Skip to content
Back to search results

Online Survey — Web-Based Survey Research

Introduced by
Mick P. Couper, Don A. Dillman (early systematic frameworks)
Year
Mid-1990s (widespread scholarly adoption ~1995–2000)
Type
Quantitative / mixed-methods data collection technique
Family
Data collection
Font size

What does it do?

An online survey is a structured data collection instrument hosted on a web platform and completed by respondents via internet-connected devices.

It enables large-scale, geographically dispersed data gathering at low cost and with rapid turnaround. Respondents self-administer the questionnaire at their convenience, which reduces interviewer bias and permits automatic data capture. Online surveys are the dominant mode of survey research in social, behavioural, health, and market research today.

The intuition behind the method

Think of an online survey as a printed questionnaire moved to a web page, but with far richer possibilities: skip logic routes respondents to different questions based on earlier answers, multimedia elements can be embedded, and responses flow directly into a database without manual entry. The trade-off is that the researcher cannot verify who is responding or observe the context in which the survey is completed. Designing a good online survey therefore requires careful attention to question wording, page flow, and the motivations of the target population to participate honestly and completely.

History

Web-based surveying emerged in the mid-1990s as internet penetration grew in academic and corporate settings. Mick Couper's 2000 review in Public Opinion Quarterly provided the first systematic scholarly assessment of web survey methodology and identified its coverage and non-response challenges. Don Dillman's Tailored Design Method, originally developed for mail surveys, was progressively adapted across editions (1978, 2000, 2007, 2014) to incorporate internet and mixed-mode protocols. By the 2010s online surveys had overtaken telephone and mail as the dominant survey mode in social and market research.

When to use it

Use an online survey when you need to collect standardised self-report data from a large or geographically dispersed sample quickly and cost-effectively, and when the research variables can be operationalised as closed-ended items. It is the preferred mode for measuring attitudes, opinions, prevalence, or self-reported behaviours in social, health, organisational, and market research. Do not use an online survey when the research question requires in-depth exploration of meaning or process (consider phenomenology, grounded theory, or semi-structured interviews instead), when the target population has low internet access or literacy (telephone or face-to-face survey is more appropriate), or when the constructs of interest cannot be adequately captured by self-report items.

Limitations and considerations

  • Coverage bias: individuals without internet access or low digital literacy are systematically excluded, limiting representativeness.
  • Response rates for unsolicited online surveys are typically low (often below 30%), which can introduce non-response bias.
  • The researcher cannot verify respondent identity or the conditions under which the survey was completed.
  • Self-report data are subject to recall bias, social desirability bias, and acquiescence bias regardless of the online mode.
  • Open-text responses require additional qualitative coding effort and are often left blank by a majority of respondents.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Reaches large, geographically dispersed samples rapidly and at low marginal cost per respondent.
  • Eliminates interviewer effects: respondents answer privately, reducing social desirability bias on sensitive topics.
  • Automatic data capture removes transcription error and enables real-time monitoring of response rates.
  • Flexible formatting: skip logic, randomised item order, embedded media, and progress bars improve data quality and respondent experience.
  • Quantitative data are immediately available in structured format, shortening the analysis timeline.

Applications

  • Measuring employee attitudes, engagement, and organisational climate in workplace research.
  • Assessing patient-reported outcomes and health behaviours in public health and clinical research.
  • Collecting customer satisfaction, product feedback, and market segmentation data in commercial research.
  • Conducting large-scale educational research on student experiences, academic achievement correlates, and instructional preferences.
  • Gathering cross-national attitudinal data in comparative political science and sociology studies.
  • Screening potential participants for experimental or clinical studies using eligibility criteria items.

How it works

How it works

1

The researcher defines the research objectives and constructs a questionnaire with closed-ended items (Likert scales, multiple choice, rankings) and, where needed, open-text fields. Each item is mapped to a variable; skip logic, mandatory fields, and response validation are configured in the survey platform. Attention to visual layout, mobile responsiveness, and estimated completion time (ideally under 15 minutes) is essential at this stage.

2

Before full deployment the questionnaire is tested with a small convenience sample (typically 5–20 people) representative of the target population. Pilot testing checks for ambiguous wording, broken skip logic, technical display issues across devices and browsers, and unusually long completion times. Feedback is used to revise items before the instrument is finalised.

3

The live survey link or embed code is distributed to the target sample via email invitation, social media, panel service, institutional mailing list, or embedded in a website. Distribution may use probability sampling (random selection from a defined frame) or non-probability sampling (convenience, snowball, quota). Personalised invitation emails with a clear subject line and incentive description improve initial response rates.

4

Response rates and partial completions are tracked in real time through the platform's dashboard. Follow-up reminders are sent to non-respondents at planned intervals (e.g., one week after initial invite). The researcher monitors for speeders (implausibly short completion times), straight-liners (identical responses across all items), and bot submissions, removing invalid cases as they are identified.

5

The survey is closed once the target sample size or deadline is reached. Response data are exported in structured format (CSV, SPSS, Excel) for cleaning and analysis. Cleaning steps include removing incomplete responses below an acceptable threshold, recoding reverse-scored items, and checking for multivariate outliers before proceeding to descriptive or inferential analysis.

Common pitfalls

  • Questionnaire too long: surveys exceeding 15–20 minutes see sharply increased dropout rates; trim items ruthlessly to only those required to answer the research question.
  • No pilot test: skipping piloting allows ambiguous items, broken skip logic, and mobile display errors to survive into the live instrument, corrupting the dataset.
  • Non-probability sampling presented as representative: convenience or snowball online samples cannot support generalisation to a defined population without explicit caveats; avoid language implying representativeness unless a probability panel was used.
  • Ignoring data quality checks: failure to screen for speeders, straight-liners, and bot responses inflates sample size artificially while degrading data quality.
  • Leading or double-barrelled items: poor item construction introduces systematic response bias that no analytical technique can fully correct.

Frequently asked questions

What sample size do I need for an online survey?

Sample size depends on the planned analysis. For descriptive statistics with a known population, standard power tables or margin-of-error formulas apply (e.g., n ≈ 385 for ±5% margin at 95% confidence in a large population). For regression or SEM analyses, common rules of thumb suggest at least 10–20 cases per predictor variable. Account for expected response rates and attrition by over-recruiting: if you expect a 30% response rate, invite roughly three times your target n.

How can I improve my response rate?

Key evidence-based strategies include: personalised invitation emails with a credible sender, a clear and short subject line, a brief explanation of the study's purpose and how long it takes, an explicit confidentiality statement, a visible deadline, and one or two polite reminders to non-respondents. Tangible incentives (vouchers, prize draws) improve rates but may attract low-effort responses; weigh this trade-off for your study context.

How do I handle missing data in an online survey?

Distinguish between item-level missing (a few items skipped within an otherwise complete response) and unit-level missing (the entire survey not submitted). For item-level missing, consider multiple imputation or full-information maximum likelihood if the data are missing at random. For unit-level missing, conduct non-response bias analysis by comparing early and late respondents or comparing available demographic variables to the target population benchmarks.

Is an online survey always quantitative?

No. Online surveys can include open-text fields that yield qualitative data, and some surveys are designed as entirely qualitative instruments. However, the dominant use case is quantitative or mixed-methods, using Likert scales, multiple-choice, and ranking items whose responses are coded numerically and analysed with descriptive or inferential statistics.

Can I use the same questionnaire online that I designed for paper administration?

Not without review. Paper questionnaires often lack skip logic, assume linear completion, and are formatted for print rather than screen. Migrating to online administration requires adding conditional display logic, checking mobile rendering, confirming that response categories display correctly across screen sizes, and re-piloting the instrument. Measurement equivalence across modes should be tested if combining online and paper data.

Related methods

Sources

  1. Couper, M. P. (2000). Web surveys: A review of issues and approaches. Public Opinion Quarterly, 64(4), 464–494.

    Open DOI reference
    Copied.
  2. Dillman, D. A., Smyth, J. D., & Christian, L. M. (2014). Internet, Phone, Mail, and Mixed-Mode Surveys: The Tailored Design Method (4th ed.). Wiley.

    ISBN: 978-1118456149
    Copied.