Sampling Bias in a University Student Satisfaction Survey
A university's quality assurance cell wants to measure overall student satisfaction. They decide to email a survey to all 12,000 enrolled students. The survey link is sent only to official university email addresses. Within two weeks, 1,800 responses are received. The response rate is 15%. The survey includes a question about library services, and 68% of respondents rate them as 'satisfied' or 'very satisfied'. The university's library reports that only 40% of students ever visit the physical library; the rest use online resources. The survey also asks about online learning platforms, and satisfaction there is 55%. The QA cell plans to present the 68% library satisfaction figure as representative of all students.
1. Identify and explain the main sampling biases in this survey. How do they affect the validity of the 68% library satisfaction figure?
Read the worked answer
The survey suffers from non-response bias and coverage/selection bias. Non-response bias: only 15% responded; those who respond to email surveys tend to be more engaged or have stronger opinions (either very satisfied or very dissatisfied). The 68% figure may over-represent satisfied students if dissatisfied students ignored the survey. Coverage bias: the sampling frame is official university email addresses, which may exclude students who rarely check email (e.g., part-time, remote, or those with IT issues), and these students may have different library usage patterns. Additionally, the question about library services likely taps into physical library use, but only 40% visit physically; the 68% satisfaction may reflect the views of the minority who use the physical library, not the majority who use online resources. Thus the 68% is not representative of all students; it is a biased estimate. To improve, the university should use stratified random sampling by faculty/year, offer multiple response modes, and weight responses by known population characteristics.
2. Suppose the QA cell wants to reduce non-response bias. Propose two concrete strategies and explain how each would help.
Read the worked answer
- Incentives and reminders: Offer a small incentive (e.g., entry into a prize draw) and send personalized reminders. This reduces non-response by motivating a broader cross-section to participate, not just those with strong opinions. 2. Mixed-mode data collection: Provide paper surveys or phone interviews for students who do not use email regularly. This addresses coverage and non-response bias by reaching students who are otherwise excluded. Both strategies aim to increase response rate and diversity of respondents, making the sample more representative. However, they do not eliminate bias entirely; weighting adjustments may still be needed.