Solidarity or self-interest? Public opinion in relation to alcohol policies in Sweden.

AIM: The aim of this article is to study how people sometimes accept policies that could in a narrow sense be seen as in conflict with their own self-interest.
DESIGN: The study is based on survey data relating to public opinion on alcohol policy in Sweden targeted at people aged 16-85 years 2016-2017. Among the 3400 people questioned, the response rate was 52%.
RESULTS: The results show that people's perception of the problematic societal consequences of alcohol, in combination with ideological norms regarding the responsibility of individuals, is much more important in explaining public opinion than self-interest factors. It is the view that there is a problem at the societal level, rather than at the personal level, that is most essential for explaining opinions on alcohol restrictions. General knowledge of alcohol-related matters has some effect, whereas personal experiences of close affiliates excessive drinking does not seem to color the opinions expressed.
CONCLUSION: Support for restrictive alcohol policies in Swedish public opinion is mainly founded on norms of solidarity and astute problem analyses at the societal level, and to a much lesser extent on egoism and personal experiences.

Why do people accept restrictive alcohol policies?

To implement policies that regulate consumption of alcohol is a challenge. For many people, alcohol is associated with good things in their private life: socialising with friends, festive celebrations, gastronomic delights, romantic adventures, overcoming inhibitions, to cope with stress of work and daily life and escaping from otherwise hard circumstances (Fry, 2011; O’Malley & Valverde, 2004). For the state to impose restrictions affecting all alcohol consumers is therefore a considerable invasion into the private sphere. Furthermore, in democratic societies, where freedom of individuals is treasured, the enforcement of such invasive policies places major demands on their legitimacy. Where popular support for public health policy in relation to alcohol is sceptical or at best equivocal, it is unlikely that such policies will ever be successful in spite of evidence-based arguments (Butler, 2009). Yet some countries succeed not just in implementing such restrictive policies, but in doing so with widespread and continuous public support.

Knowledge of how public opinion is formed in relation to restrictive social policies – and why people sometimes accept policies that in a narrow sense are in conflict with their own self-interest – is of great importance in understanding what makes such policies successful. Are there ideological beliefs that might trump self-interest – and if so which beliefs? How important is people’s understanding of the social problem at hand and to what degree is personal experience a factor? Does increased factual knowledge of the area at issue matter? The aim of this article is to seek answer to these questions and thereby increase our knowledge on what lies behind public support for restrictive policies. The study will focus on public opinion on alcohol policies in Sweden.

Public restrictions on alcohol consumption vary considerably around the world, and Sweden is an example of a country with a more restrictive approach (Brand, Saisana, Rynn, Pennoni, & Lowenfels, 2007; Karlsson & Österberg, 2007; Madureira-Lima & Galea, 2018). At the same time, as we will see, there is strong public support for this restrictive line (Leimar, Ramstedt, & Weibull, 2013; Weibull, Holmberg, Karlsson, & Arkhede, 2017), and the tendency in recent decades is towards more restrictive and less liberal positions. However, as with most political issues, public opinion is not unified – there is a substantial liberal minority. Analysing this great variation of opinions regarding alcohol-related issues gives us a greater understanding of which factors cultivate support for restrictive policies and which factors draw people towards support for liberalisation.

Possible explanations of variations in public opinion in relation to alcohol restrictions

Previous studies of attitudes towards alcohol policies have identified numerous factors that affect public opinion (see, for example, Elmeland & Villumsen, 2013; Greenfield et al., 2014; Li et al., 2017; Lucchetti, Koenig, Pinsky, Laranjeira, & Vallada, 2014; Moskalewicz, Wieczorek, Karlsson, & Österberg, 2013; Rossow & Storvoll, 2014; Stanesby, Rankin, & Callinan, 2017; Storvoll, Moan, & Rise, 2015; Storvoll, Rossow, & Rise, 2014; van der Sar et al., 2012). The results suggest that women, the elderly and religious people hold more restrictive views, while people who drink more alcohol are more liberal; experience of harm caused by the drinking of others suggests is of importance to support for restrictive policies; more intrusive policies are less popular; belief in the effectiveness of measures as well as belief in harm caused by drinking is of importance.

However, for the most part, these studies have each focused on a limited set of explanatory factors. Our ambition is to take a more comprehensive approach, testing established results and adding new perspectives from the wider field of public opinion research. The analysis will be based around six hypotheses on the importance of self-interest, public interest/ideology, knowledge, experience and problem perception. In the following sections, we will present the theoretical rationale for these hypotheses, starting with the fundamental question on the role of self-interest.

Narrow self-interest

The challenge of gaining public acceptance for invasive restrictions is not unique for alcohol policy. In fact, the core of most social policies is often to attend to the needs of the few by asking for sacrifices from the many, to support and protect less privileged groups by redistributing resources and providing tax-funded public services. The aim is commonly to minimise risks for vulnerable groups by placing restrictions on the behaviour of all people. And it is certainly not a given that people are prepared to endure costs and inconveniences in order to help strangers, especially when the sacrifices involved are burdensome. And even worse, invasive policies that are perceived as illegitimate risk undermining trust in the political system as a whole. For shorter periods of time, a restriction may be introduced in opposition to public opinion, but for such policies to succeed in the long run requires the continuous support of the majority.

Alcohol policy is a good case for illustrating this challenging problem. On the one hand, alcohol consumption is a major – but avoidable – risk for public health (Rehm et al., 2009), and there is plenty of evidence suggesting that restrictions on price and availability are effective in limiting consumption and preventing harm (Anderson, Chisholm, & Fuhr, 2009; Babor, 2010; Burton et al., 2017). The perceived downside, on the other hand, is the limiting effect that such policies have on the individual freedom of alcohol consumers.

Much social science has traditionally tended to ascribe self-interest as one of the prime sources of people’s behaviour and opinions (Downs, 1957). Even support for welfare policies that show solidarity might sometimes be described as based on self-interest and a principle of mutual insurance rather than social altruism (Rodger, 2003). However, it is also widely recognised that self-interest is multifaceted and varieties of altruistic or collective interests are equally important as drivers of people’s motivations (Mansbridge, 1990; Sears & Funk, 1991). Self-interest in these discussions is usually understood in a narrow sense, as in what is best for me in the here and now rather than recognising that self and public interest may coincide in the more distant future.

The narrow self-interest approach to alcohol policies would be to ask: Which policies are most beneficial for me – for my private financial situation and my personal convenience? In general, people are more likely to accept alcohol policies that are less intrusive and that do not affect their lives (Li et al., 2017). Restrictive alcohol polices, such as higher prices through taxation and limited availability through market restrictions, would in that respect be in conflict with the narrow self-interest of alcohol consumers. Furthermore, people’s self-interest is therefore tied to their alcohol consumption, as frequent consumers are more affected by higher prices and reduced availability than moderate consumers (Diepeveen, Ling, Suhrcke, Roland, & Marteau, 2013). Previous studies have repeatedly confirmed that there is a positive correlation between alcohol consumption and a critical view of restrictive policies (Giesbrecht, Ialomiteanu, & Anglin, 2004; Hemström, 2002; Macdonald, Stockwell, & Luo, 2011; van der Sar et al., 2012), which could be interpreted as a result of self-interest considerations.

Additionally, taking into account the public costs resulting from the adverse consequences of alcohol consumption in society, it would be in the narrow self-interest of abstainers and moderate drinkers to push such costs on to the people causing the problems by increased taxation of alcohol. Conversely, the narrow self-interest of frequent consumers of alcohol would be to ensure that such costs are shared by all and not payed for by sin taxes. Taking all of this into consideration, a theory based on self-interest would predict:

H1. (Narrow self-interest) Support for restrictive alcohol policies is negatively correlated with people’s alcohol consumption.

However, even if self-interest is a dominant candidate for explaining people’s attitudes, it is normally suggested that public opinion can also be based on public interest concerns – or combinations of the two (Funk, 2000; Lewin, 1991).

Ideology

People’s perceptions of what is in their own and in the public interest sometimes conveniently coincide, and it is difficult to determine whether people are driven by concerns for their own well-being or for the public good. But other times people favour policies that are recognisably in conflict with their narrow self-interest – such as when alcohol consumers support restrictive alcohol policies. In such cases, we must go beyond the narrow self-interest approach and consider people’s ideological positions.

When studying policy proposals that restrict the availability of alcohol, the most central ideological concern is the responsibility of the state in relation to alcohol consumption and its effects on society. This question is closely linked to the political left–right dimension and the overall responsibility of the state in relation to citizens (Karlsson, 2012; Room, 2011). Someone who is sceptical of the idea of an active state which interferes in people’s lives would probably also be less inclined to view the state as responsible for individuals’ alcohol consumption, and thereby be less likely to accept policy proposals restricting the availability of alcohol. On the other hand, people who have come to the conclusion that the government is responsible and should do more to tackle the harm done by alcohol also tend to support restrictive policies (Li et al., 2017). Previous studies have suggested that party affiliation along left–right lines is strongly correlated with attitudes on alcohol restrictions (Leimar et al., 2013). We would therefore expect:

H2. (Left–right ideology) People on the political right are more negative towards restrictive alcohol policies than people on the left.

However, there are also indications that people make a distinction between a general stance on the state–individual relationship, as in the left–right dimension, and a more specific or nuanced position in the case of alcohol. Vice control divides between liberals and restrictivists on both the right and the left (Kleiman, 1987; Widerquist, 2009). The fact that attitudes towards alcohol policies are related to, but partly detached from, left–right ideology is also evident from earlier studies on citizens and political representatives, with one obvious example being that Christian Democrats – normally to the political right – are proven to be strongly in favour of restrictive alcohol policies (Karlsson, 2012; Macdonald, Listhaug, & Rabinowitz, 1991; Rise & Halkjelsvik, 2016). We therefore propose:

H3. (Alcohol liberalism) Under control for left–right positions, people’s positions regarding a separate ideological dimension concerning individual versus state responsibility for alcohol consumption affect people’s attitudes towards alcohol policies.

Experiences, knowledge and problem perception

Another starting point for studying public opinion on specific policy proposals is people’s understanding of the problems which the policies are designed to address, as such proposals normally originate as solutions to a perceived problem. Irrespective of whether opinions on policy proposals are influenced by self-interest or more principled concerns for the public interest, the question of what actually is in someone’s interest depends on how they perceive the problem at hand. It is therefore essential to ask whether people believe that there actually is a problem, and if so, what this problem consists of, what causes it, and what consequences it has (Bacchi, 2009). Perceptions of what the problems associated with alcohol are, and how they can be addressed, have changed considerably over time (Nicholls & Greenaway, 2015), and whether the problem has been mainly perceived as an individual and/or medical issue rather than a social issue has influenced the debate and which policies are proposed (Edman & Olsson, 2014).

One possible foundation for people’s understanding of alcohol-related problems is their own knowledge of, and personal experience of, alcohol consumption and problematic use of alcohol (Laslett, Room, Waleewong, Stanesby, & Callinan, 2019). It is quite likely that personal encounters with the darker side of alcohol consumption influence people’s judgement. People who have close personal ties to a heavy drinker – or who have otherwise faced harm from the drinking behaviour of others – report that they have been affected very negatively by such experiences (Greenfield, Karriker-Jaffe, Kerr, Ye, & Kaplan, 2016; Ramstedt et al., 2016). Some results also suggest that people who have experienced harm from the drinking of others are more likely to support restrictive alcohol policies (Greenfield et al., 2014).

People’s factual knowledge of the consequences of alcohol also varies. For example, many people tend to significantly underestimate the risks of alcohol consumption (Rehm, Lachenmeier, & Room, 2014). It is also a challenge to change the minds of people who are reluctant to accept evidence of the serious harm associated with use of alcohol and who are sceptical about evidence-based policies designed to reduce this harm (Stockwell, 2013). However, people who are well informed about the social calamities caused by alcohol consumption, who perceive the problems of alcohol as real and urgent, and who have experiences of others who drink too much in their private lives, are also probably likely to support the need for restrictive alcohol policies. This gives rise to three expectations on the effects of people’s attitudes to restrictive alcohol policies based on their own experiences, factual knowledge and perceptions of the problem – both in their own personal lives (micro level) and in society (macro level):

H4. (Personal experience of harm) People who have experienced that they themselves or someone close to them drinks to excess are more prone to support restrictive alcohol policies.

H5. (Knowledge) People who have accurate knowledge regarding alcohol consumption in society are more supportive of restrictive alcohol policies.

H6. (Problem perception) People who on the whole perceive the consequences of alcohol consumption – in their personal lives as well as for society in general – to be more negative than positive are more supportive of restrictive alcohol policies.

Finally, factors relating to people’s social background such as age, gender, social class, geographic position, and religiosity, must also be taken into account. Such factors may have a direct effect on people’s attitudes but they are also important for understanding variations in ideological positions, self-interest (i.e., alcohol consumption) and knowledge about the effects of alcohol consumption.

Methods

The survey

The study is based on the Society, Opinion, Media (SOM) survey. The survey has been conducted yearly since 1986 by a research institute at University of Gothenburg with a representative sample of the Swedish population, mainly via a mail-in questionnaire, and is targeted at people aged 16–85 years. For a number of years, questions on alcohol policy have been included in the SOM surveys (Weibull et al., 2017). Data from a survey conducted September 2016–January 2017 will be used in this article. Among the 3400 people questioned, the response rate was 52%. Respondent characteristics reflect the Swedish population in terms of gender, social class and level of education. However, older people are somewhat over-represented as the response rate in the youngest groups is below average. For more details on the research design and response rates, see Arkhede, Bové, Bové, and Jansson (2017).

Alcohol policy and opinion in Sweden – the dependent variable

Swedish alcohol policy has a long tradition of being restrictive and having the aim of limiting consumption. Policy has been focused on regulating prices and availability, and restricting commercial interests (Holder et al., 1998). Even though the prohibitive line has not been officially changed, Swedish membership of the European Union in 1995 effectively reduced the range of policy instruments available to the state, the public monopoly of alcohol production and import limitations (Örnberg, 2009). In some ways, the responsibilities of local government have expanded as national instruments have been reduced (Karlsson, 2012; Nilsson, 2019). However, some policy instruments remain at national level, most notably taxation and the monopoly on distribution through the state-owned company, Systembolaget.

During the 2000s, public opinion on these policy instruments has been measured annually by the SOM surveys in the form of attitudes towards specific policy proposals. A number of such proposals have been included over the years (Weibull et al., 2017), with the longest time series concerning the two most important policy instruments: “Allowing strong beer, wine and spirits to be sold in grocery stores” (and thereby abolishing the national distribution monopoly) included 2001–2018; “lowering taxes on alcohol” 2005–2018 and “raising taxes on alcohol” 2010–2018. The responses were given on a five-point scale from very bad to very good proposal. The results from these questions are presented in Figure 1.

Figure 1.

Public opinion in relation to policy proposals concerning alcohol in Sweden 2001–2018 (percentage “very good” or “good” proposal”).

Source: National Society, Opinion, Media surveys 2001–2018.

Note. Some years the questions were part of a subset of the SOM survey and some years they were included in all questionnaires. The number of respondents for the monopoly question in 2001 was 3461, 3417 for the lower tax question in 2004, and between 3001–3334 for the raise taxes question in 2011, 2013 and 2014. Other years the number of respondents for all questions varied between 1513–1751.

In the late 1990s, following Sweden’s EU membership, Swedish alcohol policy took a liberal turn, with some support in public opinion. In the early 2000s, support for abolishing the monopoly and lowering taxes on alcohol were majority opinions. However, the SOM results for the following years showed a clear trend in the restrictive direction (cf. Rossow & Storvoll, 2014), even though during the last few years this trend seems to be levelling out. Twice as many Swedes in 2018 supported raising taxes on alcohol as there were people who supported lowering taxes. The monopoly on distribution has strong public support (see Figure 1).

In order to find an appropriate dependent variable measuring public support for restrictive policies, we have constructed an index of the mean value of two questions regarding the most important political instruments – one affecting consumers’ self-interests in terms of economic cost and the other reducing their convenience: “lowering taxes on alcohol” and “allowing strong beer, wine and spirits to be sold in grocery stores” (i.e., abolishing Systembolaget’s retail monopoly). The correlation between the two variables is Pearson’s r = +.50. The index is coded 0 (lowering taxes and abolishing the monopoly are very good proposals) to 100 (both very bad proposals). A total of 1644 respondents have answered either or both questions, and the index mean for all respondents is 65, with a standard deviation of 28. These results all refer to the SOM survey of 2016/2017, which will be used in the analyses below. The reason for focusing on just this year is the inclusion of the questions used as key independent variables presented below.

Factors explaining opinion in relation to alcohol – the independent variables

The narrow self-interest hypothesis (H1) predicts that support for restrictive alcohol policies is negatively correlated with people’s alcohol consumption. Our indicator for consumption is the survey question, “During the previous 12 months, how often have you consumed strong beer, wine or spirits?” with the response alternatives “never”, “once”, “twice a year”, “once every quarter of a year”, “once a month”, “once a week”, “several times a week”. We will use this question as two variables, the first being a scale variable recoded into times per year (0, 1, 2, 4, 12, 52, 100) and one categorical “abstainer” variable (1 = have consumed alcohol, 0 = have not). The results show that 12% of respondents abstained from alcohol during the previous year. The combined variables will illustrate whether the main effect is related to the abstention and consumption of alcohol, or whether there is a gradual effect related to the actual frequency. For H1 to be supported, we would expect a significant negative effect of the abstainer indicator and a positive effect of the frequency indicator.

H2 and H3 relate to the effects of ideology and consideration of public interest. The indicator for left–right ideology is the survey question: “People sometimes talk about locating political opinions on a left–right scale. Where would you place yourself on such a scale?”, with the responses given on a five-point scale from 1 “clearly to the left” to 5 “clearly to the right”. The mean value for all respondents was 3.1 with a standard variation of 1.1. For H2 to be supported, we would expect a negative effect of this variable (i.e., people to the right are less supportive of restrictive policies). Our indicator for an ideological dimension specific to alcohol policy is the question: “In political debates it is discussed where the main responsibility for the consumption of alcohol lies – with the individual or with society and the authorities. What is your opinion?”. The responses were given on a seven-point scale from 1 “there must be clear public restrictions on people’s alcohol consumption” to 7 “there is no need for public restrictions as individuals must take responsibility for their own alcohol consumption”. The mean value for all respondents was 3.8 with a standard variation of 2.3. For H3 to be supported, we would expect a negative effect of this variable (i.e., people who emphasise the responsibility of the individual would be less inclined to support restrictive policies).

The H4 hypothesis relates to people’s experiences of someone close to them drinking too much. The indicator here is based on the survey question: “In your opinion, are there one or more people in your close social environment who drink too much?”, with the response alternatives “no”, “yes, myself”, “yes, a family member”, “yes a close friend”, and “I don’t know”. We will use a categorical version of this variable: 0 = “no, or I don’t know”, or 1 “yes someone close to me drinks too much”. About 37% of respondents have had such experiences. For H4 to be supported, we would expect a positive effect of this variable (i.e., such experiences promote support for restrictive policies).

H5 relates to the importance of accurate knowledge. The indicator for knowledge is an index based on respondents’ opinions on three statements, two false (“alcohol consumption is a greater problem in Sweden than in other EU countries”; “a moderate consumption of wine is good for your health”) and one true (“alcohol consumption among Swedish young people is decreasing”). The responses were given on a scale from 1 “completely incorrect statement” to 5 “completely correct statement”, as well as a “no opinion” alternative. It is always a challenge to include factual questions in a survey where respondents might have the opportunity to look up the correct answer. We indicated in the questionnaire that the three statements concerned issues where people’s opinions differ, in order to deflect the respondents from realising that we were trying to test their factual knowledge. The index values represent a mean value of the three responses where 0 represents a completely wrong answer to all three and 1 a completely right answer to all three. For those who respond “no opinion” to some statements, the index is based on the statements where they have an opinion. The mean value for this index for all respondents was 0.54 with a standard deviation of 0.22. For H5 to be supported we would expect a significant positive effect of this variable (i.e., people with accurate knowledge are more likely to support restrictive alcohol policies).

H6 relates to how people perceive the consequences of alcohol. Our indicators here are two variables related to the same survey question: “Alcohol can have both positive and negative consequences. On the whole, do you find that the positive or negative consequences of alcohol consumption predominate?” The responses were given on a seven-point scale from –3 “negative consequences predominate significantly” to +3 “positive consequences predominate significantly” with 0 being the neutral answer. The respondents were asked to differentiate their answers for (a) “for you personally” and (b) “for society”. The results show that 76% perceived consequences as neutral or positive on the personal level and 74% perceived them to be negative on the societal level. For H6 to be supported, we would expect negative effects for both these variables (i.e., people who perceive that the positive effects of alcohol transcend the negative effects, on a personal as well as on a societal level, will be less supportive of restrictive policies).

A number of social characteristics, such as gender, age, income, education, social class, urban/rural location, nationality and religiosity, are also included in the analysis as control variables. All independent variables are presented in Table 1.

Table 1.

Independent variables: descriptive data (percentages and mean values) and bivariate correlation (b-values and Adj. R 2) with opinion on restrictive alcohol policies.

	Percentage of respondents (dummy variables)	Mean value (scale variables)	Correlation b-value (bivariate OLS regression)	Correlation Adj R 2 (bivariate OLS regression)	N (of independent and dependent combined)
Social characteristics
Gender: female = 1 (Control group: male and others)	47.4		+8***	.017	1.640
Age: young (< 30 years) = 1	15.6		–1	.000	1.640
Age: old (> 65) = 1	31.1		–1	.000	1.640
(Control group: middle-aged 31–64 years)
Household income: high = 1	27.4		+3	.000	1.640
Household income: low = 1	24.1		–4*	.003	1.640
(Control group: mid-level income)
Education: high = 1 (Control group = low)	41.2		+8***	.020	1.640
“Class”: blue-collar home = 1	36.2		–3	.001	1.640
“Class”: business owner/farmer = 1	10.4		–6**	.004	1.640
(Control group: white collar home = 1)	44.9		+1	.000	1.640
Geography: village, rural area = 1	32.5		-0	.000	1.640
Geography: city = 1	17.3		+2	.000	1.640
(Control group: towns)
Nationality: grown up abroad = 1 (Control group: grown up in Sweden)	8.5		–6*	.003	1.640
Religiosity: prays to God. Scale 0 (never)–1 (several times per week)		0.15	+10***	.013	1.588
Alcohol consumption
Alcohol consumption frequency: scale 0 (never)–1 (several times per week)		0.30	–16***	.030	1.601
Alcohol consumption: abstainer = 1 (control group = consume alcohol)	11.6		+16***	.038	1.640
Knowledge, perception, experience
Own experience of someone close drinking too much = 1 (Control group: no experience)	36.6		+4*	.004	1.621
Knowledge about alcohol: index scale 0 (low)–1 (high)		0.54	+15***	.011	1.475
Problem perception of alcohol consumption for you personally: 0 (negative consequences dominate)–1 (positive consequences dominate)		0.55	–25***	.063	1.363
Problem perception of alcohol consumption in society: 0 (negative consequences dominate)–1 (positive consequences dominate)		0.26	–40***	.104	1.411
Ideology
Left–right ideology: subjective scale 0 (left)–1 (right)		0.52	–21***	.044	1.598
Alcohol liberalism: scale 0 (restrictive)–1 (liberal)		0.40	–40***	.226	1.481

Source: National Society, Opinion, Media survey, 2016.

Note. OLS = ordinary least squares. All independent variables are coded 0–1, either as categorical variables or as scale variables, with 1 representing the highest possible value of each variable and 0 the lowest. The table includes the percentage of each category and mean value of each scale variable, as well as the bivariate correlation (b-value and adjusted R 2) for each independent variable on the dependent variable (the 0–100 index of support for restrictive policies). The N-value represents number of respondents for the correlation value, i.e., respondents who have values on both the independent and the dependent variables. The control group in sets of dummy variables is described in italics.

*p < .05. ** p < .01. *** p < .001.

Model design

The most important test of the six hypotheses will be made in a multiple ordinary least squares (OLS) regression model which includes the direct effects of all independent variables. However, we also recognise that the independent variables have intricate internal relationships and their effects on the dependent variable may be indirect as well as direct. As our data material consists of a cross-section survey study, we will not be able to formally determine the causal relations between the variables included. However, we will adhere to the following model strategy based on reasonable presumptions of causality between categories of variables in order to distinguish indirect as well as direct effects of the independent variables.

In our first model (Table 2 M1), alcohol consumption frequency is the dependent variable and social characteristics of respondents are independent. The results indicate that all social characteristics have significant effects on consumption: people who are male, older, richer, highly educated, not blue-collar, urban, who have grown up in Sweden and are not religious drink alcohol more frequently. This also means that effects of alcohol consumption on other variables are indirectly affected by all these social factors. The explanatory power of M1 is Adj. R2 = .11.

Table 2.

How factors explaining public opinion on alcohol policies relate to one another, OLS regression models.

Dependent variables	M1Alcohol consumption frequency	M2Knowledge index	M3Problem perception – personal	M4Problem perception – in society	M5Left–right ideology	M6Alcoholliberalism
Social characteristics
Gender: female = 1	–6***	–3**	+0	–5***	–4*	–2
Age: young (< 30 years) = 1	–7***	+1	+8***	+4*	–2	+1
Age: old (> 65) = 1	+7***	+1	+1	–4*	+1	+6*
Household income: high = 1	+7***	+1	+2	+0	+5*	–1
Household income: low = 1	–6***	–2	–2	–2	+0	+2
Education: high = 1	+8***	+3*	–1	–2	–2	–6**
“Class”: blue-collar home = 1	–4***	–05**	–4*	+0	–8***	+4
“Class”: business owner/farmer = 1	+2	–4*	+2	+1	+10**	+4
Geography: village, rural area =1	–3***	+2	–1	+0	–3	–0
Geography: city =1	+7***	+1	–2	–1	–5*	+3
Nationality: grown up abroad = 1	–9***	–9***	–4	+7**	–8**	+14***
Religiosity: prays to God, scale 0 (never)–1 (several times per week)	–9***	–1	–4	–2	+9**	–7*
Alcohol consumption
Alcohol consumption frequency: scale 0 (never)–1 (several times per week)		–1	+19***	+9***	–2	+4
Alcohol consumption: abstainer = 1		+1	–29***	+2	+2	–3
Knowledge, perception, experience
Own experience of someone close drinking too much = 1		–0	–2	–2	–1	–2
Knowledge about alcohol: index scale 0 (low)–1 (high)			–6	+2	–7	+12*
Problem perception of alcohol consumption – personal: scale 0–1				+36***	–1	+3
Problem perception of alcohol consumption – in society: scale 0–1					+13**	+30***
Ideology
Left–right ideology: scale 0 (left)–1 (right)						+24***
Constant	31***	57***	58***	07*	58***	25***
Adj R 2	.11	.04	.23	.26	.07	.15
N	7.787	1.429	1.224	1.192	1.167	1.126

Source: National Society, Opinion, Media (SOM) survey, 2016.

Note. OLS = ordinary least squares. The six dependent variables are survey questions (and one index) presented in Table 1. All dependent variables are coded 0 (lowest value) to 100 (highest value) and all independent variables are coded 0–1. The number of respondents is much higher in M1 since these questions were included in all versions of the SOM questionnaires, including those where alcohol-related questions were not included. Adj R 2-values are written in italics.

*p < .05. ** p < .01. *** p < .001.

M2 relies on the assumption that people’s knowledge of alcohol issues is related to their social background and experience of alcohol consumption as well as to experiences of people close to them (including themselves) drinking too much. The explanatory power of this model is much weaker (Adj. R 2 = .04) and we learn that people who are male, highly educated, white-collar and born in Sweden are the most knowledgeable. Experiences of alcohol consumption – own and others – do not matter.

We further assume that how people perceive the effects of alcohol consumption in their personal lives is based on social background, experience and knowledge (M3). This personal problem perception is then included as an independent variable in M4, as people are likely to let their personal experiences influence their perceptions of problems on a societal level rather than vice versa. The explanatory power of these models is quite strong (Adj R 2 = .23 for M3 and .26 for M4). Alcohol consumption is, above all, the main factor affecting personal problem perception: people who drink often perceive the effects of alcohol on their personal lives as more positive, and the effect increases with consumption frequency. People who are younger and white-collar also have a more positive perception at the personal level. Surprisingly, experience of someone close to them drinking too much has no effect.

Problem perception at the societal level (M4) is influenced by private experiences and thereby indirectly by personal alcohol consumption. The direct effect of alcohol consumption is lower. We also note that people who are female, older and raised in Sweden are more likely to perceive the negative consequences of alcohol in society as more dominant.

In M5 and M6, ideological beliefs are dependent variables as it is reasonable to assume that political beliefs are products of social background as well as experiences and problem perception. In M5 we note that people who are male, highly educated, white-collar or a business owner, urban, who have grown up in Sweden and who are religious are more likely to be to the political right. These results are not surprising to anyone who is familiar with public opinion research. We also note that people who perceive the societal effects of alcohol as more negative are more likely to be to the political left. This aligns with the assumption that people to the left are more inclined to highlight structural social problems. People’s experience of alcohol consumption has no correlations with their left–right position.

In M6, left–right position is included among the independent variables explaining people’s position in relation to the alcohol-specific ideological dimension: alcohol liberalism. As expected, we note that people to the political right tend to emphasise the responsibility of the individual rather than public responsibility in relation to the consequences of alcohol consumption. The two ideological dimensions are therefore clearly related. On the other hand, this correlation could also be described as surprisingly weak. Along with other independent variables – which also contribute significantly to the model – the explanatory power of M6 is only .15. This means that alcohol liberalism is also very much an ideological dimension apart from left and right. The strongest effect on this variable, stronger than that of left–right position, is the perception of a problem at the level of society. People who perceive societal problems to be greater are more inclined to emphasise social responsibility, and so are people who are older, have a lower level of education, have grown up abroad and have a higher level of knowledge about alcohol. Remarkably, alcohol consumption and perception of a problem at the personal level have no significant effect on alcohol liberalism.

In the next section, we will test our hypotheses by introducing opinions in relation to restrictive alcohol policies as the dependent variable in Table 3. In the first four models of this analysis, groups of independent variables are introduced separately in order to gauge their respective explanatory power. In M5, social characteristics are included along with alcohol consumption. In M6, variables relating to knowledge, perception and experience are introduced and in M7 the two ideological variables are added.

Table 3.

Factors explaining public opinion in relation to restrictive alcohol policy, OLS regression models.

	M1	M2	M3	M4	M5	M6	M7
Social characteristics
Gender: female = 1	+7***				+6***	+6***	+5**
Age: young (< 30 years) = 1	–1				–2	–2	2
Age: old (> 65) = 1	+1				+1	+0	+2
Household income: high = 1	+2				+3	+4*	+5**
Household income: low = 1	–2				–3	–3	–4
Education: high = 1	+7***				+9***	+7***	+5**
“Class”: blue-collar home = 1	–1				–2	–2	–2
“Class”: business owner/farmer = 1	–8**				–7**	–8**	–4
Geography: village, rural area =1	+1				+0	+1	–0
Geography: city =1	+0				+1	+1	–1
Nationality: grown up abroad = 1	–7**				–11***	–7*	–4
Religiosity: prays to God, scale 0 (never)–1 (several times per week)	+10***				+7**	+5*	+4*
Alcohol consumption
Alcohol consumption frequency: scale 0 (never)–1 (several times per week)		–11***			–15***	–8**	–8**
Alcohol consumption: abstainer = 1		+12***			+13***	+7**	+6*
Knowledge, perception, experience
Own experience of someone close drinking too much = 1			+2			+2	+1
Knowledge about alcohol: index scale 0 (low)–1 (high)			+18***			+15***	+10*
Problem perception of alcohol consumption – in private life: scale 0–1			–13***			–9**	–9***
Problem perception of alcohol consumption – in society: scale 0–1			–33***			–30***	–18***
Ideology
Left–right ideology: subjective scale 0 (left)– 1 (right)				–13***			–14***
Alcohol liberalism: scale 0 (restrictive)–1 (liberal)				–38***			–30***
Constant	58***	67***	71***	87***	61***	64***	84***
Adj R 2	.05	.05	.14	.24	.11	.20	.34
N	1.588	1.601	1.230	1.452	1.579	1.198	1.141

Source: National Society, Opinion, Media survey, 2016.

Note. OLS = ordinary least squares. The dependent variable is an index of the mean value of two questions regarding the most important political instruments: “lowering taxes on alcohol” and “allowing strong beer, wine and spirits to be sold in grocery shops” (i.e., abolishing Systembolaget’s retail monopoly). The index is coded 0 (lowering taxes and abolishing the monopoly are both very bad proposals) to 100 (both very good proposals). All independent variables are coded 0–1 and presented in Table 1. Adj R 2-values are written in italics.

*p < .05. ** p < .01. *** p < .001.

Results: Test of six hypotheses on support for restrictive alcohol policies

Our first hypothesis on the effects of narrow self-interest (H1) predicted that support for restrictive alcohol policies would be negatively correlated with people’s alcohol consumption. This hypothesis is clearly supported in Table 3. The combined explanatory power of our two consumption indicators is Adj. R 2 = .05. The effects of the indicators are reduced in M6 and M7, where variables pertaining to knowledge, perception, experiences and ideology are introduced, but they remain significant. However, the overall impression is that the effects of alcohol consumption – here interpreted as self-interest – are relatively modest in relation to other explanatory factors.

The second hypothesis (H2), predicting that people to the political right are more negative towards restrictive alcohol policies, is clearly supported by the results. However, the left–right effect, whilst substantial, is overshadowed by the effect of alcohol liberalism, as there is also strong support for the prediction of the third hypothesis (H3) – that, taking into account left–right position, people’s position vis-a-vis a separate ideological dimension regarding individual versus state responsibility for alcohol consumption would affect their attitudes.

The fourth hypothesis (H4), that people who have experience of someone close to them drinking would be expected to be more supportive of restrictive alcohol policies, is not supported. There are no indications that such personal experiences have any effect on opinions, directly or indirectly. However, the importance of another kind of knowledge is substantiated in Table 3. People who have accurate knowledge of alcohol-related issues in society are, as expected, more supportive of restrictive alcohol policies. H5 is thus confirmed, even though the effect of knowledge in the final model is quite weak in relation to other independent variables.

The sixth hypothesis (H6) predicted that people who generally perceive the consequences of alcohol consumption to be more negative than positive would be supportive of restrictive alcohol policies. This hypothesis is also thoroughly supported. Furthermore, the results indicate that the effect of perceiving it to be a problem at a societal level is also clearly more important for explaining political opinions than at a personal level. The effect of perceiving it to be a societal problem is somewhat reduced in M7. However, since the results in Table 2 show that perceiving it to be a societal problem is an important explanatory factor behind people’s position on left–right ideology as well as on alcohol liberalism, part of the already substantial effect on policy opinions of perception of it as a societal problem is indirect via ideology.

The effects of people’s social characteristics on their opinions regarding alcohol policies are not particularly substantial. People who are female, come from high-income households, are highly educated and religious are more supportive of restrictive policies. It should be noted that a number of effects of social background are indirect, primarily via alcohol consumption, problem perception and ideology.

Discussion

Approximately 88% of the adult population in Sweden drink alcohol at least once a year, and three quarters of them perceive the consequences of this consumption in their private lives as positive or at least neutral. Restrictive policies make this pleasurable consumption more expensive and less convenient. How is it that so many Swedes still want to keep – or increase – such restrictions?

This article has shown that people’s self-interest does indeed have an impact: frequent alcohol consumers who are most affected by restrictions are less likely to support restrictive policies. However, the effect of narrow self-interest is on the whole quite limited.

Instead, the results show that perceptions of problems caused by alcohol consumption, in combination with ideological values and norms, are much more important for explaining public opinion. People to the left are more supportive of restrictive policies, but ideological beliefs specifically relating to the responsibility of the individual alcohol consumer versus the state are even more important than left–right position.

Furthermore, even though most people perceive the consequences of alcohol consumption as positive or neutral in their own lives, there is an overwhelming consensus around the view that the consequences for society are mostly negative. And it is a perception that there is problem at the societal level, rather than the personal level, that is most essential for explaining opinions on alcohol restrictions. General knowledge of alcohol-related matters has some effect, whereas personal experiences of close affiliates drinking too much have no effect on opinion.

The conclusion must therefore be that the firm support for restrictive alcohol policies in Swedish public opinion is mainly founded on norms of solidarity, accurate knowledge and astute problem analyses at the societal level, and, to a much lesser extent, on egoism and personal experiences.

The findings of this study may also be of interest for policy makers, especially those who seek to retain a restrictive line on alcohol. In an ideal democracy, public opinion is expected to be the base for public policy, and it has been suggested that public opinion might indeed be the single most important explanation for the development of alcohol policy (Baggott, 1986). Earlier studies have also previously suggested that changes in how problems are perceived over time can explain changes in attitudes towards alcohol policy (Storvoll et al., 2015). Policy makers should therefore be advised to closely follow developments in public opinion on alcohol. Furthermore, our results suggest that working to preserve norms and reminding people of the downsides of drinking are much more likely to be more relevant as policy strategies than worrying about how adjusted taxes or sales restrictions might affect people’s convenience or wallets.

A final reflection is that even though opinion in regard to alcohol shapes policy, opinion is probably also shaped by policy (Saglie, 1996). This is likely one of the explanations for why the Swedish case studied in this article differs from the neighbouring Danish case. Denmark is often described as the country most similar to Sweden in terms of welfare state organisation. However, it has been argued that Denmark’s traditional lack of political intervention in relation to alcohol has led to an understanding of problems associated with alcohol as part of the private rather than the collective sphere (Elmeland & Villumsen, 2013). Sweden’s long tradition of a restrictive line, historically initiated by a vigorous temperance movement, has possibly contributed to reproducing different norms.