Reconciling Visitor Data for a Heritage-Tourism Study
A state cultural department reports 1.2 million visitors in 2019 across five heritage sites. ASI ticket counters record 0.7 million. A quick survey of 500 visitors finds 62% came “for photography,” 18% for “religious reasons,” 12% for “education,” and 8% “other.” Two sites complicate the count: Konark has free entry on three national holidays, and Hampi has four ungated entry points with no ticketing. A consultant uses the 500-person survey to claim that “heritage tourism raises local household incomes by 20%,” and proposes a ₹6 lakh, 12-month impact study. The department asks you to review the claim and the study design before sanction. You have no additional budget for a census, but you can add questions to the existing survey and use local interviews.
1. Explain the discrepancy between 1.2 million and 0.7 million visitors, and recommend how to measure footfall at the two problem sites without a full census; discuss coverage and measurement error.
Read the worked answer
The gap is unlikely to be a single counting error. The state figure may include free-entry days, non-ticketed access, school groups, and local worshippers, while the ASI figure counts only tickets sold. Konark's three free national holidays would create a real spike that ticket data miss; Hampi's four ungated points mean that ticket counts measure only one segment of visitors. There may also be double counting on multi-site passes, or differences in the reporting year and the treatment of children. The two totals may even measure different constructs: “visits” versus “ticketed visits.”
Without a census, use stratified direct observation: at each ungated or free-entry point, conduct 15-minute manual counts in eight randomly selected time blocks per day, repeated on weekdays, weekends, and holidays, and rotate the blocks across a month. Combine these with gate counts where available and with a short exit interview to estimate the proportion of local worshippers versus tourists. Report a range, not a false precision, and state the coverage error explicitly. The department should publish a definition sheet-what counts as a visit, how repeat visits are handled, and whether free-entry days are included-so future figures are comparable.
2. Critically evaluate the consultant's causal claim from the 500-visitor survey; identify at least three threats to validity and propose a stronger design within the budget.
Read the worked answer
The 500-visitor survey cannot support a 20% household-income claim. First, it samples visitors, not households; it says nothing about local incomes and suffers selection bias because it misses people who avoid tourist areas. Second, it is cross-sectional: a correlation between tourism and income at one moment cannot establish that tourism caused the income change. Third, omitted variables-road access, remittances, agricultural prices, and government schemes-could explain both tourism and income. Fourth, self-reported reasons for visiting (“photography”) are not a measure of economic impact, and the 62% figure may reflect the survey's location and timing. Finally, 20% is a point estimate without a confidence interval or a stated baseline.
A stronger design within ₹6 lakh: a repeated household survey in two matched clusters-one near the heritage site and one comparable but farther away-with a baseline and a 12-month follow-up; a short enterprise survey of homestays, guides, vendors, and transport operators; and 20–25 semi-structured interviews. Use difference-in-differences if baseline data can be collected, or at least compare changes in the two clusters. Pre-specify income definitions and report uncertainty.
3. Propose a mixed-methods 12-month design with sampling, instruments, and ethical safeguards, and state two indicators that would show whether tourism income reaches local households.
Read the worked answer
Design: a sequential mixed-methods study. Months 1–2: scoping, site mapping, and translation of instruments. Months 3–4: baseline household survey in two matched clusters, with a target of 400 households (200 per cluster), sampled by random walk from updated voter or ration-card lists; stratify by distance to the site and by caste/occupation to avoid capturing only the entrepreneurial fringe. Months 5–11: monthly enterprise diaries from 60 businesses, visitor exit surveys (n≈600) using the added questions, and 25 interviews with workers, women's groups, and local officials. Month 12: analysis and validation workshop.
Instruments: household income and expenditure module, employment module, enterprise revenue and wage module, visitor spending module, and an interview guide. Ethical safeguards: informed consent in Odia and English, no inducement that coerces participation, anonymised income data, secure storage, and community feedback before publication.
Indicators that income reaches households: (1) the share of local workers in tourism enterprises who report regular cash wages and the median wage, tracked against the comparison cluster; (2) the proportion of households reporting tourism-related income and the change in that proportion from baseline, disaggregated by gender and caste. These are direct transmission indicators, not just aggregate visitor numbers.