Identifying and controlling extraneous variables, control groups and matched pairs, sources of bias, sensitivity of content, and the random response technique for sensitive questions.

What this dot point is asking

Edexcel codes 1d.03 and 1d.07 require you to identify factors that lead to bias, including the sensitivity of the subject matter, and to know how to minimise distortion; and to understand the importance of identifying and controlling extraneous variables. At Higher tier this extends to using control groups, the advantage of matched pairs, and the random response technique for sensitive questions. The big idea is a fair test: control everything except the variable being investigated.

Explanatory, response and extraneous variables

In an experiment on fertiliser and plant growth, the fertiliser is explanatory and growth is the response. Extraneous variables (light, water, soil, temperature) must be kept the same for every plant. If they vary, you cannot tell whether a difference in growth was caused by the fertiliser or by, say, more sunlight. Controlling extraneous variables is what makes the experiment a fair test.

Control groups and matched pairs

A control group receives no treatment (or the standard treatment) and provides a baseline for comparison. Comparing the treated group with the control isolates the effect of the explanatory variable. Matched pairs (Higher tier) improve fairness further: each individual in the treatment group is paired with a similar individual (same age, sex, starting weight) in the control group, so differences between the people themselves are reduced and the comparison is cleaner.

Sources of bias

Common sources Edexcel expects you to identify and reduce:

• Sampling bias. An incomplete frame or a non-random method (opportunity sampling) misses part of the population.
• Non-response bias. People who do not respond may differ systematically from those who do.
• Self-selection bias. Volunteers (for example a phone-in poll) are not representative.
• Question bias. Leading or emotive wording steers answers.
• Response bias from sensitivity. On embarrassing or risky topics, people may not answer honestly.

To minimise distortion you use random sampling from a good frame, neutral questions, follow-ups to reduce non-response, and special techniques for sensitive questions.

Sensitive questions and the random response technique

When a question is sensitive (illegal, embarrassing or private), people may refuse or lie, biasing the data. The random response technique protects privacy: the respondent secretly uses a random device (a coin, a card) to decide whether to answer the real sensitive question or a harmless one. Because the interviewer does not know which question each person answered, no individual is exposed, yet because the probability of getting each question is known, the overall proportion giving the sensitive answer can still be estimated.