Survey of Sulfite Attitudes: 18% Response and a Funded Chair
A candidate surveys 600 wine professionals about sulfite limits, recruiting through LinkedIn and two trade newsletters. After six weeks, 108 responses (18%) are received. The survey promises anonymity, yet asks for job role, region and company size; several combinations have fewer than five respondents. The study is funded by a producer that markets 'low-sulfite' wines, and the funder's logo appears on the landing page. The candidate wants to conclude that 'the industry supports lower sulfite limits' and to recommend a policy change. The questionnaire uses a five-point agreement scale, but the key item reads: 'Sulfites are a problem for consumers.' No pilot was conducted. The candidate has a 10-week deadline and no budget for follow-up. The dissertation question is: 'What do wine professionals think about sulfite regulation?'
1. Evaluate the claim that 'the industry supports lower sulfite limits' based on 108 responses from 600 invited.
Read the worked answer
The claim is not supported as stated. An 18% response rate leaves 492 non-respondents, and respondents are self-selected through LinkedIn and trade newsletters; they may differ systematically in role, region, sulfite attitudes or engagement with the topic. The funder's logo on the landing page may attract people sympathetic to low-sulfite messages and deter others. The sample cannot be treated as representative of 'the industry' without a sampling frame and non-response analysis. The key item is also leading: 'Sulfites are a problem for consumers' asks respondents to endorse a premise, not to state a limit. Even if 80% agreed, the correct inference is: 'Among 108 self-selected respondents, X% agreed with this statement.' To support an industry claim, the candidate would need a defined population, a probability or quota sample, a pilot-tested neutral item, and a non-response follow-up. With 10 weeks and no budget, the dissertation should be framed as an exploratory survey and should not recommend policy change on this basis.
2. What ethical and reporting issues arise from the funding and the anonymity promise?
Read the worked answer
The funding creates a conflict of interest that must be disclosed in the dissertation, the consent form and any publication. The funder should have no role in questionnaire design, analysis or veto over findings; if that cannot be guaranteed, say so. The anonymity promise is at risk because job role, region and company size can identify individuals in small cells, especially where fewer than five respondents share a combination. The candidate should remove or aggregate identifying variables, suppress small cells, and store data securely. Consent must state the funder's involvement and the limits of anonymity. The leading item and the absence of a pilot are reporting issues: the questionnaire should be reproduced in an appendix, with the exact wording. The candidate should also report the response rate, recruitment channels and any exclusions. If the funder's logo was on the landing page, that is a potential recruitment bias and must be described. The dissertation should include a limitations section that names these issues rather than burying them.
3. Propose an analysis and reporting plan that makes the most of the data without overclaiming.
Read the worked answer
Begin with a descriptive table: response count by role, region, company size and recruitment channel; report the 18% response rate and the unknown denominator for each channel. For the key item, report the full distribution, not just agreement, and use exact confidence intervals for proportions. Cross-tabulate by role and region, but suppress cells with fewer than five and label the rest as exploratory; do not run a battery of significance tests without correction. If a neutral item was not asked, do not infer support for a specific limit; report only agreement with the statement as worded. Include a non-response sensitivity paragraph: how different would the non-respondents have to be to overturn the headline? Present the funder disclosure and the anonymity limitations. In the discussion, distinguish 'respondents' from 'the industry' and recommend a probability survey or Delphi panel as the next step. The strongest honest conclusion is that the survey identifies themes worth testing, not a mandate for policy change.