Sampling, recruitment and data collection

Data were from UK participants in wave 1 of the International Food Policy Study, conducted in Australia, Canada, Mexico, the UK and the USA. Data were collected via self-completed web-based surveys in December 2017 with adults aged 18–64 years. Respondents were recruited through Nielsen Consumer Insights Global Panel and their partners’ panels. Email invitations (with a unique link) were sent to a random sample of panellists (after targeting for age and country criteria); panellists known to be ineligible were not invited. The mean survey time across countries was 33 min.

Respondents provided consent prior to completing the survey. Respondents received remuneration in accordance with their panel’s usual incentive structure (eg, points-based or monetary rewards, or chances to win prizes). A full description of the study methods can be found in the International Food Policy Study: Technical Report—Wave 1 (2017) at www.foodpolicystudy.com/methods.

Variables used in the analysis

The variables used in the analysis, the survey items they were derived from, response options and how response options were collapsed for analysis are described in table 1.

Alongside single-item measures of attitudes, knowledge and social norms related to sugary drinks; we included single items measures of trust in advice on sugary drinks from health experts, and the food and beverage industry; and single item measures of support for, and perceived effectiveness of, the SDIL. As previous research has indicated that the acceptability of food taxes varies with the stated intentions of these,11 12 19 20 we included a preamble to the questions about support for, and perceived effectiveness of, the SDIL outlining the intention of the levy and the stated use of revenue generated.

Sociodemographic variables considered were age in years, sex at birth, whether or not participants had children and socioeconomic position. Parental status was a potentially important variable because the SDIL is included as a flagship component of England’s Childhood Obesity Plan and has particularly been framed in terms of potential benefits to children.1 23 Socioeconomic position was measured using participants’ highest educational qualification and perceived income sufficiency.

The Beverage Frequency Questionnaire is a 7-day food record that assesses consumption for 17 beverage categories, including caloric and non-caloric beverages.24 For each beverage category, respondents report the number of drinks and the usual portion size, using category-specific images of beverage containers, adapted from the ASA24 dietary recall.25 Participants who reported any consumption of regular fizzy drinks (including alcoholic drinks that contained regular fizzy drinks as a mixer), sweetened fruit drinks, sports drinks or energy drinks over the previous 7 days were considered SSB consumers in the analysis.

Inclusion criteria

UK resident participants in wave 1 of the International Food Policy Study, aged 18–64 years, who correctly responding to a data integrity question in which participants were asked to identify the current month, and provided usable information on all other variables of interest were included in the analysis. Data from countries other than the UK were not included as comparable questions on support for, and perceived effectiveness of, the SDIL were not asked of these participants.

Analysis

Data were weighted with poststratification sample weights constructed using population estimates from the UK census based on age group, sex and region. These sample weights were used throughout the analysis to reduce the effects of non-response and selection bias and return the sample to population representativeness.

Descriptive statistics were used to quantify all variables of interest. Logistic regression models were fitted to explore associations between other variables and support for, and perceived effectiveness of, the SDIL. We used separate models to explore support for the SDIL and perceived effectiveness of the SDIL where support for, or perceived effectiveness of, the SDIL were the outcome variables and all other variables were included as explanatory variables. Unless otherwise noted, adjusted OR (and 95% CI) of support for, or perceived effectiveness of, the SDIL are presented adjusted for all other variables included.

Data were analysed using R V.3.3.1.

Patient and public involvement

Patients and the public were not involved in design, conduct, analysis or interpretation of the study.

Summary of findings

To our knowledge, this is the first study of a range of sociodemographic, consumption and psychological correlates of both support for, and perceived effectiveness of, an SSB tax. Unlike previous studies, our research was conducted in the context of a definite, rather than hypothetical, SSB tax. We found that the majority of UK adults aged 18–64 years were supportive of the SDIL and believed it would be effective, have a positive attitude to sugary drinks, have good knowledge about the links between sugary drinks and obesity, and trust messages from health experts, but not the food and beverage industry, about sugary drinks. Around half reported social norms about not drinking sugary drinks.

Social norms towards not consuming sugary drinks and trusting health expert messages on sugary drinks were both associated with greater support for and perceived effectiveness of the SDIL. In addition, having dependent children and trusting messages from the food and beverage industry on sugary drinks were associated with less support for the SDIL, while older age, not consuming SSBs and knowledge of the link between sugary drinks and obesity were associated with greater support. Older age was associated with lower perceived effectiveness of the SDIL, and more negative attitudes towards sugary drinks were associated with greater perceived effectiveness. There were no associations between gender, education or income sufficiency and either support for, of perceived effectiveness of, the SDIL.

Strengths and weaknesses of methods

Key strengths of the analysis are the use of a large, population representative, sample; inclusion of a range of sociodemographic, consumption and psychological variables; and the context of a definite, rather than hypothetical, SSB tax announced 20 months before data collection (although not implemented until 4 months after). Given previous findings that support is greater when revenues are used for health-promoting activities,11 12 19 20 we were careful to present the SDIL as an intervention targeting manufacturers rather than consumers, with revenues ear-marked for health-promotion activities. Social desirability bias may be less likely to occur in more anonymous settings such as online surveys.26

Participants were not recruited using probability-based sampling meaning the findings do not provide nationally representative estimates, although this was reduced by applying sampling weights. The results are, therefore, likely to be generalisable to the UK, but may not be more widely generalisable. This is a cross-sectional analysis and we cannot be sure of the direction of causation between putative explanatory variables and outcomes. Nor have we explored more complicated causal networks linking the variables included. All variables were self-reported. While all have strong face validity, we have not explored other aspects of validity or reliability of any of the measures used. However, all were derived from existing instruments and in many cases it would be hard to know what a ‘gold standard’ measure should be for validation. Although a high proportion of participants who completed the survey were included in the analysis, we do not know what proportion of those invited to participate were included.

Comparison to previous results and interpretation of findings

Most people in our survey (90%) knew that there was an association between sugary drink consumption and obesity. This reflects previous findings where 89%–91% agreed that SSB consumption increased the risk of obesity.3 5 Despite this, there were also high positive attitudes towards sugary drinks with almost two-thirds of respondents agreeing that sugary drinks taste good, and less than half had social norms about not drinking sugary drinks. In the UK, SSBs appear to remain a pleasurable and positive part of life, despite their known health harms.

Similar to previous research which found that only 30% of Americans gave favourable ratings to soda companies,22 we found low levels of trust in messages about sugary drinks from the food and beverage industry. Levels of trust in similar messages from health experts were higher, but still less than two-thirds. Low levels of trust in experts may reflect a general public mistrust of nutritional epidemiology.27

Despite less than perfect trust in messages about sugary drinks from health experts, there was a high level of support for the SDIL (70%) and even higher belief that it would be effective (71%). This is higher than previous research which, as far as we are aware, reports maximum support of 60%.8 28 Even in the context of an existing tax on sweetened drinks in France, only 49% supported the tax.19 The high level of support we found may reflect the combined effect of previous findings that support for more intrusive public health interventions such as taxes on food and tobacco often increases after implementation,2 and that support for SSB taxes is often greater when revenues are used for health-promoting activities.11 12 19 20 28 29 Although the SDIL had not been implemented at the time of data collection, impending implementation had been known of for 20 months. Further, we were careful to inform participants that SDIL revenues would be spent on school breakfast clubs and sports activities. In addition, the SDIL is unique in being targeted at manufacturers rather than consumers, and intended to promote reformulation rather than necessarily reduce consumption.1 Previous qualitative work has found that those who do not support generic SSB taxes often cite excessive personal taxation and government intrusion into individual’s lives as reasons for this.11 30 This is much less applicable to the SDIL than to consumer-facing SSB taxes.

Low acceptability of SSB taxes has previously been ascribed to a perception that they are unlikely to achieve significant behaviour change or public health benefit.11 30 Previous research has reported perceived effectiveness (to improve population health or decrease SSB consumption) in the range of 39%–58%.5 12 19 In contrast, we found much higher levels of perceived effectiveness (71%). This may again reflect the unique nature of the SDIL with an explicit intention to change manufacturer, rather than consumer, behaviour— and our focus on effects on industry, rather than consumer, behaviour.

Higher support for, and perceived effectiveness of, the SDIL here compared with previous work may also reflect cultural differences between the UK and other countries where previous data has been collected. Unlike previously, we used population weighting which increases confidence that results are population representative. Finally, it is possible that the unique design and framing of the SDIL makes it more acceptable and increases perceived effectiveness compared with previous taxes proposed to research participants.

The pattern of associations between attitudes, social norms, trust and support for, and perceived effectiveness of, the SDIL are, for the most part, intuitive. It might be expected that non-consumers, who are less likely to be negatively financially effected by the tax, would be more supportive. In other contexts, those who stand to gain most from financial incentive interventions are most supportive.31 Social norms to not drink sugary drinks, negative attitudes towards sugary drinks, greater knowledge about the health harms of sugary drinks, greater trust in health experts and less trust in the food and beverage industry all reflect more ‘public health’ orientated patterns that would be expected to be associated with greater support for, or perceived effectiveness, of the SDIL.

It is somewhat surprising that those with children under the age of 18 years were less supportive of the SDIL than those without. The SDIL was particularly framed in terms of potential benefits to children.1 23 If one’s own consumption is likely to influence support for the SDIL, then parents’ support for the SDIL may also be influenced by their children’s consumption. If children are greater consumers of sugary drinks,32 then this may explain why parents with children under the age of 18 years were less supportive. As described above, previous research on the association between psychological variables and support for, and perceived effectiveness of, SSB taxes is sparse.

We did not find that gender or markers of socioeconomic position were associated with support for, or perceived effectiveness of, the SDIL in mutually adjusted models. This reflects some, but not all, previous findings.5 8 10 19 22 Unlike most previous work we included a wide range of sociodemographic, consumption and psychological variables in mutually adjusted models and it may be that gender or socioeconomic differences operate entirely through the other variables included in our models.

Implications of findings

Many structural public health policies require government action, which may be limited by perceptions concerning public acceptability of such policies—often uninformed by evidence. Greater understanding of public acceptability of a range of structural public health policies, and how this changes over time and over the course of implementation, may help to develop strategies to address public concerns and build public support for these policies.