Survey Research & Questionnaire Design
Surveys are simple to distribute and easy to design badly. This sub-cluster covers the mechanics that separate a defensible survey instrument from a flawed one: questionnaire design (question wording, ordering, and response-option construction), Likert-scale construction and analysis, response-rate management, and sampling-frame design — deciding who the survey can reach and what that implies about who it can generalize to. It complements the sampling and measurement sub-clusters above with the specific practicalities of the survey instrument as a data-collection tool.
Guides
Survey Weighting: Design Weights, Post-Stratification, and Raking
Design weights correct for unequal selection probability, post-stratification corrects for a known population-margin mismatch, and raking (iterative proportional fitting) matches several margins at once without a joint population table. This guide separates the three, shows when each is needed, and works a seeded, reproducible simulation through all three plus the resulting design effect and weight trimming.
Thurstone Scaling: Equal-Appearing Intervals and Judge Ratings
Thurstone scaling (equal-appearing intervals) uses a judge panel to rate statement favorability; the median becomes the scale value and the semi-interquartile range (Q) flags ambiguous items. It is Likert scaling’s direct historical predecessor — and the judge-panel labor it requires is exactly why Likert’s simpler respondent-rated method displaced it in practice.
Conjoint Analysis for Research Surveys: Profiles, Utilities, and Relative Importance
How to build attribute-and-level profiles, collect ranking/rating data, estimate part-worth utilities with dummy-variable regression, and calculate relative importance — with every number in the worked example independently computed.
Best-Worst Scaling: Case 1, 2, and 3 Designs
Best-worst scaling has three distinct case designs, each measuring something different. This guide distinguishes object, profile, and multi-profile BWS, works a Case 1 choice task end to end, and covers counting-based versus modeled (logit/HB) analysis.
Discrete Choice Experiments: Attributes, Levels, and Experimental Design
How to select attributes and levels, construct choice sets with a fractional factorial or D-efficient design, and analyze the results with conditional logit — including a fully worked, independently computed design and analysis example.
Translation and Back-Translation of Research Instruments
Forward translation, reconciliation, and back-translation catch linguistic errors — but a linguistically perfect back-translation doesn’t prove the instrument still measures the same thing in the new language. That’s what cognitive debriefing with target-population respondents is for.
Guttman Scaling: Cumulative Items, Scalogram Analysis, and the Coefficient of Reproducibility
How Guttman scaling orders cumulative items, scores a scalogram, and uses the coefficient of reproducibility (and scalability) to test whether a scale is genuinely unidimensional, with a fully worked example.
Cognitive Interviewing: Think-Aloud vs. Verbal Probing for Questionnaire Pretesting
Think-aloud and verbal probing are the two core cognitive interviewing techniques for catching a survey item’s comprehension problems before fielding. This guide gives a real probe bank by probe type and the actual decision rule for how many interview rounds are enough.
Leading Questions in Surveys: How to Spot Them and Write Neutral Rewrites
A practical diagnostic for spotting a leading survey question — embedded assumption, loaded framing, or one-sided response options — with before/after neutral rewrites across five question formats and a reviewer checklist.
Acquiescence Bias: Balanced Scales, Reverse-Coding, and Its Real Cost
Acquiescence bias (yea-saying) inflates agree/disagree survey responses. Balanced, reverse-coded scales make it detectable — but reverse-coding brings its own costs: respondent confusion and a spurious wording factor in your factor analysis. This guide covers the mechanism, the fix, how to detect acquiescence in data you already collected, and how to weigh the reverse-coding tradeoff.
Semantic Differential Scale: Construction, Scoring & When to Use It
Constructing valid bipolar adjective pairs, Osgood’s evaluation/potency/activity structure, scoring conventions, and when a semantic differential beats a Likert item.
Double-Barrelled Questions: How to Spot and Split Them
A double-barrelled survey item asks two things at once but allows only one answer. This guide gives the and/or diagnostic test for spotting one, plus worked before/after rewrites across Likert, yes/no, and frequency question types.
Non-Response Bias: How to Measure It and What to Report
Response rate is a weak proxy for non-response bias. Three measurement procedures — wave analysis, a non-respondent follow-up sub-sample, and benchmark comparison — plus AAPOR RR1-RR6 reporting and the exact limitations wording reviewers expect.
Likert Scale Survey Design & Statistical Analysis Guide
How to design and analyse Likert data: the difference between a Likert item and a Likert scale, how 5-point, 7-point and forced-choice formats behave, and when the ordinal-versus-interval question decides between non-parametric and parametric tests.
Likert Scale: Construction, Examples & Analysis
How to build a Likert item or scale correctly — unipolar vs. bipolar, how many response points, worked example items, and the ordinal-vs-interval analysis debate.
Survey Question Types: A Practical Guide with Examples
A worked-example catalogue of survey question types: multiple choice and multi-select, dichotomous, category, filter/skip-logic, Likert/rating/ranking, matrix, open-ended, demographic, and attention-check items, with a quick-reference table.
Closed-Ended Questions in Research: Types & Examples
Closed-ended survey questions explained: what they are, closed vs. open-ended, the six main types (dichotomous, multiple choice, Likert, rating, ranking, checklist) with worked examples, when to use each, and how to code the data.
Questionnaire Design: Writing Survey Questions That Work
A construction guide to the questionnaire-design decisions that determine data quality: question wording, response-option and Likert scale design, question order, sensitive-question handling, pretesting, and the scale-vs-independent-items distinction.
