Survey Research: How to Create a Questionnaire Survey

Getting your Trinity Audio player ready...

Contents

• What Is Survey Research?
• Types of Surveys
• Advantages and Disadvantages of Survey Research
• Types of Questions in a Survey
• Reliability and Validity in Survey Research
• How to Design a Questionnaire: Step-by-Step
• Examples of Survey Questions
• How to Ensure Rigor in Survey Research
• Tips for Maximizing Response Rates
• Key Takeaways
• Frequently Asked Questions (FAQs)
• Glossary of Key Terms in Survey Research

What Is Survey Research?

Survey research is a systematic method of collecting data from a defined group of people, called respondents or participants, to describe, compare, or explain their knowledge, attitudes, behaviors, or outcomes. It is one of the most widely used research methods across disciplines, from public health and clinical medicine to sociology, psychology, and education.

At its core, survey research involves:

• Defining a research question or hypothesis
• Designing an instrument (questionnaire) to measure relevant variables
• Administering the instrument to a sample from a target population
• Analyzing the collected data
• Interpreting and reporting findings

Types of Surveys

By Mode of Administration

Type	Description	Example Use
Self-administered (Paper)	Respondents fill in a printed questionnaire	Patient satisfaction forms in a clinic
Online/Web-based	Delivered via platforms like SurveyMonkey, REDCap, Google Forms	Large-scale national health surveys
Telephone	Interviewer asks questions over a call	Political polling, health behavior surveillance
Face-to-face (Structured Interview)	Interviewer reads questions in person	Demographic surveys in low-literacy populations
Computer-Assisted Personal Interview (CAPI)	Tablet/laptop used during an in-person interview	Census fieldwork
Mixed-mode	Combination of two or more modes	National Health Interview Survey (USA)

By Research Design

Design	Description	Example
Cross-sectional	Data collected at a single point in time	Prevalence of depression among university students
Longitudinal/Panel	Same respondents surveyed repeatedly over time	British Cohort Study tracking health outcomes
Retrospective	Asks about past events or behaviors	Lifetime tobacco use among adults
Prospective	Follows participants forward to capture future events	Monthly surveys tracking COVID-19 symptom onset

Advantages and Disadvantages of Survey Research

Advantages

• Cost-effective: especially online surveys; reach large samples at minimal cost
• Scalability: can be administered to hundreds or thousands simultaneously
• Standardization: every participant answers the same questions, reducing interviewer bias
• Anonymity: encourages honest responses on sensitive topics (e.g., substance use, sexual behavior)
• Versatility: applicable across disciplines and topics
• Quantifiability: structured responses can be easily coded and analyzed statistically
• Wide geographical reach: online and telephone surveys cross geographical barriers

Disadvantages

• Response bias: social desirability, acquiescence, and recall bias can distort answers
• Low response rates: particularly for online surveys (often below 30%), threatening representativeness
• Inflexibility: closed-ended questions cannot capture nuance or unexpected responses
• Causality limitations: surveys establish associations, not causal relationships
• Sampling bias: non-probability or convenience samples limit generalizability
• Self-report limitations: participants may misremember, misunderstand, or misreport
• Question order effects: earlier questions can prime or influence later answers

Types of Questions in a Survey

Closed-Ended Questions

• Dichotomous: Yes / No options
  • “Have you ever been diagnosed with hypertension?”
• Multiple choice (single answer): Select one from several options
  • “What is your highest level of education?”
• Multiple response: Check all that apply
  • “Which of the following chronic conditions have you been diagnosed with? (Select all that apply)”
• Rating scales: Numerical response (e.g., 1–10)
  • “Rate your current pain level from 0 (no pain) to 10 (worst possible pain).”
• Likert scale: Agreement continuum (typically 5 or 7 points)
  • “I feel supported by my supervisor.” (Strongly Disagree → Strongly Agree)
• Semantic differential: Bipolar adjective pairs
  • “This public health campaign was: Ineffective [1-2-3-4-5] Effective”
• Rank-order: Respondents rank items by preference or priority

Open-Ended Questions

• Allow free-text responses; used to capture rich, unanticipated data
• “Please describe any barriers you face in accessing mental health services.”
• Best used sparingly; harder to analyze at scale

Measurement Scales

Scale	Properties	Example
Nominal	Categories with no order	Religion, blood type, gender
Ordinal	Categories with order but unequal intervals	Likert scales, education level
Interval	Equal intervals, no true zero	Temperature (°C), IQ score
Ratio	Equal intervals with a true zero	Age, weight, number of hospitalizations

Reliability and Validity in Survey Research

These are the two cornerstones of measurement quality. Reliability refers to consistency; validity refers to accuracy (measuring what you intend to measure).

Types of Reliability

Type	Definition	How to Assess	Example
Test-retest reliability	Consistency of scores over time when conditions are unchanged	Administer same survey twice; compute Pearson’s r or ICC	PHQ-9 depression scale re-administered 2 weeks apart
Inter-rater reliability	Agreement between two or more raters coding responses	Cohen’s Kappa, ICC	Two researchers independently coding open-ended trauma responses
Intra-rater reliability	Consistency of a single rater across time	Cohen’s Kappa comparing ratings at two time points	One clinician re-scoring the same audio-recorded interviews
Internal consistency	Extent to which items in a scale measure the same construct	Cronbach’s alpha (α ≥ 0.70 acceptable); McDonald’s omega	All items in an anxiety subscale should correlate with each other
Parallel forms reliability	Equivalence between two different versions of the same test	Pearson’s r between Form A and Form B scores	Two alternate versions of a medical knowledge quiz
Split-half reliability	Consistency between two halves of a single instrument	Spearman-Brown corrected correlation	Odd-numbered vs. even-numbered items on a self-esteem scale

Types of Validity

Type	Definition	How to Assess	Example
Face validity	Appears to measure the construct on the surface (subjective)	Expert or respondent review	Survey on loneliness “looks like” it covers loneliness
Content validity	Items adequately cover the full domain of the construct	Expert panel review; Content Validity Index (CVI)	All WHO health dimensions represented in a quality-of-life scale
Criterion validity	Scores predict or correlate with an external criterion	Pearson’s r or AUC vs. gold standard	AUDIT-C alcohol screen validated against structured clinical interview
— Concurrent validity	Correlation with a criterion measured at the same time	Concurrent correlation	New anxiety scale correlated with GAD-7
— Predictive validity	Ability to predict a future outcome	Longitudinal correlation or regression	Baseline depression score predicts rehospitalization at 6 months
Construct validity	Instrument truly measures the theoretical construct	Factor analysis, SEM, hypothesis testing	—
— Convergent validity	Correlates positively with theoretically similar measures	Multi-trait multi-method (MTMM)	Social support scale correlates with quality-of-life scale
— Discriminant validity	Does not correlate with theoretically unrelated measures	MTMM, AVE > MSV	Anxiety scale should NOT strongly correlate with cognitive ability
Structural validity	Factor structure matches theoretical model	Confirmatory Factor Analysis (CFA)	Single-factor structure of PHQ-9 confirmed in a new population
Cross-cultural validity	Instrument performs equivalently across cultural/language groups	Measurement invariance testing (CFA)	SF-36 translated and validated in Hindi for Indian populations
Ecological validity	Findings generalize to real-world settings	Expert judgment; comparison with field data	Lab-based stress survey reflects real occupational stress
Internal validity	Extent to which observed relationships are causal (more applicable to experimental designs)	Research design controls	Ruling out confounders in a pre-post survey intervention study
External validity	Generalizability of findings to other populations, settings, or times	Sampling strategy, replication	National survey results applicable to state-level population

How to Design a Questionnaire: Step-by-Step

Step 1: Define Your Research Objectives

Clearly specify what you want to measure before writing a single question.

• Social science example: “To assess the association between perceived social support and academic burnout among postgraduate students.”
• Biomedical example: “To measure adherence to antiretroviral therapy (ART) and identify its predictors among HIV-positive adults in a tertiary hospital.”

Step 2: Identify Your Target Population and Sample

• Define inclusion/exclusion criteria
• Choose a sampling strategy:
  • Probability: Simple random, stratified, cluster, systematic
  • Non-probability: Convenience, purposive, snowball, quota

Step 3: Decide on the Mode of Administration

Consider your population’s literacy level, internet access, sensitivity of the topic, and available resources.

Step 4: Select or Develop Your Instrument

Option A: Use a validated existing scale

Always prefer validated instruments when available. Here are some popular ones in research:

Scale	Domain	Used in
PHQ-9	Depression	Primary care, epidemiology
GAD-7	Generalized anxiety	Mental health, clinical settings
AUDIT	Alcohol use disorder	Public health, addiction medicine
SF-36 / SF-12	Health-related quality of life	Biomedical, chronic disease research
Perceived Stress Scale (PSS)	Psychological stress	Social and health sciences
Rosenberg Self-Esteem Scale	Self-esteem	Developmental and social psychology
Morisky Medication Adherence Scale (MMAS-8)	Medication adherence	Clinical and pharmacoepidemiology research

Option B: Develop new items

If no validated tool exists:

• Conduct a literature review to identify the construct’s dimensions
• Generate an item pool (3–5 items per dimension initially)
• Conduct expert review for content validity
• Pilot test on a small subsample
• Analyze psychometric properties

Step 5: Write Clear, Unbiased Questions

Rules for writing good survey questions:

• ✅ Use simple, everyday language
• ✅ Ask one thing per question (avoid double-barreled questions)
• ✅ Be specific about time frames (“In the past 30 days…”)
• ✅ Provide mutually exclusive and exhaustive response options
• ✅ Match the scale to the construct
• ❌ Avoid leading questions (“Don’t you agree that…”)
• ❌ Avoid loaded/emotionally charged language
• ❌ Avoid double negatives (“Do you not disagree…”)
• ❌ Avoid jargon or technical terms unfamiliar to respondents

Step 6: Structure the Questionnaire

A well-organized questionnaire follows this general flow:

1. Title and brief introduction: state the purpose, confidentiality assurance, and estimated time
2. Consent section (if applicable)
3. Screening questions (if needed)
4. Sociodemographic/background items: age, sex, education, occupation
5. Core study variables: primary outcome and predictor measures
6. Sensitive items: place later once rapport is established
7. Open-ended items: place at the end; they require more effort
8. Closing/thank-you statement

Step 7: Pilot Test the Questionnaire

• Administer to a small group (typically n = 10–30) similar to your target population
• Assess: clarity, comprehension, time to complete, missing responses
• Revise problematic items
• Assess internal consistency (Cronbach’s alpha) for multi-item scales

Step 8: Administer the Survey

• Obtain ethical approval/IRB clearance if required
• Train interviewers if using face-to-face or telephone modes
• Track response rates; follow up with non-respondents
• Monitor data quality in real time (for online surveys)

Step 9: Manage and Analyze Data

• Establish a data management plan before collection begins
• Code and enter data (or export from online platforms)
• Check for missing data, outliers, and entry errors
• Conduct descriptive statistics, then inferential analyses per your objectives

Examples of Survey Questions

Social Science Example: Academic Burnout Study

Background variable:

“What is your current year of postgraduate study?” ○ First year ○ Second year ○ Third year ○ Fourth year or above

Perceived Social Support (adapted from MSPSS):

“Please indicate how much you agree with each statement.” (1 = Strongly Disagree, 7 = Strongly Agree)

Item	1	2	3	4	5	6	7
There is a special person who is around when I am in need	○	○	○	○	○	○	○
My family really tries to help me	○	○	○	○	○	○	○
I can count on my friends when things go wrong	○	○	○	○	○	○	○

Burnout (adapted from MBI-GS):

“How often do you experience the following? (0 = Never, 6 = Every day)”

Item	0	1	2	3	4	5	6
I feel emotionally drained by my studies	○	○	○	○	○	○	○
I feel used up at the end of a study day	○	○	○	○	○	○	○

Biomedical Science Example: ART Adherence Study

Clinical background:

“How long have you been on antiretroviral therapy (ART)?” ○ Less than 6 months ○ 6–12 months ○ 1–3 years ○ More than 3 years

Adherence measure (adapted from MMAS-8):

“Please answer the following about your ART medications.”

Item	Yes	No
Do you sometimes forget to take your HIV medicine?	○	○
Over the past two weeks, were there any days when you did not take your HIV medicine?	○	○
Have you ever cut back or stopped taking your medicine without telling your doctor because you felt worse when you took it?	○	○

Barrier assessment (open-ended):

“What are the main reasons, if any, that you have difficulty taking your HIV medications as prescribed? Please describe:”

How to Ensure Rigor in Survey Research

Sampling Rigor

• Use probability sampling wherever feasible to ensure representativeness
• Calculate adequate sample size a priori using power analysis (e.g., G*Power, OpenEpi)
• Report response rates and compare respondents vs. non-respondents on key variables

Instrument Rigor

• Use validated, culturally adapted instruments
• Report Cronbach’s alpha or McDonald’s omega for all multi-item scales
• Conduct confirmatory factor analysis (CFA) to test structural validity
• For translated tools, conduct forward-backward translation and test measurement invariance

Data Collection Rigor

• Standardize administration procedures with a protocol/manual
• Train and certify all data collectors
• Use double data entry or built-in logic checks (for electronic surveys)
• Monitor for satisficing (respondents giving careless answers): embed attention checks

Reporting Rigor

• Follow established reporting guidelines:
  • CHERRIES checklist: for online surveys
  • STROBE: for observational studies including surveys
  • COSMIN: for studies validating patient-reported outcome measures
• Report: sampling frame, recruitment method, response rate, missing data handling, and analytic approach

Minimizing Common Biases

Bias	Description	Mitigation Strategy
Social desirability bias	Responding in a socially acceptable way	Anonymity, indirect/projective questions
Acquiescence bias	Tendency to agree regardless of content	Include reverse-scored items
Recall bias	Inaccurate memory of past events	Shorter recall periods; event anchoring
Non-response bias	Systematic difference between responders and non-responders	Follow-up of non-respondents; compare on known variables
Order/context effects	Earlier items influence later responses	Randomize item/block order
Interviewer bias	Interviewer behavior shapes responses	Standardized scripts; blind interviewing
Sampling bias	Sample not representative of population	Probability sampling; weighting

Tips for Maximizing Response Rates

• Keep the survey as short as possible (aim for ≤15 minutes)
• Send a pre-notification before the survey arrives
• Use personalized invitations rather than generic mass emails
• Include a brief, compelling rationale for participation
• Assure confidentiality and data protection clearly
• Offer reminders at 1 week and 2 weeks post-invitation
• Consider incentives (monetary, vouchers, entry into a prize draw) where ethical
• Optimize for mobile devices for online surveys
• Use a trusted sender (institutional email rather than a commercial survey platform address)

Key Takeaways

• Survey research is a versatile, cost-effective method for collecting data from large samples on knowledge, attitudes, behaviors, or outcomes.
• Surveys can be cross-sectional, longitudinal, retrospective, or prospective, and administered in multiple modes (online, face-to-face, telephone).
• Questionnaire design requires clear objectives, appropriate sampling, well-written questions, and a logical structure before any data collection begins.
• Use validated instruments from the literature whenever available; develop new tools only when necessary, following rigorous psychometric procedures.
• Reliability (test-retest, internal consistency, inter-rater) and validity (content, criterion, construct) are both essential and must be reported.
• Pilot testing is non-negotiable: it catches ambiguous questions, technical glitches, and respondent burden issues.
• Major biases (social desirability, acquiescence, recall, non-response) can be anticipated and mitigated through thoughtful design and administration.
• Response rates should be maximized through pre-notification, reminders, brevity, and appropriate incentives.
• Reporting should follow established checklists (CHERRIES, STROBE, COSMIN) to ensure transparency and reproducibility.
• Sample size should be determined a priori based on power analysis, not convenience.

Frequently Asked Questions (FAQs)

What is the difference between a questionnaire and a survey?

A questionnaire is the data collection instrument: the set of questions. A survey is the broader research process that includes the questionnaire, sampling, administration, and analysis. A survey uses a questionnaire, but a questionnaire is not itself a survey.

How many response options should a Likert scale have: 5 or 7?

Both are commonly used. A 5-point scale is simpler and works well when cognitive burden is a concern or populations have lower literacy. A 7-point scale offers finer discrimination and may be more sensitive to change over time. Research generally shows 5- and 7-point scales perform similarly in terms of reliability and validity. Avoid even-numbered scales (which force a choice) unless you have a specific reason to eliminate a neutral midpoint.

When should I use skip logic (conditional branching) in a survey?

Use skip logic when certain questions are only relevant to a subset of respondents. For example: “Have you ever smoked cigarettes? (Yes/No) → If Yes: At what age did you start smoking?” Skip logic reduces respondent burden, minimizes irrelevant questions, and improves data quality. It is easily implemented in platforms like REDCap, Qualtrics, and SurveyMonkey.

What is the minimum acceptable Cronbach’s alpha for a scale?

The conventionally accepted threshold is α ≥ 0.70 for research purposes, and α ≥ 0.90 for clinical decision-making tools. However, alpha is sensitive to the number of items: longer scales artificially inflate it. Consider also reporting McDonald’s omega (ω), which is a more robust and less biased estimate of internal consistency, particularly when scale items have unequal factor loadings.

What is the difference between anonymous and confidential surveys?

• Anonymous: No identifying information is collected at all; it is impossible to link responses to individuals.
• Confidential: Identifying information may be collected (e.g., for follow-up), but the researcher commits to keeping it separate from responses and not disclosing it. Anonymity is stronger protection and tends to elicit more honest responses on sensitive topics. Confidentiality is necessary in longitudinal studies where participants must be tracked over time.

How do I handle missing data in survey research?

First, distinguish between missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR): the mechanism affects which approach is appropriate. Common strategies include:

• Listwise deletion: simple but can introduce bias and reduce power
• Mean/median imputation: easy but underestimates variance
• Multiple imputation (MI): preferred for MAR; generates several plausible datasets and pools results
• Full information maximum likelihood (FIML): preferred in SEM contexts Always report the extent and pattern of missing data transparently.

Can survey research be used to establish causality?

Traditional cross-sectional surveys cannot establish causality: they can only demonstrate association. To strengthen causal inference with survey data, researchers can use: longitudinal/panel designs (establishing temporal precedence), natural experiments with survey data, or propensity score matching to control for confounders. Ultimately, experimental designs (RCTs) remain the gold standard for causal claims, but well-designed longitudinal surveys contribute substantially to causal evidence in fields like epidemiology and sociology.

What ethical considerations apply to survey research involving sensitive topics?

When surveying sensitive topics (mental health, substance use, sexuality, trauma, chronic illness):

• Obtain informed consent clearly explaining the nature of questions before respondents begin
• Provide trigger warnings if content may be distressing
• Include resource signposting (e.g., helpline numbers) at the end of the survey
• Ensure data security: encrypt data, use secure platforms (REDCap is preferred in health research)
• Obtain IRB/ethics committee approval prior to data collection
• Allow respondents to skip any question without penalty
• Debrief participants after completion, especially in psychological research

Glossary of Key Terms in Survey Research

Term	Definition
Acquiescence Bias	The tendency of respondents to agree with survey statements regardless of their actual views. Mitigated by including reverse-scored items in scales.
Alpha (Cronbach’s α)	A coefficient (ranging 0–1) measuring the internal consistency of a multi-item scale. Values ≥ 0.70 are generally considered acceptable for research purposes.
Anonymity	A data collection condition in which no identifying information is collected, making it impossible to link responses to specific individuals.
Attrition Bias	In longitudinal surveys, the systematic dropout of participants over time, which can skew results if those who leave differ from those who remain.
Bipolar Scale	A rating scale anchored at both ends by opposing adjectives or statements (e.g., Ineffective ↔ Effective). Also called a semantic differential scale.
Branching / Skip Logic	Conditional routing in a questionnaire that directs respondents to different questions based on their previous answers, reducing respondent burden and irrelevant questions.
CAPI	Computer-Assisted Personal Interviewing — a data collection method in which an interviewer uses a tablet or laptop to administer a structured questionnaire face-to-face.
CHERRIES Checklist	A reporting guideline specifically designed for internet-based surveys, ensuring transparency in design, sampling, response rates, and analysis.
Closed-Ended Question	A question with a predefined set of response options (e.g., Yes/No, multiple choice, Likert scale). Easier to analyze than open-ended questions.
Cluster Sampling	A probability sampling method in which the population is divided into natural groups (clusters) and whole clusters are randomly selected. Useful for geographically dispersed populations.
Confidentiality	A researcher’s commitment not to disclose identifiable participant information, even when such information has been collected.
Construct	A theoretical concept or attribute being measured (e.g., depression, social support, medication adherence).
Construct Validity	The degree to which an instrument truly measures the theoretical construct it purports to measure. Assessed through convergent and discriminant validity, and factor analysis.
Content Validity	The extent to which a questionnaire’s items adequately cover all dimensions of the construct being measured. Assessed by expert panel review and the Content Validity Index (CVI).
Content Validity Index (CVI)	A quantitative measure of content validity, calculated as the proportion of experts rating each item as relevant. Item-CVI ≥ 0.78 and Scale-CVI ≥ 0.90 are standard thresholds.
Convergent Validity	Evidence that an instrument correlates positively and substantially with other measures of theoretically similar constructs.
COSMIN	COnsensus-based Standards for the selection of health Measurement INstruments — a methodological framework and checklist for evaluating the quality of patient-reported outcome measures (PROMs).
Cross-Sectional Survey	A survey design that collects data from a sample at a single point in time; useful for estimating prevalence but cannot establish temporal relationships.
Cross-Cultural Validity	Evidence that an instrument performs equivalently across different cultural, linguistic, or demographic groups. Tested through measurement invariance analysis.
Data Saturation	In mixed-methods or qualitative survey analysis, the point at which new responses no longer yield new themes or insights.
Dichotomous Question	A closed-ended question with only two response options (e.g., Yes / No; True / False).
Discriminant Validity	Evidence that an instrument does not correlate strongly with measures of theoretically unrelated constructs, demonstrating specificity.
Double-Barreled Question	A flawed question that asks about two separate issues simultaneously, making it impossible to know which issue the response addresses. Example: “Do you find the course content interesting and the teaching effective?”
Ecological Validity	The extent to which survey findings reflect real-world conditions and can be generalized to naturalistic settings outside the study context.
External Validity	The degree to which survey findings can be generalized beyond the study sample to other populations, settings, or time periods.
Face Validity	The superficial, subjective appearance that an instrument measures what it is intended to measure. The weakest form of validity; assessed informally by experts or respondents.
Exploratory Factor Analysis (EFA)	A statistical technique used when the factor structure is unknown, to identify how items cluster into underlying dimensions.
Confirmatory Factor Analysis (CFA)	A statistical technique used to test whether a hypothesized factor structure fits an observed dataset.
FIML	Full Information Maximum Likelihood — a method for handling missing data that uses all available data without imputation, producing unbiased estimates under the MAR assumption.
GAD-7	The Generalized Anxiety Disorder 7-item scale; a widely validated self-report instrument for screening and measuring anxiety severity.
Gold Standard	In criterion validity testing, an established, highly accurate measure against which a new instrument is compared.
ICC (Intraclass Correlation Coefficient)	A reliability statistic used to assess test-retest and inter-rater reliability, particularly for continuous data. Values > 0.75 are generally considered good.
Incentive	A reward (monetary, gift voucher, prize draw entry) offered to encourage survey participation. Must be proportionate and not coercive.
Internal Consistency	The degree to which all items within a scale measure the same underlying construct. Typically assessed with Cronbach’s alpha or McDonald’s omega.
Internal Validity	The extent to which observed associations within a study reflect true relationships rather than confounding variables or methodological artifacts.
Inter-Rater Reliability	The degree of agreement between two or more independent raters when coding or scoring the same data. Assessed with Cohen’s Kappa or ICC.
Interval Scale	A measurement scale with equal distances between values but no true zero point (e.g., temperature in Celsius, IQ scores).
Intra-Rater Reliability	The consistency of a single rater’s judgments when scoring the same material on two separate occasions.
IRB	Institutional Review Board — an ethics committee that reviews and approves research involving human participants to ensure it meets ethical standards. Equivalent to a Research Ethics Committee (REC) in the UK/India.
Kappa (Cohen’s Kappa)	A statistic measuring inter-rater or intra-rater agreement, corrected for chance. Values: < 0.20 = slight; 0.21–0.40 = fair; 0.41–0.60 = moderate; 0.61–0.80 = substantial; > 0.80 = almost perfect.
Leading Question	A question phrased in a way that implies a desired answer, introducing bias. Example: “Don’t you agree that exercise is important for health?”
Likert Scale	An ordinal rating scale — typically 5 or 7 points — measuring degree of agreement or disagreement with a statement. Named after psychologist Rensis Likert.
Longitudinal Survey	A survey design that follows the same participants over time, allowing assessment of change, trends, and temporal relationships between variables.
MCAR	Missing Completely at Random — a missing data pattern in which the probability of missingness is unrelated to any observed or unobserved variable. The least problematic type.
MAR	Missing at Random — a missing data pattern in which the probability of missingness is related to observed (but not missing) data. Amenable to imputation methods.
MNAR	Missing Not at Random — a missing data pattern in which the probability of missingness is related to the missing values themselves. The most problematic type.
McDonald’s Omega (ω)	A reliability coefficient that, unlike Cronbach’s alpha, does not assume equal factor loadings, making it a more robust measure of internal consistency.
Measurement Invariance	A property indicating that an instrument measures the same construct in the same way across different groups. Tested via multi-group CFA.
Mixed-Mode Survey	A survey that uses two or more administration modes (e.g., online and telephone) to maximize coverage and response rates.
MMAS-8	Morisky Medication Adherence Scale (8-item) — a self-report instrument widely used in clinical research to assess adherence to prescribed medications.
MSPSS	Multidimensional Scale of Perceived Social Support — a 12-item validated scale measuring perceived support from family, friends, and a significant other.
Multiple Imputation (MI)	A statistical method for handling missing data in which multiple plausible datasets are generated, analyzed separately, and pooled using Rubin’s rules.
Nominal Scale	The most basic measurement level; classifies data into discrete, unordered categories with no numerical meaning (e.g., blood type, religion, ethnicity).
Non-Probability Sampling	A sampling approach in which participants are not selected through random procedures, limiting generalizability. Includes convenience, purposive, snowball, and quota sampling.
Non-Response Bias	A systematic distortion in results that occurs when individuals who do not respond differ meaningfully from those who do.
Open-Ended Question	A question that allows respondents to answer in their own words, generating qualitative data. Useful for capturing nuance but harder to analyze at scale.
Operationalization	The process of defining how an abstract construct (e.g., “depression,” “social cohesion”) will be concretely measured in a survey.
Ordinal Scale	A measurement scale in which categories have a meaningful order but intervals between them are not necessarily equal (e.g., Likert scales, education level).
Panel Survey	A longitudinal design in which the same respondents are surveyed repeatedly over time, enabling analysis of change at the individual level.
Parallel Forms Reliability	The consistency of scores between two different but equivalent versions of the same instrument, administered to the same group.
PHQ-9	Patient Health Questionnaire-9 — a 9-item validated self-report instrument for screening and measuring depression severity, widely used in primary care and epidemiology.
Pilot Test	A small-scale preliminary administration of a questionnaire to identify problems with clarity, length, response options, or technical functioning before full deployment.
Population	The entire group of individuals to whom a researcher wishes to generalize findings. Surveys collect data from a sample drawn from this population.
Power Analysis	A statistical procedure used to determine the minimum sample size needed to detect an effect of a given size with a specified level of confidence (typically 80% power at α = 0.05).
Predictive Validity	A subtype of criterion validity; the extent to which scores on an instrument predict a future outcome or criterion.
PROM	Patient-Reported Outcome Measure — any measure of a patient’s health status reported directly by the patient, without interpretation by a clinician.
Purposive Sampling	A non-probability sampling strategy in which participants are selected deliberately based on specific characteristics relevant to the research question.
Questionnaire	The instrument — the actual set of questions — used to collect data in survey research. Distinct from the broader survey process itself.
Random Sampling	A probability-based sampling method in which every member of the population has a known, non-zero chance of being selected, minimizing selection bias.
Ratio Scale	The highest measurement level; has equal intervals and a true, meaningful zero point (e.g., age, weight, number of hospital admissions).
Recall Bias	Systematic error arising from respondents’ inaccurate or incomplete memory of past events or behaviors, particularly problematic in retrospective surveys.
REDCap	Research Electronic Data Capture — a secure, web-based platform widely used in biomedical and social research for survey administration and data management.
Reliability	The consistency or stability of a measurement instrument across time, raters, or items. A necessary but not sufficient condition for validity.
Response Rate	The proportion of sampled individuals who complete and return a survey. Calculated as: (Completed responses ÷ Eligible individuals contacted) × 100.
Reverse-Scored Item	A questionnaire item worded in the opposite direction of most items in a scale, used to detect acquiescence bias and ensure respondents read carefully.
Sample	A subset of individuals drawn from a target population and actually surveyed. Findings are generalized from the sample to the population.
Sampling Frame	The complete list or database from which a sample is drawn (e.g., a hospital patient registry, university enrollment list, electoral roll).
Satisficing	A response behavior in which respondents invest minimal cognitive effort, selecting the first plausible answer or consistently choosing the same response option.
Scale	A composite measure formed by combining scores from multiple related items to create a single score representing an underlying construct.
Semantic Differential Scale	A rating scale anchored by two bipolar adjectives at each end, used to measure attitudes or perceptions (e.g., Ineffective [1–2–3–4–5] Effective).
SF-36 / SF-12	Short Form-36 / Short Form-12 — widely validated instruments measuring health-related quality of life across eight domains, used extensively in biomedical research.
Simple Random Sampling	A probability sampling method in which every individual in the sampling frame has an equal probability of selection, typically via random number generation.
Snowball Sampling	A non-probability technique in which existing participants recruit further participants from their networks; useful for hard-to-reach populations.
Social Desirability Bias	The tendency of respondents to answer in ways they believe are socially acceptable rather than truthfully.
Split-Half Reliability	A measure of internal consistency obtained by dividing a scale into two halves and correlating their scores, corrected using the Spearman-Brown formula.
STROBE	STrengthening the Reporting of OBservational studies in Epidemiology — a reporting checklist applicable to observational research including surveys.
Stratified Sampling	A probability sampling method in which the population is divided into subgroups (strata) and random samples are drawn from each, ensuring representation.
Structural Validity	Evidence that the factor structure of an instrument matches its theoretical model, assessed via confirmatory factor analysis (CFA).
Survey	The complete research process of collecting, analyzing, and interpreting data from a sample using a questionnaire. Encompasses instrument design, sampling, administration, and analysis.
Systematic Sampling	A probability sampling method in which every kth individual from a sampling frame is selected (e.g., every 10th patient from a clinic register).
Target Population	The specific group to whom the researcher intends to generalize findings (e.g., “adults aged 18–65 living with Type 2 diabetes in urban Maharashtra”).
Test-Retest Reliability	The stability of scores from the same instrument administered to the same participants on two separate occasions. Assessed with Pearson’s r or ICC.
Triangulation	The use of multiple methods, data sources, or investigators to cross-validate findings and strengthen the overall rigor of a study.
Validity	The degree to which an instrument measures what it is intended to measure. Encompasses multiple subtypes including content, criterion, and construct validity.
Variable	Any characteristic or attribute that can vary across individuals or observations. Includes outcome variables (dependent) and predictor variables (independent).
Weighting	A statistical adjustment applied to survey data to correct for unequal probabilities of selection or differential non-response, improving representativeness of estimates.

---

Terms are listed alphabetically. Definitions are contextualized for survey and questionnaire research.