05], β = .21, p = .007, f²= .05). Biological Vulnerability had a small positive association with positive alcohol expectancies (B = .09 [.04], β = .19, p = .014, f²= .04) and a considerably larger moderate association with negative alcohol expectancies (B = .16 [.04], β = .35, p < .001, f²= .13).

Believing that AI/AN people have greater alcohol problems was not significantly related to self-efficacy (B = −.02 [.05], β = −.02, p = .753), attempts to control drinking (B = .33 [.30], β = .09, p = .281), temptation to drink heavily (B = .09 [.05], β = .14, p = .083), positive expectancies (B = −.02 [.03], β = −.05, p = .529), or negative expectancies (B = .001 [.04], β = .001, p = .989). Effect sizes for the associations between the Alcohol Problems subscale and these alcohol-related variables were trivial at f² < .003 for all but temptation, which showed a small effect size (f² = .02).

Models of alcoholism

The multiple regression analysis predicting Biological Vulnerability scores from beliefs about alcoholism revealed a unique positive association with belief in a disease model of alcoholism that was moderate in size (B = .47 [.09], β = .39, p < .001, f²= .18), but no unique association with belief in a psychosocial model of alcoholism (B = .18 [.13], β = .10, p = .157, f²= .01). In contrast, there was a small significant relationship between Alcohol Problems scores and belief in a psychosocial model of alcoholism (B = .38 [.14], β = .20 p = .010, f²= .04), but no significant association between Alcohol Problems and belief in a disease model of alcoholism (B = .15 [.10], β = .12, p = .132, f²= .015).

Guilt related to drinking

Only 14.6% of the sample reported experiencing any level of guilt associated with having one or two beers, glasses of wine, or wine coolers; while 35% of the sample reported any level of guilt for having one or two shots of hard liquor was (M = .61, SD = .97). The majority of the sample (61.6%) also reported some guilt for having three or four beers, glasses of wine, or wine coolers, although ratings associated with drinking this amount of this type of alcohol were low (M = .87, SD = .86). The majority of participants reported guilt for having 5 or more beers, glasses of wine, or wine coolers (71%; M = 1.71, SD = 1.17) or for having 5 or more shots of liquor (84.1%; M = 2.06, SD = 1.12).

Separate logistic regression analyses revealed that RFMS Biological Vulnerability (OR = 1.62, p = .048) and Alcohol Problems (OR = 1.76, p = .016) subscales were significantly associated with greater likelihood of expressing guilt over drinking one or two beers, glasses of wine, or wine coolers. Partial correlations revealed a significant positive relationship between guilt for from having one or two shots of liquor and Biological Vulnerability (r = .24, p = .002), but no significant association with Alcohol Problems (r = .09, p = .245). There also were no significant associations between the RFMS subscales and drinking greater amounts of alcohol, with correlation coefficients ranging from .01 to .13.

Local option law

Over a third of participants (35.9%) reported having lived in an Alaskan village or town that had voted to adopt a local option law that controlled the sale, importation, and/or possession of alcohol. Of these individuals, some reported having lived in more than one village or town with different local option laws. Of participants who reported having lived in a local option law village or town: 47.3% reported that sale, importation, and possession was prohibited; 40.0% reported the sale of alcohol was prohibited, but importation for personal use was allowed; 16.4% reported that sale and importation was prohibited, while possession was not; and 9.1% reported the sale of alcohol was limited to only one specific, licensed store.

Multiple regression analyses were conducted examining the effect of having lived in a local option law village on the RFMS Biological Vulnerability and Alcohol Problems subscales. Having lived in a village or town that had adopted any local option law was not significantly associated with Biological Vulnerability (B [SE] = .11 [.19] β = .05, p = .569, f² =.002) or with Alcohol Problems (B [SE] = −.09 [.20] β = −.04, p = .654, f² =.001). Analyses examining the effect of each local option law revealed that regardless of the local option law, there was no association with either RFMS subscale and effect sizes were trivial.

Discussion

One aim of this study was to develop a measure of AI/ANs’ belief in the FM that would allow us to examine initial hypotheses regarding the association of the FM with drinking behavior and problems. To this end, we revised the FMS developed by LaMarr (2003) to be more acceptable and less offensive to AI/AN participants. Our examination of the AI/AN-related questions of the revised FMS revealed one factor measuring belief in a biological and genetic vulnerability to alcohol problems among AI/AN people, which was the aim of the revised scale. Two other factors also emerged from the exploratory factor analysis. The second factor was simply comprised of two items relating to AI/AN people’s ability to “hold their liquor.” These items did not load onto the first factor despite being related to the notion of physiological differences in how alcohol affects AI/AN people. This factor was not significantly related to the other factors on the scale. It seems that these items primarily loaded together given the very similar item content, and in this sample were not a part of the FM. Given these findings, we recommend that these items, as well as item 10, which showed a low communality and factor loading, be omitted in future use. The final scale, with these items omitted, is included in Appendix A.

The third factor that emerged from the EFA was comprised of items relating to belief that AI/AN people do not drink more than non-Natives and do not have more problems with alcohol. These items were included in the scale as positively worded items to reduce the overall reactance to the scale and allow for a more accurate measurement of AI/AN participants’ belief in the FM. However, higher scores indicate disagreement and therefore greater endorsement of the idea that AI/AN people have more issues with alcohol than non-Native people. Not surprisingly, this factor showed a moderate positive association with belief in a biological or genetic propensity toward alcohol problems among AI/AN people.

On average, the sample tended to “slightly agree” with the idea that AI/AN people have a biological vulnerability to alcohol problems and have greater alcohol problems than non-Natives. Further, our finding that scores on both subscales were normally distributed shows that belief in the FM was not restricted to a minority of the sample.

Using the RFMS, we examined whether there was evidence of a potential self-fulfilling prophecy at play by examining the link between belief in the FM and alcohol use and consequences. Consistent with our hypotheses, greater belief in the biological vulnerability to alcohol problems aspect of the FM was associated with greater alcohol use, heavy episodic drinking, and alcohol consequences. We also examined potential mechanisms that may help to explain why belief in a biological vulnerability to alcohol problems would be related to alcohol use and consequences.

Consistent with a drinking restraint model and with the idea of stereotype threat, greater belief in a biological vulnerability to alcohol problems was associated with greater attempts to control drinking, guilt for drinking even small amounts of alcohol, temptation to drink heavily, and positive alcohol expectancies. Consistent with findings from a prior study using the original FMS (LaMarr, 2003) and lending support for the convergent and discriminant validity of the RFMS Biological Vulnerability subscale, greater endorsement of a biological vulnerability to alcohol problems for AI/AN people was associated with greater endorsement of a disease model of alcoholism, but was not associated with belief in a psychosocial model of alcoholism. Both the disease model and the FM, which is similar to an ethnic/racial specific version of the disease model, attribute alcohol misuse to factors that are internal, global, and stable (Walters, 2002). This may foster hopelessness regarding one’s ability to drink moderately or control one’s intake (Walters, 2002). Consistent with this notion, the Biological Vulnerability scores were associated with lower self-efficacy to resist drinking heavily. Belief in a biological vulnerability to alcohol problems also was associated with attempts to control drinking, but given its association with greater alcohol use and consequences, it may affect the strategies being used. Further studies are needed to examine the association between belief in a biological vulnerability to alcohol problems and efforts taken to control alcohol use and avoid alcohol-related problems among AI/ANs who drink. Such research could evaluate whether belief in a biological vulnerability to alcohol problems results in less use of protective behavioral strategies or harm reduction techniques because they are deemed ineffective for a disease-based attribution for alcohol problems, and if instead, greater attempts to establish abstinence are used as a means of avoiding alcohol-related harm.

In contrast to the findings for RFMS Biological Vulnerability subscale, greater belief that AI/AN people have more alcohol problems and drink more than non-Native people was associated with belief in a psychosocial model of alcoholism, which focuses on the role of the environment and social learning. This same RFMS Alcohol Problems subscale was not associated with greater drinking or alcohol problems, nor was it associated with self-efficacy, temptation to drink, or alcohol expectancies. Thus, in this sample of AI/AN college students who drink, believing that AI/AN people drink more and have more alcohol problems did not have a significant association with drinking or attitudes toward alcohol, with one exception: it was associated with greater guilt for drinking a small amount of alcohol (i.e., one or two drinks). The guilt associated with drinking just one or two drinks may be a reflection of all-or-nothing thinking regarding alcohol among FM believers; either one is completely abstinent or is drinking problematically.

LaMarr (2003) was the first to directly measure and examine belief in the FM, an important step in critically evaluating this stereotype that anecdotally is well accepted by AI/AN and non-Native people alike. However, this is the first study to our knowledge to directly examine the associations of the FM with alcohol-related behavior. More research is needed to carefully examine the nuances in the relationships between FM beliefs and alcohol consequences and to tease apart the mechanisms of action. Although we theorize that potential mechanisms linking the biological vulnerability aspect of the FM with greater alcohol use and problems are consistent with a drinking restraint model, as well as the effect that the FM may have on the use of strategies to moderate drinking, further research is needed to draw causal conclusions.

Another possible explanation for the pattern of associations found is that belief in the biological vulnerability aspect of the FM is a post hoc attribution for one’s difficulty moderating alcohol intake or for alcohol consequences. As an AI/AN individual experiences problems with alcohol, these may be attributed to their biological vulnerability and taken as confirmatory evidence, resulting in these beliefs increasing as alcohol problems develop. If this were the case, we would likewise predict that experiencing greater problems with alcohol oneself would lead to greater belief that AI/AN people experience greater problems with alcohol, owing to the assumed biological vulnerability for AI/AN people in general. However, no associations were found between personal alcohol use and consequences and scores on the Alcohol Problems subscale. Belief that AI/AN people have more alcohol problems also was not associated with efforts to control drinking, alcohol expectancies, or self-efficacy to resist drinking heavily or temptation to do so. Thus, believing that AI/AN people have more alcohol problems did not appear to lead to a self-fulling prophecy.

Contrary to our hypothesis, having previously lived in a village or town that had voted to restrict the sale, importation, and/or possession of alcohol was not associated with any aspect of belief in the FM. Our hypothesis was based on the notion that having lived in such a village or town would reinforce the FM. This lack of association between having lived in village or town that restricted access to alcohol and the FM might be owing to the FM being a prevalent stereotype that is not any more reinforced in such villages or towns than in locations that have not adopted such laws.

This study was limited to AI/AN college students in Alaska who drink. It is unknown if our findings would generalize to a non-college student sample. Another important limitation of this study was the cross-sectional design. Future studies are needed to examine the issue of directionality. Future studies also are needed to examine how belief in the biological vulnerability aspect of the FM may affect efforts to control drinking, the use of harm reduction techniques, and use of protective behavioral strategies to avoid negative alcohol consequences among AI/AN students who drink. If belief in the biological vulnerability aspect of the FM causes greater drinking, alcohol problems, and lower self-efficacy to control drinking, then prevention and treatment efforts with AI/AN college students should focus on challenging beliefs in the FM – particularly given the absence of evidence that AI/AN people are genetically or biologically more vulnerable to alcohol problems than are other ethnic or racial groups. Even if it the case that belief in the biological vulnerability aspect of the FM is formed only after alcohol problems develop, challenging this belief may be beneficial as it could undermine intervention efforts and self-efficacy for change. Given that the FM is a particular type of alcohol expectancy, expectancy challenge interventions (e.g., Lau-Barraco & Dunn, 2008) could potentially be useful for correcting beliefs in the FM.

In conclusion, this study is the first to our knowledge to directly examine belief in the FM and alcohol outcomes. Although belief the biological vulnerability aspect of the FM was associated with attempts to control drinking, our findings suggest that these beliefs may have harmful effects on alcohol use and consequences among AI/ANs who drink, leading to a self-fulfilling prophecy that may in turn strengthen belief in the myth. It is possible that belief in the biological vulnerability aspect of the FM may undermine self-efficacy to drink moderately or use other strategies to avoid alcohol-related consequences among those who drink. While further research is still needed to establish a causal relationship between belief in the biological vulnerability aspect of the FM and negative alcohol outcomes, interventions aimed at AI/AN college drinkers may benefit from educational components regarding the lack of evidence for the FM and how belief in this myth may inadvertently affect alcohol use, consequences, and self-efficacy for change.

Appendix A

Revised Firewater Myth Scale

In this measure we’re interested in how people think about alcohol and the degree to which they believe there are differences in how people react to alcohol. Your responses are confidential and will never be tied to your identity. This measure asks you to rate how much you agree with the following statements based on what you think, not how sure you are that the statements are 100% true or proven.

1. People of Asian descent feel the effects of   alcohol, or feel intoxicated, more easily than   people of European descent.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
2. In general, African Americans drink more   than European Americans.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
3. Alcohol causes more harm than good for   people in general, regardless of ethnicity or   race.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
4. It is better for people who come from ethnic   or racial groups with high rates of alcoholism   to not drink at all.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
5. Drinking even small amounts of alcohol is   harmful for many Alaska Natives and   American Indians.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
6. People of Irish descent are proud of how well   they can hold their liquor.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
7. Alaska Natives and American Indians are   more likely to be born with a tendency to   develop problems with alcohol.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
8. People of Irish descent are more likely to start   a fight after drinking	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
9. French people drink frequently, but don’t   suffer from alcoholism to the same extent as   Americans.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
10. People of Irish descent can hold their liquor   better than non-Irish people.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
11. Alaska Natives and American Indians don’t   drink any more than any other ethnic or racial   group.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
12. African Americans are more likely to have   alcohol problems than are European   Americans.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
13. Most people of Irish descent who drink don't   have a problem with alcohol.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
14. Alaska Natives and American Indians are as   biologically capable of handling alcohol as   non-Native people.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
15. Because of biological or genetic differences,   it would be best for people of Irish descent to   never drink or stop drinking completely.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
16. Alaska Natives and American Indians feel the   effects of alcohol, or feel intoxicated, more   easily than people of European descent.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
17. Many people of Irish descent don’t drink at   all.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
18. Alaska Natives and American Indians have as   much control over their drinking as non-   Native people.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
19. Most Alaska Natives and American Indians   who drink don't have a problem with alcohol.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
20. People of Scottish descent are more likely to   have problems with alcohol.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
21. Alaska Natives and American Indians   metabolize alcohol differently than non-   Native people.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
22. Hispanics drink more heavily than do non-   Hispanics.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
23. European Americans are more likely to have   problems with alcohol than are Asian   Americans.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
24. Hispanics are more likely to be born with a   tendency to develop problems with alcohol.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
25. Because of biological or genetic differences,   it would be best for Alaska Natives and   American Indians to never drink or stop   drinking completely.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
26. Drinking even small amounts of alcohol is   harmful for many people of Irish descent.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
27. Hispanics drink less often than European   Americans, but drink more heavily when they   do drink.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
28. European Americans use alcohol more   heavily than many ethnic minority groups in   the US.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
29. Alaska Natives and American Indians are   more likely to have a genetic vulnerability to   problems with alcohol.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
30. Many African Americans don’t drink at all.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
31. People of Irish descent are more likely to   have a genetic vulnerability to problems with   alcohol.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree
32. Alaska Natives and American Indians have a   shorter history of exposure to alcohol   compared to people of European descent,   therefore are more susceptible to the effects   of alcohol.	1 Strongly Agree	2 Moderately Agree	3 Slightly Agree	4 Slightly Disagree	5 Moderately Disagree	6 Strongly Disagree