A Survey That Doesn't Match the Behavior
A high school surveys 600 students about recycling. Results show 82% say recycling is 'very important,' but building audits find only 31% of recyclable material actually reaches the correct bin. The principal wants to launch an awareness campaign based on the survey. A teacher points out that the survey measured stated attitudes, not behavior, and that students may have answered to look good. The school has a small budget and one semester to show improvement.
1. Why is the gap between stated attitude and observed behavior a measurement problem, and what concepts explain it?
Read the worked answer
Model reasoning:
- Attitude–behavior gap. Stated attitudes are weak predictors of behavior, especially when behavior is inconvenient or unobserved. The 82% figure measures endorsement, not action.
- Social desirability bias. Students answering a survey about recycling have an incentive to report the socially approved answer. The 82% is likely inflated.
- Measurement validity. The survey instrument does not measure the construct the principal cares about (recycling behavior). The audit does - it is a behavioral measure, though it has its own limits (contamination, bin placement).
- Ecological inference risk. Concluding 'students don't care' from the audit, or 'students care, so bins will fill' from the survey, are both unsupported.
The judgment: the two data sources answer different questions, and the principal should treat the audit as the outcome measure and the survey as, at best, a diagnostic of awareness.
2. Design an evaluation the school could run in one semester that would actually test whether a campaign works.
Read the worked answer
Model reasoning - the key move is a comparison, not a before/after only:
- Baseline: audit several comparable locations (e.g., cafeteria, two hallways) before any intervention.
- Comparison group: run the campaign in some locations and not others (or stagger the rollout), so improvement can be attributed to the campaign rather than to a seasonal trend or a simultaneous change.
- Outcome measure: contamination rate in bins, measured the same way at baseline and end.
- Secondary measure: a short behavioral or observational check (e.g., counts at bins) rather than another attitude survey.
- Limits to state: one semester is short; novelty effects may inflate early results; a small number of locations limits statistical power.
The teacher's caution is operationalized: the school measures behavior, uses a comparison, and does not rely on self-report to declare success.