Section 1.310 min read

Survey Studies

Core summary

Surveys collect self-reported data from participants using questionnaires. They are the most accessible study type for beginners but are prone to response bias, social desirability bias, and low response rates if poorly designed.

Detailed explanation

Surveys are a subtype of cross-sectional studies where data collection relies on self-report. They can be paper-based, electronic (Google Forms, REDCap, SurveyMonkey), phone-based, or face-to-face interviews. Key design principles include using validated instruments when available (e.g., PHQ-9 for depression), writing clear and unambiguous questions, avoiding leading or double-barreled questions, piloting the questionnaire before full deployment, and planning for adequate response rates (typically >60% for credibility). Question formats include Likert scales (strongly agree to strongly disagree), visual analog scales, multiple choice, and open-ended text. The statistical analysis depends on the type of data collected: Likert data is ordinal (use non-parametric tests), while continuous measurements use parametric tests if normally distributed.

Clinical example

You want to assess burnout among residents at your hospital. You use the validated Maslach Burnout Inventory (MBI), distribute it electronically, and collect responses anonymously to reduce social desirability bias. You achieve a 72% response rate. Your results show that 45% of residents score high on emotional exhaustion.

Research example

Many knowledge, attitudes, and practices (KAP) studies use surveys. A KAP survey about antibiotic resistance among primary care physicians might use a self-administered questionnaire with knowledge questions (scored), attitude Likert items, and practice frequency questions. These studies are common in public health and medical education research.

Knowledge check

Q1. Why should you use a validated questionnaire when available?

Q2. What is a 'double-barreled' question?

Q3. What is the minimum response rate generally considered acceptable for survey credibility?