Primary Sources of Data: Quantitative Methods
Self-Completion Questionnaires: Instruments delivered via post, email, or by hand that respondents fill out independently. This method is highly cost-effective for reaching large, geographically dispersed samples.
Structured Interviews: A researcher reads questions from an 'interview schedule' to a respondent in person or over the phone. This allows the researcher to clarify questions but introduces the risk of interviewer bias.
Closed Questions: Fixed-choice questions (e.g., 'Yes/No' or Likert scales) that provide quantitative data. These are easily analyzed using software to generate statistical summaries.
Longitudinal Designs: Studies conducted over long periods to track changes. Panel studies follow a specific group on multiple occasions, while cohort studies follow a group sharing a specific characteristic at regular intervals.
| Feature | Questionnaire | Structured Interview |
|---|---|---|
| Cost | Relatively low; minimal staff needed | High; requires trained interviewers |
| Response Rate | Often low (postal/email) | Higher; harder for respondents to refuse |
| Clarification | Impossible; leads to misunderstandings | Possible; researcher can explain terms |
| Bias | Low researcher influence | High risk of 'Interviewer Effect' |
Reliability vs. Validity: Always distinguish between these two terms. Reliability refers to the consistency and replicability of the method (surveys are usually high in this), while Validity refers to the truthfulness or accuracy of the data.
The Positivist Preference: In exam questions about 'why' a method is chosen, explain that positivists prefer surveys because they allow for the testing of hypotheses and the discovery of social laws through numerical evidence.
Verification: When evaluating a method, consider the 'three Rs': Is it Representative, is it Reliable, and is the Response Rate sufficient to avoid bias?
Interviewer Bias/Effect: Students often forget that the mere presence of an interviewer (age, gender, ethnicity) can unconsciously influence the respondent's answers, reducing the validity of the data.
Social Desirability Bias: This occurs when respondents give 'correct' or 'positive' answers to look good in the eyes of the researcher, a common issue in personal interviews compared to anonymous questionnaires.
Longitudinal Confusion: Do not confuse panel studies (same individuals) with cohort studies (same starting characteristic but not necessarily identical individuals over time).