Content
Summary
Surveys gather data from a sample to infer characteristics of a larger population, requiring careful design to minimize bias and error. Effective implementation involves defining objectives, structuring question flow, and selecting appropriate sampling techniques such as simple random, stratified, or cluster sampling.
Intuition
Think of a survey as a window into a large population. You cannot see the whole picture at once, so you look through a small, carefully chosen pane (the sample). If the pane is dirty (biased questions), cracked (sampling error), or only shows one side of the room (convenience sampling), your view of the whole room (population) will be distorted. The goal is to construct a clear, unbiased window that accurately reflects the diversity and reality of the entire population, using methods like random selection to ensure every part of the room has an equal chance of being seen.
Derivation
- Define the research objective and target population.
- Select a sampling method (e.g., simple random, stratified) to ensure representativeness.
- Design questions (open/closed) and structure the survey flow (simple to complex).
- Pilot test the instrument to identify confusion or bias.
- Administer the survey and analyze data, accounting for potential sampling and measurement errors.
Worked example
A researcher wants to study mental health in college students. They use student ID numbers to randomly select 100 participants. What sampling method is this, and what are two potential sources of error mentioned in the case study?
- Identify the method: Using student ID numbers and random selection gives each student an equal chance, which is simple random selection.
- Identify errors: The case study mentions self-selection bias (if students choose to participate), under-coverage bias (if only one university is sampled), measurement error (self-reported habits), and non-response bias (if dropouts differ from completers).
Answer Simple random selection. Potential errors include self-selection bias, under-coverage bias, measurement error, and non-response bias.
Common mistakes
- Assuming convenience sampling provides a representative sample.
- Using leading questions that presuppose a preference.
- Ignoring the risk of bias in systematic sampling if the list has a hidden pattern.
- Confusing sampling error (chance) with sampling bias (unrepresentative frame).