Research and Analysis: Design, Measurement, and Interpretation
Core concepts
Research questions test the research process: define the problem, design the study, collect data, analyze, and report. Distinguish primary from secondary data and qualitative (focus groups, depth interviews, observation) from quantitative (surveys, experiments) methods. Know the three designs: exploratory (clarify a fuzzy problem), descriptive (measure and describe), and causal (test cause and effect, usually via experiment).
Understand sampling: probability methods (simple random, stratified, cluster, systematic) versus non-probability (convenience, judgment, quota, snowball). Know that sampling error falls as sample size grows, while non-sampling error (frame, measurement, response, non-response) does not automatically shrink. Grasp validity (measuring what you intend) and reliability (consistency). A scale can be reliable but invalid. Know Likert, semantic differential, and constant-sum scales, and the difference between nominal, ordinal, interval, and ratio data, which determines permissible statistics.
Traps
Watch for questions that imply a bigger sample fixes a biased frame - it does not. Confusing correlation with causation is a common distractor. Be careful with leading or double-barreled questions in questionnaire design. Do not confuse exploratory with descriptive when the goal is to size a market (that is descriptive).
Worked rule of thumb
Match design to question type: "Why are sales falling?" (unclear cause) suggests exploratory; "What share of buyers prefer X?" suggests descriptive; "Does price change affect units sold?" suggests causal/experimental.
Key points
- Exploratory, descriptive, and causal designs answer different question types.