Short answers, pulled from the story.
A survey in human research is a predetermined set of questions given to a sample of individuals to extract specific data that can be generalized to a larger population. Surveys are used across social science, medicine, marketing, and official statistics to assess opinions, behaviors, and factual information.
The Statistical Society of London used the first recorded written questionnaire in 1838, when a committee printed questions designed to document the history of strikes. The application of probability sampling in the 1930s later made surveys a standard empirical research tool.
Surveys are conducted in person, by telephone, by mail, and online. Common modes include computer-assisted personal interviewing (CAPI), computer-assisted telephone interviewing (CATI), mail questionnaires, web surveys, mobile surveys, and SMS surveys. Over 50% of surveys today are opened on mobile devices.
Probability-based web surveys and face-to-face surveys with show-cards show similar levels of measurement quality; telephone surveys have been found to perform worse. Online surveys offer faster data collection, often completing in little more than a month, but suffer from low and highly variable response rates.
Online survey response rates range from less than 1% in broad enterprise email surveys to nearly 100% in targeted membership surveys. Non-coverage (not all people have internet access), acquiescence bias, and sampling frame problems are the main structural challenges researchers must address.
The link between self-reported behavior and actual behavior, while positive, is not always strong. Dishonesty is especially pronounced in sex-related questions, where men tend to inflate reported partner counts and women tend to understate them. Researchers recommend comparing verbal reports against direct observation where possible.