Overview
This lecture focused on identifying different types of survey bias, explaining their impact on data quality, and practicing example scenarios to recognize bias.
Importance of Survey Quality
- The reliability of population summaries depends on whether surveys are conducted well.
- Surveys can be good or bad, largely determined by the presence of bias.
- Biased surveys produce untrue or misleading data, so identifying bias is crucial.
Types of Survey Bias
- Sampling bias occurs when the sample does not accurately represent the population.
- Voluntary response bias happens when only people with strong opinions choose to participate.
- Non-response bias arises when people choose not to answer, skewing results.
- Convenience bias occurs when the sample is chosen based on ease of access, not representativeness.
- Measurement bias results from poorly worded or misleading survey questions that influence responses.
Example Scenarios & Bias Identification
- Asking Facebook friends about Facebook use = convenience bias (and possibly voluntary response).
- Loaded question about taxes and unemployment = measurement bias (and voluntary response).
- CNN survey on military on July 4th = voluntary response bias, timing, and source effects.
- Asking about STDs = non-response bias due to privacy/shame.
- Surveying grocery shoppers about eating out = convenience bias (negative skew).
- Random calls from the Yellow Pages = non-response and convenience bias (excludes non-landline users).
Key Terms & Definitions
- Bias — Systematic error in data collection leading to unrepresentative results.
- Sampling Bias — When the chosen sample does not accurately reflect the population.
- Voluntary Response Bias — Only people with strong feelings respond, skewing results.
- Non-response Bias — Bias from people refusing or failing to answer.
- Convenience Bias — Sample is chosen for ease, not representativeness.
- Measurement Bias — Survey questions are worded to influence or confuse respondents.
Action Items / Next Steps
- Practice identifying types of bias in sample survey scenarios.
- Be prepared to discard data from surveys exhibiting clear bias.