Descriptive Research Design: Types, Methods and Examples

Descriptive research design is a structured approach used to describe the characteristics, distribution, frequency or development of a population, phenomenon or experience without manipulating the conditions being studied. It may use quantitative, qualitative or mixed data and commonly answers questions such as what exists, who is affected, how often it occurs and how it changes over time.

Introduction

Researchers do not always need to determine why something happens. In many projects, the first and most important task is to establish what is happening, who or what is involved, how common it is, where it occurs and whether it is changing.

A descriptive research design provides the plan for producing that account. It connects a clearly defined research question with an appropriate population, sample, measurement process, data source, analytical method and reporting strategy.

This article explains the meaning, characteristics and types of descriptive research design. It also shows how to select a design, collect and analyse data, avoid common errors and write a proposal-ready methodology.

Key takeaways

• Descriptive research documents conditions, characteristics, distributions or experiences without manipulating an intervention.
• “Descriptive” identifies the study’s main objective; it is not a single data-collection method.
• A descriptive study may be quantitative, qualitative or mixed and may be cross-sectional or longitudinal.
• Surveys, observations, interviews, records and case reports are possible methods or data sources.
• Descriptive findings do not by themselves demonstrate causation.
• High-quality description still requires defensible sampling, valid measurement, transparent analysis and appropriate reporting.

What is descriptive research design?

Descriptive research design is the overall plan for systematically observing, measuring and presenting the characteristics of a defined population, event, setting or phenomenon. The researcher describes what is present rather than deliberately assigning an intervention or manipulating an independent variable.

In health research, commonly recognised descriptive designs include case reports, case series, descriptive cross-sectional studies and ecological descriptions. Qualitative researchers may also use a qualitative descriptive design when they seek a clear, experience-near account rather than an abstract theory or a phenomenological interpretation.

A strong design specifies:

• What will be described.
• The population, setting and time period.
• The unit of analysis.
• How participants or records will be selected.
• Which variables or categories will be examined.
• How information will be collected.
• How results will be summarised.
• Which limitations affect interpretation.

Characteristics of descriptive research design

1. It is guided by a clearly bounded descriptive question

Descriptive questions often begin with:

• What?
• Who?
• Where?
• When?
• How many?
• How often?
• How much?
• How do participants describe the experience?

Examples include:

• What proportion of first-year students uses generative AI for academic work?
• How often do patients miss scheduled outpatient appointments?
• What barriers do rural teachers report when adopting online learning systems?
• How did the demographic profile of doctoral candidates change from 2020 to 2026?

2. Variables are observed rather than experimentally manipulated

The researcher records characteristics or experiences as they occur. Participants are not randomly assigned to a treatment and control condition merely for the purpose of description.

This does not mean that the study lacks structure or control over quality. Researchers still control inclusion criteria, measurement procedures, interviewer training, coding rules, data-cleaning decisions and analytical methods.

3. The population, unit and context must be defined

A statement such as “this study describes students’ attitudes” is incomplete. It should identify:

• Which students.
• At which institutions.
• During which period.
• What “attitude” means.
• How it will be measured.
• Whether the unit of analysis is the student, class, department or institution.

4. It may use numerical or textual evidence

Quantitative descriptive studies report frequencies, percentages, averages, rates, distributions and confidence intervals.

Qualitative descriptive studies use interviews, focus groups, observations or documents to produce a close-to-the-data account in accessible language. Qualitative description is a legitimate methodological approach, not merely an incomplete version of another qualitative tradition.

5. It may examine one time point or change over time

A one-time survey can provide a cross-sectional snapshot. Repeated surveys can show population-level changes, while longitudinal observation can describe change among the same participants or units.

6. Its conclusions are primarily descriptive

A study may describe apparent differences or patterns, but causal statements require stronger assumptions and designs. Statements such as “students who sleep longer reported higher grades” do not establish that additional sleep caused the grades.

What descriptive research is not

Descriptive research is not the same as descriptive statistics

Descriptive research is a study objective or design logic. Descriptive statistics are techniques used to summarise data.

Means, medians and percentages may appear in an experiment, cohort study or randomised trial. Their use does not make the overall study descriptive.

A cross-sectional study is not automatically descriptive

Cross-sectional refers to the timing of measurement. A descriptive cross-sectional study estimates the frequency or distribution of a condition. An analytical cross-sectional study evaluates an association between an exposure and an outcome measured within the cross-sectional structure.

Cross-sectional designs are especially useful for prevalence estimates, but their lack of inherent temporal ordering usually limits causal interpretation.

A survey is not automatically a descriptive design

A survey is a data-collection method. It can support:

• A descriptive study of public attitudes.
• A correlational study of relationships among attitudes.
• An experimental survey using randomised question wording.
• A longitudinal panel study.
• A programme evaluation.

Descriptive and correlational research are not identical

A purely descriptive study focuses on distributions or characteristics. A correlational study explicitly examines the direction or strength of a relationship between variables.

A project can contain both components. It might first report the prevalence of academic stress and then explore its association with study hours. The first objective is descriptive; the second is analytical or correlational.

Types of descriptive research design

There is no single universally accepted list because researchers classify designs along several dimensions. The following framework avoids mixing unlike categories.

Classification by methodological approach

Approach	Main evidence	Typical output	Example
Quantitative descriptive	Numerical measurements or coded categories	Frequencies, percentages, means, rates and intervals	Percentage of employees working remotely
Qualitative descriptive	Interviews, observations, text or documents	Categories and accessible accounts of experiences	How caregivers describe barriers to home support
Mixed descriptive	Numerical and textual evidence	Statistical profile plus contextual explanation	Frequency of service use and users’ descriptions of access problems

Classification by time structure

Time structure	Meaning	Suitable question
Cross-sectional	Data collected for one defined time or short period	What percentage of students currently uses the library weekly?
Repeated cross-sectional	Comparable samples surveyed on multiple occasions	How did weekly library use change between 2024 and 2026?
Longitudinal	The same units are followed across time	How does each student’s library use change throughout the degree?
Retrospective descriptive	Existing information from an earlier period is reviewed	What were the characteristics of admissions recorded from 2020 to 2025?
Prospective descriptive	Information is collected forward from a defined starting point	What symptoms are reported during the next six months?

A longitudinal project can remain descriptive when its objective is to document change rather than identify a causal effect.

Classification by study strategy or data source

Descriptive survey

A descriptive survey collects standardised answers from a defined sample. It is useful for measuring opinions, behaviours, experiences, characteristics and self-reported conditions.

Survey quality depends on the sampling frame, recruitment process, questionnaire wording, administration mode, nonresponse and processing—not simply the number of responses. AAPOR recommends defining the target population carefully, selecting an appropriate sampling approach, asking one clear concept at a time and reporting probability and nonprobability methods transparently.

Descriptive observational study

Researchers systematically observe behaviour, events or settings without introducing an intervention.

Observation may be:

• Naturalistic or structured.
• Participant or non-participant.
• Overt or, where ethically defensible, unobtrusive.
• Human-recorded or sensor-assisted.
• Conducted continuously or through time sampling.

An observation protocol should define the behaviour, observation period, recording rules and procedures for checking agreement among observers.

Case report and case series

A case report describes one unusual, informative or clinically important case. A case series describes multiple cases sharing a defined feature.

These designs can detect emerging events and generate hypotheses, but they normally lack a comparison group and cannot estimate population prevalence unless the case-ascertainment process and denominator are known.

Ecological descriptive study

An ecological study describes groups, regions, schools, organisations or countries rather than individuals.

For example, a researcher could compare regional distributions of broadband access. Conclusions about regions should not automatically be applied to every individual within them, because doing so may produce an ecological fallacy.

Document, record and secondary-data analysis

Researchers may describe:

• Census datasets.
• Administrative records.
• Clinical files.
• School attendance records.
• Government reports.
• Social-media posts.
• Policy documents.
• Published articles.
• Archived photographs or media.

The researcher must evaluate how and why the original data were created, whether definitions changed, which cases are missing and whether access or reuse is ethically and legally permitted.

Qualitative descriptive study

Qualitative description is appropriate when the goal is a direct, practical account of participants’ experiences, perspectives or processes.

It commonly uses purposive sampling, individual interviews, focus groups, observations or documents. Analysis may involve systematic coding and thematic or content-oriented categorisation while remaining relatively close to participants’ language.

When should descriptive research design be used?

Use a descriptive design when the principal objective is to:

• Estimate prevalence, frequency or distribution.
• Establish a baseline before an intervention.
• Profile a population, organisation, service or setting.
• Document behaviours, experiences or conditions.
• Monitor changes or trends.
• Describe an unusual case or emerging event.
• Summarise routine or archival data.
• Generate hypotheses for later analytical research.
• Assess current practice against an established benchmark.
• Produce a practical qualitative account.

Do not choose the design merely because it appears easier. It should be selected because it matches the research question.

Descriptive research compared with related designs

Design	Main purpose	Manipulation	Main question	Appropriate conclusion
Exploratory	Clarify an underdeveloped problem	Usually no	What might be important?	Candidate concepts or questions
Descriptive	Characterise a defined phenomenon	No experimental manipulation	What exists, how much or how often?	Distribution, profile or account
Correlational	Examine relationships	No experimental manipulation	Are variables associated?	Direction or strength of association
Explanatory	Investigate why or how	May be observational or experimental	Why does the pattern occur?	Explanatory evidence, subject to design
Experimental	Estimate causal effects through manipulation	Yes	What is the effect of an intervention?	Stronger causal inference when assumptions hold
Evaluation	Judge programme merit or performance	Varies	Is the programme effective, efficient or appropriate?	Context-specific evaluative judgement

How to conduct descriptive research

Step 1: Define the descriptive objective

State exactly what must be described.

Weak objective:

To study social media and students.

Improved objective:

To describe the frequency, purposes and reported academic uses of social-media platforms among first-year undergraduate students at three universities during the 2026 spring semester.

Step 2: Define the population, unit, setting and period

Specify:

• Target population.
• Accessible population.
• Inclusion and exclusion criteria.
• Unit of analysis.
• Geographic or institutional setting.
• Reference period.

The unit of observation and unit of analysis may differ. A researcher might collect records from hospitals but analyse individual admissions.

Step 3: Select the design dimensions

Decide whether the project is:

• Quantitative, qualitative or mixed.
• Cross-sectional, repeated cross-sectional or longitudinal.
• Survey-based, observational, case-based or record-based.
• Prospective or retrospective.
• Conducted at the individual or aggregate level.

Step 4: Operationalise variables or qualitative concepts

An operational definition explains how a concept will be observed or measured.

For example, “frequent AI use” could mean:

Self-reported use of a generative-AI application on four or more days during the previous seven days.

The definition should identify the behaviour, threshold and reference period. Avoid vague concepts such as “regular,” “good performance” or “high engagement” unless they are clearly defined.

Step 5: Develop a sampling plan

Identify:

• The sampling frame.
• Sampling method.
• Expected sample size.
• Recruitment procedure.
• Anticipated nonresponse.
• Whether weights will be needed.
• How the achieved sample will be compared with the target population.

Probability sampling supports design-based population inference when properly implemented. Nonprobability samples can still provide useful description, but researchers should avoid presenting them as automatically representative.

Step 6: Select or develop the instrument

Possible instruments include:

• Questionnaire.
• Interview guide.
• Observation schedule.
• Data-extraction form.
• Coding framework.
• Sensor or measurement device.

Use established measures where suitable. When creating a new instrument, seek feedback, conduct cognitive testing or pilot work and document revisions.

Step 7: Plan ethics and data governance

Determine whether institutional or ethics approval is required. Prepare informed-consent procedures, confidentiality protections, access controls, retention schedules and a plan for sensitive or identifiable data.

Secondary data are not automatically free from ethical obligations. Data ownership, consent, licensing, re-identification risk and local law still matter.

Step 8: Pilot the procedures

A pilot can reveal:

• Ambiguous questions.
• Missing response options.
• Excessive completion time.
• Recruitment difficulties.
• Technical failures.
• Inconsistent observations.
• Unusable records.
• Coding disagreements.

Do not silently combine pilot data with the main study unless the protocol permits it and the measurement procedure remained compatible.

Step 9: Collect and monitor data

Use written procedures for recruitment, administration, observation, data entry and quality checks. Record deviations and monitor missingness, response patterns and interviewer or observer performance.

Survey guidance emphasises quality checks throughout the entire process, including sampling, programming, interviewing, editing, coding and analysis.

Step 10: Analyse, interpret and report transparently

Present the results that directly answer each objective. Report denominators, missing values, uncertainty and limitations. Do not convert an observed pattern into a causal claim.

Sampling and sample size in descriptive studies

Sampling choices

Common probability methods include:

• Simple random sampling.
• Systematic sampling.
• Stratified sampling.
• Cluster sampling.
• Multistage sampling.

Common nonprobability methods include:

• Convenience sampling.
• Purposive sampling.
• Quota sampling.
• Snowball or network-based recruitment.
• Volunteer online panels.

The best method depends on the target population, available frame, resources and intended inference.

Sample-size formula for a proportion

For a large population and a simple random sample, an initial sample size for estimating one proportion can be calculated as:$$n_0=\frac{z^2p(1-p)}{e^2}$$n0​=e2z2p(1−p)​

Where:

• $n_0$n0​ = initial required sample.
• $z$z = critical value for the selected confidence level.
• $p$p = expected proportion.
• $e$e = desired absolute margin of error.

When no credible value of $p$p is available, $p=0.50$p=0.50 produces the largest sample under this formula.

For a finite population of size $N$N:$$n=\frac{n_0}{1+\frac{n_0-1}{N}}$$n=1+Nn0​−1​n0​​

The calculation may then require adjustment for:

• Expected nonresponse.
• Cluster design effects.
• Stratification.
• Subgroup estimates.
• Unequal selection probabilities.
• Resource or ethical constraints.

The formula does not validate the sampling frame or eliminate selection and nonresponse bias.

Qualitative sample adequacy

Qualitative descriptive studies do not normally use a margin-of-error formula. Sample adequacy depends on the specificity of the question, diversity of relevant experiences, richness of the data, sampling strategy and analytical purpose.

Researchers should explain why the sample was sufficient rather than relying on a universal numerical threshold.

Data-collection methods

Questionnaires

Suitable for collecting standardised information from many participants.

Good questions should:

• Address one concept at a time.
• Use language the population understands.
• Avoid leading or loaded wording.
• Provide balanced and exhaustive response options.
• Specify an appropriate reference period.
• Permit “do not know” or “prefer not to answer” where justified.
• Minimise unnecessary burden.

These principles align with AAPOR’s survey best-practice guidance.

Interviews and focus groups

Useful for qualitative description or mixed designs. The guide should align with the research objective without forcing participants into predetermined conclusions.

Researchers should document:

• Interviewer background and training.
• Setting and mode.
• Recording and transcription.
• Duration.
• Language or translation.
• Coding process.
• Reflexive considerations.

Observation

Observation is suitable when behaviour can be recorded more accurately in context than through self-report.

Researchers should define:

• Observable units.
• Recording intervals.
• Start and stop rules.
• Observer position.
• Reactivity risks.
• Inter-observer consistency.

Existing records

Record-based studies may be efficient and cover long periods, but their quality is constrained by the original recording system. Variables may be missing, inconsistently defined or collected for administrative rather than research purposes.

Data analysis in descriptive research

Categorical variables

Report:

• Frequency counts.
• Percentages or proportions.
• Rates.
• Cross-tabulations.
• Confidence intervals.
• Missing values.

Percentage:$$\text{Percentage}=\frac{f}{n}\times100$$Percentage=nf​×100

Where $f$f is the frequency in a category and $n$n is the relevant denominator.

Prevalence

$$\text{Prevalence}=\frac{\text{existing cases in the defined population}}{\text{total population assessed}}$$Prevalence=total population assessedexisting cases in the defined population​

Always specify the population, condition definition and time reference. A percentage without a clear denominator can be misleading.

Continuous variables

Common summaries include:

• Mean and standard deviation for suitably distributed data.
• Median and interquartile range for skewed data.
• Minimum and maximum where informative.
• Percentiles.
• Graphical distributions.

Mean:$$\bar{x}=\frac{\sum x_i}{n}$$xˉ=n∑xi​​

Do not select the mean automatically. Examine the distribution, measurement scale and presence of extreme values.

Confidence intervals

A point estimate describes the observed sample. A confidence interval communicates sampling uncertainty under the selected statistical assumptions.

For descriptive objectives, estimates and their uncertainty are generally more informative than presenting p-values alone.

Weighting

Survey weights may account for:

• Unequal selection probabilities.
• Nonresponse adjustment.
• Post-stratification.
• Calibration to known population totals.

Weighted analysis should also use variance-estimation procedures appropriate to the sample design.

Qualitative descriptive analysis

A transparent qualitative process may include:

1. Familiarisation with the data.
2. Initial coding.
3. Development of a codebook.
4. Comparison and refinement of codes.
5. Grouping codes into descriptive categories.
6. Examination of contradictory or uncommon cases.
7. Writing a coherent account supported by selected excerpts.
8. Checking interpretations against the dataset and research question.

Software can assist organisation, retrieval and audit trails, but it does not replace interpretation.

Worked example: a descriptive cross-sectional survey

Research question

What proportion of first-year university students used generative AI for academic tasks during the previous 30 days, and for which tasks was it used?

Design

A quantitative descriptive cross-sectional survey.

Population

All first-year undergraduate students enrolled at three participating universities during the 2026 spring semester.

Sampling

A stratified sample drawn by university and faculty, using current enrolment lists where access is permitted.

Variables

• Any AI use during the previous 30 days.
• Number of days used.
• Academic tasks.
• Type of access.
• Self-reported training received.
• Age group, study field and enrolment status.

Analysis

• Number and percentage reporting any use.
• Confidence interval for the principal proportion.
• Median number of use days.
• Frequency of each task.
• Description by university and faculty.
• Missing-data report.
• Weighted estimates if selection probabilities differ.

Appropriate conclusion

In the participating universities, a specified proportion of responding first-year students reported using generative AI for academic work during the reference period.

Inappropriate conclusion

Generative AI improves the academic performance of first-year students.

The study did not manipulate AI use or establish a causal effect on performance.

Worked example: qualitative descriptive research

Research question

How do first-generation university students describe barriers to using academic-library services?

Possible design

• Qualitative descriptive approach.
• Purposive sampling of students with relevant experience.
• Semi-structured interviews.
• Descriptive coding and category development.
• Attention to differences by study mode or campus.
• Reflexive documentation of researcher decisions.

Suitable output

The findings might describe barriers involving opening hours, uncertainty about available services, digital access, prior educational experience and discomfort seeking assistance. These would be empirical categories derived from the collected accounts—not assumed findings written in advance.

Proposal-ready methodology template

This study will use a [quantitative/qualitative/mixed-method] descriptive research design to describe [phenomenon, characteristics or experiences] among [target population] in [setting] during [time period]. A [cross-sectional/repeated cross-sectional/longitudinal/retrospective] structure will be used because the objective is to [estimate a distribution, document current conditions, describe experiences or examine change] rather than establish a causal effect.

The target population will consist of [population definition], and the unit of analysis will be [unit]. Participants or records will be selected using [sampling method] from [sampling frame or recruitment source]. The planned sample size is [number and justification], with allowance for [nonresponse, design effect or subgroup analysis, where applicable].

Data will be collected using [questionnaire, interview guide, observation schedule or extraction form]. The instrument will measure or examine [key variables, indicators or categories]. Quality procedures will include [pilot testing, interviewer training, reliability assessment, cognitive testing, duplicate coding or data validation].

Quantitative data will be summarised using [frequencies, percentages, means, medians, rates and confidence intervals]. Qualitative data will be analysed through [coding and descriptive category development]. Missing data, achieved sample characteristics and study limitations will be reported transparently. Ethical procedures will include [approval, consent, confidentiality and data-security arrangements].

Advantages of descriptive research design

It establishes a reliable baseline

A descriptive study can show the starting distribution of a condition before a programme, policy or intervention is introduced.

It estimates prevalence and frequency

Population-based descriptive studies are important for determining how common a condition, behaviour or characteristic is.

It accommodates diverse evidence

Researchers can use surveys, observations, interviews, documents, administrative records or mixed data.

It preserves real-world context

Because an intervention is not deliberately imposed, researchers can document naturally occurring practices and conditions.

It supports planning and hypothesis generation

Clear descriptions can identify important subgroups, unusual patterns and unanswered questions for subsequent research.

It can be repeated

Repeated measurements using comparable definitions can help monitor change. Replication, however, depends on sufficiently detailed reporting and stable measurement.

Limitations of descriptive research design

It cannot establish causation by description alone

An observed difference may reflect confounding, selection, measurement error, reverse direction or chance.

Representativeness may be weak

A large sample is not necessarily representative. Coverage problems, volunteer recruitment and nonresponse can create systematic differences between the sample and target population.

Measurement error can distort the description

Poor questions, inconsistent observers, inaccurate records or changing definitions may produce biased results.

Cross-sectional data may lack temporal order

When exposure and outcome are measured at the same time, researchers may not know which occurred first.

Self-reports are vulnerable to response effects

Recall error, social desirability, misunderstanding and sensitivity can influence answers.

Aggregate data can invite ecological fallacy

A relationship observed across regions or organisations may not hold among individuals.

Qualitative description is context-dependent

A rich account from a purposive sample should not automatically be converted into a population percentage or universal claim.

Existing records constrain the researcher

Administrative datasets may omit important variables and may use categories that do not match the research question.

Common mistakes

Calling the study descriptive without stating what is described

Identify the population, phenomenon, context, variables and time period.

Treating every survey as descriptive

Classify the study according to its objectives, not merely its instrument.

Using causal verbs

Avoid “affects,” “leads to,” “influences” or “results in” unless the design supports the claim. Prefer “was reported with,” “co-occurred with” or “was associated with” where appropriate.

Confusing a convenience sample with the population

State the recruitment source and limit generalisation.

Reporting percentages without denominators

Readers need to know how many observations contributed to each estimate.

Ignoring missing values

The denominator can change across questions. Report item-level missingness and explain how it was handled.

Listing methods without design coherence

The question, sampling strategy, instrument and analysis should form one logical plan.

Reporting only p-values

A descriptive project should prioritise distributions, effect estimates, uncertainty and practical interpretation.

Describing qualitative software as the analytical method

NVivo, ATLAS.ti, MAXQDA or similar applications organise data. The researcher must still identify and explain the actual analytical process.

Reliability, validity and research quality

Measurement validity

Ask whether the instrument measures the intended concept. Content validity, construct validity and criterion-related evidence may be relevant, depending on the measure.

Reliability

Reliability concerns consistency. Suitable evidence might include:

• Internal consistency for multi-item scales.
• Test–retest reliability.
• Inter-rater or inter-observer agreement.
• Stability of coding decisions.
• Instrument calibration.

A high reliability coefficient does not prove that the correct construct was measured.

Internal credibility

Even without causal inference, the description should be credible. Researchers should minimise selection, information, interviewer, observer and processing errors.

External validity and transferability

Quantitative generalisation depends heavily on sampling and coverage. Qualitative researchers instead explain the context and participant characteristics so readers can judge transferability.

Reflexivity

In qualitative description, researchers should record how their roles, assumptions and interactions may have shaped data generation and interpretation.

Ethics and data protection

Descriptive research can create substantial risks even when it does not manipulate an intervention.

Important considerations include:

• Voluntary participation and informed consent.
• Protection of vulnerable participants.
• Sensitive questions.
• Confidentiality and anonymity.
• Secure storage and controlled access.
• Re-identification risk.
• Use of administrative or online data.
• Permission for recording.
• Retention and deletion procedures.
• Responsible reporting of small subgroups.

Publicly accessible information should not automatically be treated as ethically unrestricted research data. Researchers must consider expectations of privacy, platform terms, jurisdictional law and potential harm.

Digital research tools and artificial intelligence

Survey and data-collection tools

Digital platforms can support questionnaire programming, skip logic, multilingual administration, mobile data collection, audit trails and secure export. Tool selection should be based on security, accessibility, offline capability, data location, institutional policy and the needs of the population.

Statistical tools

R, Python, SPSS, Stata and similar applications can calculate descriptive estimates, confidence intervals, survey-weighted statistics and visualisations. Reproducible syntax is preferable to undocumented manual editing.

Qualitative-data tools

Qualitative software can support coding, retrieval, memos, codebooks and team comparison. It should not be presented as an automatic substitute for methodological judgement.

Appropriate uses of generative AI

AI may assist with:

• Brainstorming non-final questionnaire wording.
• Identifying duplicated or ambiguous draft items.
• Generating test cases for survey logic.
• Supporting code development.
• Producing preliminary classifications of open-text responses.
• Checking documentation for omissions.
• Improving the clarity of non-sensitive prose.

AAPOR reports that AI is already being applied to literature review, questionnaire development, interviewing, open-response analysis and scientific writing. It recommends disclosing the model, task and verification procedures.

Essential safeguards

Researchers should:

1. Avoid entering confidential or identifiable data into unapproved systems.
2. Validate AI classifications against human-reviewed data.
3. Check generated citations and factual statements.
4. Retain version information, prompts and decision logs where relevant.
5. Examine differential error across demographic or linguistic groups.
6. Disclose material AI assistance.
7. Keep human researchers accountable for final methodological decisions.

AI output should be treated as unverified until it has been checked.

Reporting descriptive research

A transparent report should include:

• The descriptive design in the title or methods.
• Research objectives.
• Setting and dates.
• Population and eligibility criteria.
• Sampling frame and sampling method.
• Sample-size rationale.
• Recruitment and response information.
• Instrument development and testing.
• Operational definitions.
• Data-collection procedures.
• Missing-data handling.
• Statistical or qualitative analysis.
• Participant or unit characteristics.
• Estimates with appropriate denominators and uncertainty.
• Biases and limitations.
• Ethics approval and consent.
• Funding and conflicts of interest.
• AI or digital-tool disclosures where material.

For observational cross-sectional research, STROBE provides a 22-item reporting framework covering the title, abstract, introduction, methods, results, discussion and funding. STROBE is a reporting guideline, not a tool for scoring methodological quality or designing the study.

For survey studies, the CROSS checklist addresses study design, questionnaire development, sample-size calculation, administration, ethical considerations, analysis and reporting.

Conclusion

Descriptive research design is appropriate when the principal aim is to produce an accurate, transparent account of a defined population, phenomenon, condition or experience. It can use quantitative, qualitative or mixed evidence and may be cross-sectional or longitudinal.

Its apparent simplicity should not be mistaken for a lack of rigour. Credible description requires a precise question, defensible sampling, valid measurement, systematic analysis, ethical data handling and conclusions that remain within the limits of the design.