Surveys in clinical research: methodological aspects and practical guidance

Bianchini, Larissa; Pereira, Aline Braz; Tomazini, Bruno Martins; Righy, Cássia; Maia, Israel Silva; Ramos, João Gabriel Rosa; Rosa, Regis Goulart; Roepke, Roberta Muriel Longo; Ferreira, Juliana Carvalho; Besen, Bruno Adler Maccagnan Pinheiro

doi:10.62675/2965-2774.20260322

Tabla de contenido Anterior Actual Siguiente

Tabla de contenido

Narrative Review • Crit. Care Sci. 38 • 2026 • https://doi.org/10.62675/2965-2774.20260322 linkcopy

Surveys in clinical research: methodological aspects and practical guidance

Authorship SCIMAGO INSTITUTIONS RANKINGS

Use of surveys in clinical research allows investigators to explore stakeholders’ perspectives, measure implementation of interventions, and inform future decision-making. Surveys are versatile and accessible, but they require methodological rigor to yield adequate results. Their development involves a sequence of decisions that influence both data quality and interpretability - from defining objectives and selecting a sample to designing the questionnaire and choosing the method of administration. Questionnaires must balance clarity with precision, capturing relevant constructs without overburdening respondents. In many cases, frameworks such as Knowledge, Attitude, and Practice questionnaires are employed to structure questions and explore relationships between what individuals know, believe, and do. Online platforms increase the ability to disseminate surveys to a broader, more diverse target population and to improve data-collection workflows. Beyond the technical aspects, using surveys for clinical research faces practical challenges, such as low response rates and variability in engagement across formats. Addressing these issues requires planning and the use of strategies to encourage participation without compromising data quality. This review offers a practical overview intended to guide researchers in designing and conducting survey studies in clinical research.

Keywords:
Survey; Questionnaire; Survey design

Do
Know the denominator of your sampling frame to allow the calculation of non-response rates
Avoid confusing or ambiguous questions. You will not be able to analyze and interpret these results later
Remember to consider cultural characteristics and social biases. Avoid social desirability bias
Remember that less is more during item generation and reduction! Do consider using dummy tables to visualize the end product of your survey
Pretest your survey after item generation and reduction. It will save you time and effort in the future
Pilot test your survey. It will avoid deployment of a survey with technical issues
Don't
Deploy a survey with unclear research goals. Do define if your goal is to describe, predict or make inferences
Start an online survey if you can't have a clear sampling plan. Sampling errors will lend your results subject to much criticism
Interpret what survey respondents reported to do as an actual observed behavior or attitude. Cognitive biases may influence how humans respond to surveys
Report only interesting findings. Follow your initial research plan to avoid selective reporting
Deploy surveys without ethical approval or consultation of its need with research Ethics Committees

Term	Definition
Acquiescence	A response bias in which individuals tend to agree with survey statements without carefully considering their content, often leading to inflated correlations
Acronym BRUSO	Acronym highlighting key principles for writing effective survey questions: "brief", "relevant", "unambiguous", "specific", and "objective"
"Ceiling" or "floor" effect	Occur when survey responses cluster at the top ("ceiling") or bottom ("floor") of a scale, thereby reducing data variability
Close-ended questions	Questions that offer respondents a fixed set of answer options to choose from
Construct validity	How well a survey or test measures the theoretical concept or construct it is intended to assess
Content validity	The extent to which a survey or test fully represents the entire domain of the concept being measured
Coverage error	Occurs when some members of the target population are not included in or have no chance of being selected from the survey sampling frame
Criterion validity	How a measure correlates with an external standard or outcome (the criterion), indicating how accurately it predicts or reflects real-world performance
Don't know answer	A response option that allows participants to indicate they have no opinion, are unsure, or lack sufficient information to answer a question
Double-barreled question (combined questions)	A survey item that asks about two or more issues within a single question, making it unclear which part the respondent is answering
Measurement error	The difference between the true value of what is being measured and the value obtained by the survey or instrument, caused by flaws in question design, respondent interpretation, or data collection
Negative worded items	Survey questions phrased with negation or in a negative form, which can be harder for respondents to interpret. They should be avoided, unless they are explicitly part of the research question (e.g., to evaluate framing effects)
Nonresponse error	When individuals selected for a survey do not participate or fail to answer certain questions, the results can be biased if the nonrespondents differ systematically from respondents
Open form question	Allows respondents to answer in their own words, providing detailed information but requiring more effort to analyze and code, usually with qualitative synthesis methods
Pilot test	Before administering a survey, a pilot test with a small sample of the target population is necessary to identify any issues that need to be corrected before deployment. This includes the cover page, duration, formatting and any other technical issues
Pretesting the questionnaire	Pretesting the survey occurs before pilot testing. During pre-testing, the other authors of the group or close colleagues should read the survey attentively, looking for questionnaire wording and formatting that may lead to incorrect answers. This includes but is not limited to possible induction of social desirability bias, double-barreled questions, acquiescence and possible ceiling and floor effects
Recall bias	A type of measurement error that occurs when respondents do not accurately remember past events or experiences
Sampling bias	Occurs when the sample selected for a study is not representative of the target population
Social desirability bias	Occurs when respondents answer questions in a way they believe is more socially acceptable or favorable, rather than being truthful, which can distort survey results. This may be especially true for sensitive issues that the survey investigator needs to be aware of

Design	Definition	Methods	Advantages	Disadvantages
Simple random sample	Each element in the population has a known, equal, and non-zero chance of being included in the sample	Random number selection or drawing lots	Requires little prior knowledge about the studied population	May not capture certain groups of interest. Might be inefficient
Systematic random sample	Similar to simple random sampling The researcher selects a random starting point and, at a specified interval, systematically selects cases from a sampling list	Systematic selection of cases using a fixed interval from a random start point	Easy to analyze data and compute sampling errors. High precision	Periodic ordering in the sampling list may introduce bias. May not capture certain groups. Might be inefficient
Stratified sample	Used when there is a particular interest in ensuring that certain groups will be included in the study or sampled at a higher (or lower) rate than others	The sample is divided into homogeneous subgroups (strata), and simple or systematic random sampling is used within each stratum	Captures key subgroups of interest. Allows disproportionate sampling and optimal allocation within strata. Maximum precision	Requires prior knowledge of the population. Increases complexity of data analysis and error computation
Cluster sample	Aims to maximize sample dispersion across the community to fully represent existing diversity while minimizing costs	The sample is divided into heterogeneous groups/clusters (e.g., people in certain geographic areas), and a random sample is drawn from the clusters	Reduces fieldwork costs. Allows sampling of groups for which individual-level data may not be available	Greater complexity in data analysis and error computation. Lower precision

location_on

Associação de Medicina Intensiva Brasileira - AMIB Rua Arminda, 93 - 7º andar - Vila Olímpia, CEP: 04545-100, Tel.: +55 (11) 5089-2642 - São Paulo - SP - Brazil
E-mail: ccs@amib.org.br

rss_feed Acompañe los números de esta revista en su lector de RSS