The One-Week Dropout Survey
A Class XII student in Pune wants to study why students drop out after Class X. She has one week and no budget. She plans to print 100 questionnaires, hand them out during a single lecture at her own junior college, and ask: 'Don't you agree that family poverty is the main reason students drop out?' She will write each respondent's name on the form 'for attendance', but tells students the survey is anonymous. She intends to conclude that 'poverty is the main cause of dropout in Maharashtra' and submit the report to a district-level competition. Her teacher has approved the plan without reading it carefully.
1. Identify three design errors in this plan and explain the specific distortion each produces.
Read the worked answer
Three errors, each with a distinct distortion.
- Sampling frame produces survivor bias. Surveying only students still enrolled in Class XI-XII at her own college excludes exactly the people who dropped out. No matter how large the sample, it cannot answer the question. The finding would describe continuing students' opinions about dropout, not dropouts' reasons.
- The question is leading. 'Don't you agree that family poverty is the main reason?' invites acquiescence: respondents tend to agree with a stated proposition, especially in a classroom where the researcher is a senior student. It measures agreement with the researcher, not the cause.
- Anonymity is broken. Collecting names 'for attendance' while promising anonymity is a contradiction; students may fear that answers reach teachers. This produces social-desirability bias and is effectively coercive, since participation is required.
- Overgeneralisation. A one-week convenience sample of 100 Pune college students cannot support a claim about all Maharashtra. External validity is close to zero. The design is not merely small - small samples can be defensible. It is unrepresentative and biased.
2. Rewrite one question and propose a feasible sampling strategy she could complete in one week.
Read the worked answer
A neutral question, and a feasible week-long sample. Rewritten item: 'In your experience, what makes students leave school after Class X? Tick up to two: (a) cost or fees, (b) need to earn, (c) distance or safety, (d) marriage or domestic work, (e) repeated failure, (f) illness, (g) other ___.' This avoids naming a cause in advance and allows the respondent's own ranking. Sampling: she cannot reach dropouts through her college. In one week she can (i) ask two local teachers and an anganwadi worker to identify 15-20 recent dropouts; (ii) use snowball referral from those; (iii) add a comparison group of 20 continuing students from the same locality; (iv) stratify by gender and ward. Thirty dropouts and thirty continuing students, interviewed at home with consent, is far more defensible than 100 captive classmates. She should report the limitation honestly rather than claim state-wide validity.
3. What ethical issues arise, and how should each be fixed?
Read the worked answer
Ethics: consent, confidentiality, no harm, honesty.
- Informed consent: explain purpose, time, voluntary nature, right to skip questions, and that refusal will not affect marks. For minors, obtain guardian consent.
- Confidentiality: no names on questionnaires; use codes. Attendance and survey should be separate sheets, or attendance collected before or after, never linked.
- No coercion: distributing forms in a lecture and implying attendance is coercive; a classroom is a captive setting. Allow take-home return.
- No harm: questions about poverty and failure can distress a respondent; keep them sensitive and provide referral information.
- Honesty in reporting: do not call a convenience sample 'representative'.
- Permission: take the principal's permission and check the competition's research guidelines. The core principle is that the researcher's convenience never overrides the respondent's autonomy.