The protocol for conducting this survey was reviewed by the Queen's University Research and Ethics Board and approved April 19, 2012. Each respondent was provided contact information for the principal investigator and the chair of the Research and Ethics board. Consent was assumed by the respondent agreeing to participate in the survey.
The questionnaire was designed to obtain responses from a random sample of the Kingston population on their opinions and attitudes towards alcohol use and effects on the community.
The development of the survey was informed by a review of the literature3-6,8,9 and input from community partners involved in the local alcohol misuse and injury prevention coalition. Based on the literature review and the needs of the coalition, it was determined the questionnaire would solicit the following from respondents:
There were six sections to the questionnaire; one section for each of the areas outlined above plus a section to gather demographic information. In the first question, respondents were asked to use a four- point scale to rate how common or uncommon they thought each of twelve reasons were for people to drink. A neutral response (i.e. neither common nor uncommon) was not a choice option; however, they could respond with 'I don't know' or could refuse to answer the question. In the second question respondents were asked to use the same four-point scale (very common to very uncommon) to rate eight factors thought to contribute to alcohol problems in a community.
The intent of the third question was to gauge the community's perception of whether there are alcohol- related issues in Kingston. Respondents were asked to rate how much of an issue they thought fifteen items were. The four-point scale consisted of the response choices: not at all, slight, moderate, and very much an issue.
The second hand effects of alcohol use were measured by determining how many people are affected by someone else's drinking. Respondents were asked whether they had experienced or not, in the previous 12 months, five different interactions with people who had been drinking alcohol.
The next section of questions used different question formats to ask about support for and awareness of actions that could be taken in a community to affect the alcohol and drinking environment. Likert-type format was used to find out the level of agreement with community actions (strongly agree to strongly disagree). As with the other scale-response questions, no neutral response choice was given to respondents.
The final section of the questionnaire asked respondents to provide key demographics that would be used for grouping respondents into meaningful categories such as sex, age, education, and income. Respondents were also asked for their postal code in order to classify respondents by a social and material deprivation index.10
Frequencies were tabulated for demographic variables - age, sex, education, income, and deprivation index (material and social combined and separately) by using postal codes. These frequencies were compared to the City of Kingston population statistics as available (income and education data were not available at the city level) in order to determine the generalizability of information collected by the survey. The chi-square goodness of fit test was used to determine whether the sample differed significantly from the City of Kingston population. If the p-value for the test was less than or equal to 0.05, then the survey was considered significantly different from the actual population and therefore generalizations should not be made.
Cross tabulations were done between sex and each of age, income, and education (independent variables). These tabulations were used to make decisions on further analysis of the dataset. Response categories were collapsed or removed when the frequency within the category was too small to provide meaningful results. Once the appropriate dataset was created all alcohol-related questions (dependent variables) were cross tabulated with age, sex, and material and social deprivation indexes. These crosstabs (or relationship between two variables) were measured for significance by performing a chi-square test. The Cramer's V was the post-test used to provide more information on the strength of the relationship.