Which Variables Should Be Measured to Describe the Study Sample?

What Should Readers Know About the People or Units Included in Your Study?

You have recruited participants and are preparing the descriptive information that will appear in your methodology and results sections. You know that readers need to understand who participated, but deciding which characteristics to report is less straightforward.

Should you collect age, gender, educational attainment, socioeconomic status, employment, geographic location, and institutional affiliation? What about prior experience, baseline performance, technological access, or characteristics specific to the research topic?

Many published studies present a table of participant characteristics, often called Table 1. Its familiar appearance can make the selection of descriptive variables seem routine. Yet there is no universal collection of demographic and background characteristics appropriate for every research question.

A study of university students' experiences with online learning may need information about academic program, study modality, connectivity, and previous online learning experience. A clinical study may require information about disease severity, treatment history, and baseline health status. An organizational study may depend on professional roles, employment arrangements, and institutional context.

How can researchers identify the characteristics necessary to describe their sample meaningfully without turning the baseline questionnaire into an inventory of every personal detail that could be collected?

Describe the Characteristics Relevant to Your Population, Design, and Interpretation

Selecting Descriptive Variables That Make the Study Sample Interpretable

What Are Descriptive Variables?

Descriptive variables are characteristics used to summarize the individuals, groups, organizations, observations, or other units included in a study.

They help establish the composition and relevant circumstances of the sample.

For example, researchers may describe participants by age, educational background, professional experience, baseline health status, or prior exposure to the phenomenon being investigated.

Descriptive variables are not necessarily independent variables, dependent variables, or statistical controls. They may serve primarily to characterize the sample.

However, a variable can have more than one role. Baseline writing proficiency, for example, may describe participants and also serve as a prespecified covariate in an intervention analysis.

The distinction concerns why the information is collected and how it is interpreted.

Start With the Target Population

The target population provides the first basis for choosing descriptive variables.

Readers need sufficient information to understand the population to which the research findings are intended to apply.

For example, a study involving undergraduate students should ordinarily clarify relevant educational characteristics rather than describe participants exclusively through general demographic categories.

Depending on the research question, these may include academic program, year level, enrollment status, learning modality, or institutional setting.

A study involving teachers may require professional role, teaching experience, subject area, educational level taught, and instructional context.

Researchers should identify which characteristics meaningfully distinguish participants within the intended population.

The objective is not to create a comprehensive biography of every participant. It is to provide enough information to interpret the study's evidence.

Distinguish Eligibility Variables From Descriptive Variables

Eligibility criteria determine who may enter the study. Descriptive variables characterize those who actually participate.

These functions can overlap, but they should not be confused.

For example, a study may require participants to be licensed teachers with at least one year of classroom experience.

Teaching experience is then relevant to eligibility. The researchers may also report the distribution of experience among enrolled teachers.

However, eligibility screening may already establish certain characteristics without requiring researchers to collect additional detail.

If all participants are confirmed to be undergraduate students, the study may not need to collect extensive educational histories unless these serve another purpose.

Researchers should avoid repeating questions when reliable information has already been obtained through appropriate screening or approved records.

Demographic Variables Should Be Selected Deliberately

Age, gender, educational attainment, socioeconomic circumstances, and geographic location are frequently reported in research.

However, their relevance differs across studies.

Age may be central to a developmental investigation but less informative in a narrowly defined adult professional sample.

Gender may be relevant to understanding access, experiences, or disparities in a particular setting, but collecting it should not be treated as an automatic requirement for every project.

Socioeconomic information may be essential when studying educational inequality, while detailed household financial information may be unnecessary for a study of a specific technical procedure.

Researchers should also consider whether categories are meaningful, inclusive, appropriate to the population, and compatible with the intended analysis.

Where sensitive characteristics are collected, participants should understand why the information is requested and how it will be protected.

Contextual Variables May Be More Informative Than Conventional Demographics

In some studies, characteristics of the research environment are more directly relevant than standard demographic categories.

For example, a study of online learning may benefit from describing participants' internet connectivity, access to suitable devices, study environment, and previous experience with online courses.

These variables may help readers understand the conditions under which the observed experiences occurred.

However, researchers should avoid assuming that any one contextual indicator fully represents a broader construct.

Owning a laptop does not necessarily imply reliable digital access. Similarly, having internet connectivity does not establish that a participant has a private or suitable learning environment.

Contextual variables should be operationalized according to the aspects of the environment relevant to the research question.

Baseline Characteristics Help Describe the Starting Conditions

Studies involving interventions or longitudinal follow-up may need information about participants' circumstances before the relevant exposure or intervention.

Baseline outcome measures, previous experiences, health status, prior performance, and other characteristics may help establish these starting conditions.

For example, researchers evaluating an instructional intervention may describe students' baseline writing performance and previous experience with the instructional technology.

In a randomized trial, baseline tables can show the observed characteristics of the assigned groups.

However, observed baseline differences should not automatically be interpreted as evidence that randomization failed. Chance differences can occur even under valid random assignment.

CONSORT guidance discourages routine significance testing of baseline differences in randomized trials because randomization already explains their probabilistic origin.

Descriptive reporting should support understanding rather than become an informal test of whether the groups were properly randomized.

Sample Description Is Not the Same as Representativeness

Describing the characteristics of participants does not establish that they represent the target population.

Suppose a study reports that its sample includes students from several academic programs and year levels.

This information describes sample diversity, but it does not show whether the proportions correspond to those in the eligible student population.

Assessing representativeness requires attention to the sampling frame, recruitment process, participation probabilities, and relevant population benchmarks.

Even close agreement on observed demographic characteristics does not guarantee similarity on unmeasured variables.

Researchers should distinguish sample description from claims about sampling bias and the population applicability of research findings.

Choose Variables That Help Readers Evaluate Applicability

A useful sample description allows readers to assess whether the participants and conditions resemble those relevant to another population or setting.

For example, the results of a digital learning intervention evaluated among students with reliable personal computers may not apply in the same way to students who rely on shared devices and intermittent connectivity.

Describing technological access can therefore be more informative than reporting several unrelated personal characteristics.

Similarly, an intervention evaluated among experienced teachers may have different implementation implications for novice teachers.

These considerations relate to external validity and transportability, but descriptive similarities alone cannot establish that findings will generalize.

The aim is to provide information relevant to assessing applicability, not to claim that a descriptive table proves generalizability.

Use Reporting Guidelines to Identify Relevant Characteristics

Reporting guidelines can help researchers identify information that readers need to evaluate a study.

The STROBE statement recommends reporting participant characteristics, including demographic, clinical, and social information where relevant, together with exposures and potential confounders.

CONSORT reporting guidance addresses baseline characteristics and participant flow in randomized trials.

These guidelines are not universal questionnaires. They do not prescribe the same demographic variables for every study.

Researchers should use the reporting guidance appropriate to their design and interpret its recommendations in relation to the study's objectives.

Planning descriptive measurements before data collection is preferable to discovering during manuscript preparation that important contextual information is unavailable.

What Should Appear in a Sample Characteristics Table?

A sample characteristics table should present information that is meaningful for understanding the enrolled participants or observational units.

It may include sample size, relevant demographic characteristics, baseline outcomes, prior experiences, and contextual factors.

For comparative studies, characteristics may be reported separately by group when this helps readers understand the sample and design.

For continuous variables, appropriate summaries may include means and standard deviations or medians and interquartile ranges, depending on the distribution and reporting purpose.

For categorical variables, counts and percentages are commonly useful.

Researchers should make denominators clear, especially when variables have missing observations.

For example, reporting that 60% of participants have prior online learning experience may be misleading if only half of the sample answered the relevant question.

Do Not Assume Every Descriptive Variable Needs Statistical Testing

Descriptive tables often contain many characteristics, but this does not mean that each requires a hypothesis test.

Testing every demographic difference may introduce unnecessary multiplicity and encourage interpretations that are not central to the research question.

In randomized trials, significance tests of baseline differences are generally discouraged as a routine practice.

In observational studies, group differences may be relevant, but statistical significance alone does not determine whether a characteristic is an important confounder or whether adjustment is necessary.

Researchers should distinguish describing the sample from testing substantive hypotheses.

Descriptive reporting can be valuable without attaching a p-value to every row.

How Much Detail Is Appropriate?

The level of detail should reflect the research question, sample size, and disclosure risks.

For example, reporting exact ages may be unnecessary when age categories adequately serve the descriptive purpose.

However, categories should not be chosen merely to conceal important variation or create favorable patterns.

In small samples, combinations of characteristics may make individuals identifiable even when names are omitted.

Researchers should consider whether highly detailed tables could reveal sensitive information about participants or small institutions.

Appropriate aggregation, suppression, or other disclosure-control procedures may be necessary.

The relevant privacy and ethics requirements should guide these decisions.

Descriptive Variables Are Not Automatically Confounders

A common analytical mistake is to treat every characteristic included in Table 1 as a variable that must be controlled statistically.

However, descriptive relevance and causal adjustment are different considerations.

Age may be useful for describing a sample without necessarily being a confounder of the specific exposure-outcome relationship.

Conversely, a variable that is not conventionally demographic may be an important confounder.

Researchers should identify confounders through substantive causal reasoning rather than automatically including every descriptive characteristic in a regression model.

This distinction is particularly important when considering which variables should be measured to control for confounding.

Consider Sampling Design Variables

Some information needed to describe and analyze a sample concerns how observations were selected.

Examples include sampling strata, clusters, recruitment sites, and inclusion probabilities.

These variables may be essential to estimating population quantities or calculating appropriate standard errors.

For example, a national survey using stratified multistage sampling may require design weights and cluster identifiers.

Such variables are not necessarily demographic characteristics, but they are important to understanding the sample and the estimates derived from it.

Researchers should distinguish participant characteristics from sampling design information while ensuring that both are available when needed.

Missing Descriptive Data Should Be Reported Transparently

Not every participant will necessarily provide every background characteristic.

Some questions may be sensitive, optional, or unavailable from existing records.

Researchers should report missingness where relevant and avoid silently calculating percentages from inconsistent denominators.

For example, if 180 of 200 participants report their employment status, percentages should be clearly based on the 180 observed responses or otherwise explained.

Researchers should not assume that missing demographic information is random.

In some settings, participants with particular circumstances may be less willing to disclose sensitive information.

The implications depend on how the variable is used and the reasons for missingness.

Collect Descriptive Information Without Unnecessary Intrusion

Descriptive data collection should be proportionate to the scientific purpose.

Researchers may be tempted to include extensive demographic sections because these are common in published studies.

However, familiar questionnaire items can still impose burden or create privacy concerns.

The Philippines' Data Privacy Act of 2012 and other applicable data protection frameworks establish requirements concerning the appropriate processing of personal information.

Researchers should evaluate whether each descriptive characteristic is necessary, whether less intrusive alternatives are available, and how the information will be protected.

The general principle is consistent with avoiding unnecessary collection of available research variables.

Build a Descriptive Variable Specification Matrix

A descriptive variable matrix can connect each proposed characteristic to its reporting purpose.

The matrix should be adapted to the actual study. No variable in this example is universally required.

The objective is to identify characteristics that help readers understand the sample and the boundaries of the findings.

Describing Students in a Digital Learning Research Study

Preparing the Sample Characteristics Table

Suppose the following hypothetical data are obtained.

All percentages in this hypothetical table use the full sample of 240 participants as the denominator. The categories within each characteristic are mutually exclusive and exhaustive for the example.

The table shows the composition of the recruited sample but does not establish whether these distributions match those of the target population.

Interpreting the Description

The resulting description is relatively focused, but it provides information relevant to interpreting the study.

If the research concerned another population or outcome, a different set of descriptive variables could be more appropriate.

Common Misconceptions About Describing a Research Sample

Describe the Sample in Ways That Help Readers Interpret the Evidence

Begin with the target population and identify characteristics that are relevant to the research setting, outcomes, and intended applicability.

Then distinguish information needed for sample description from variables required for causal adjustment or other analyses.

Before finalizing the instrument, draft a preliminary sample characteristics table with the variables you intend to report.

This can reveal missing definitions, inappropriate categories, and information that may be unnecessary.

However, the table should remain a consequence of the research design rather than the starting point for indiscriminate data collection.

Before Selecting Descriptive Variables

Frequently Asked Questions About Descriptive Variables

Describe the Sample According to What Readers Need to Understand

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.