03 · What You Need to Know
Description and Explanation Answer Different but Compatible Questions
Description establishes what needs to be explained
Descriptive research characterizes a phenomenon. It may establish its prevalence, frequency, distribution, characteristics, patterns, trajectories, or observable features.
For example, a study might determine that 62% of respondents report using generative AI at least weekly, that usage is more frequent among students in particular disciplines, and that the most common uses involve brainstorming and language revision.
Those findings can be useful in their own right. They tell us something about what is occurring.
But they can also create explanatory questions. Why does usage differ across disciplines? Why do some students use AI frequently while others avoid it? What processes account for the observed pattern?
In this sense, description can establish the phenomenon that explanation attempts to account for. Research-methods literature commonly distinguishes descriptive questions about what exists from explanatory questions concerned with how or why a phenomenon occurs.
Explanation asks what accounts for the pattern
Explanatory inquiry moves beyond documenting an observed pattern and attempts to account for it.
That explanation might concern causes, mechanisms, processes, relationships, contextual conditions, or theoretically meaningful factors. The appropriate form depends on the research question and disciplinary tradition.
Suppose students with greater confidence in evaluating AI-generated information use generative AI more frequently. That association might become part of an explanatory account, but several interpretations remain possible. Confidence might encourage use. Frequent use might increase confidence. Prior digital competence might influence both. Disciplinary expectations might also matter.
Explanation therefore requires more than finding variables that correlate with an outcome. Strong explanatory research considers competing interpretations and selects a design appropriate to the kind of explanation being claimed.
One study can contain descriptive and explanatory research questions
There is no methodological requirement that every research question in a study serve exactly the same purpose.
Consider a study examining student absenteeism:
| Research question |
Primary purpose |
What the answer contributes |
| How frequently are students absent during the semester? |
Descriptive |
Establishes the prevalence or frequency of absence |
| Which student groups show different patterns of absence? |
Descriptive |
Characterizes how absence is distributed |
| Which measured factors are associated with higher absenteeism? |
Potentially explanatory, depending on framing and design |
Identifies relationships that may contribute to an explanatory account |
| How do transportation difficulties, employment demands, and course experiences contribute to students' decisions to miss classes? |
Explanatory |
Investigates processes or conditions that may account for the observed patterns |
These questions can belong together when they contribute to one coherent investigation. The descriptive questions establish the pattern; the explanatory questions investigate what may account for it.
Descriptive results do not automatically become explanatory
This distinction becomes particularly important during analysis.
Imagine that a survey finds students who work more hours per week also report more absences. That is an observed association.
You might reasonably report that employment hours are associated with absenteeism. You cannot automatically conclude that employment causes absenteeism.
Perhaps students with financial difficulties are both more likely to work and more likely to encounter transportation problems. Perhaps course schedules influence both employment patterns and attendance. Perhaps the direction of the relationship is more complicated than expected.
An explanatory interpretation must remain proportionate to the design and evidence.
Watch Out
Regression coefficients, statistical significance, or associations between variables do not by themselves establish causal explanation. If your explanatory claim is causal, the design must provide a defensible basis for addressing temporal order, confounding, selection, and plausible alternative explanations.
Explanatory does not always mean causal
Explanation is broader than one form of causal effect estimation.
A qualitative study might explain how a decision develops through interactions among participants, institutional rules, and contextual conditions. A case study might develop an explanation for why an organizational reform unfolded differently across departments. A mixed methods study might combine an observed quantitative pattern with qualitative evidence about mechanisms or implementation.
Explanatory research may therefore be quantitative, qualitative, or mixed methods. Experimental and quasi-experimental designs are particularly important when the intended explanation is explicitly causal, but explanation can also concern processes, mechanisms, meanings, and contextual relationships.
The relevant question is not simply whether a study calls itself explanatory. It is what exactly the researcher claims to have explained.
A study can move from description to explanation
One coherent structure is to begin by establishing a pattern and then investigate what accounts for it.
Describe the phenomenon Establish what is happening, to whom, how frequently, where, or over what period.
Identify the pattern requiring explanation Determine which differences, relationships, trajectories, or unexpected findings deserve further investigation.
Develop or specify possible explanations Use theory, prior research, contextual knowledge, or earlier findings to identify plausible mechanisms or explanatory factors.
Generate appropriate explanatory evidence Use a design capable of distinguishing among relevant explanations to the extent required by the research question.
Match the conclusion to the evidence State what the study describes confidently and what it can explain, while preserving uncertainty where competing explanations remain possible.
This sequence is intuitive, but it is not mandatory. A study may begin with a theoretically specified explanation and still include descriptive analyses needed to characterize the sample or phenomenon.
The descriptive and explanatory components may use the same data
You do not necessarily need separate datasets.
A longitudinal dataset, for example, might be used first to describe how an outcome changes over time and then to investigate factors associated with those trajectories. A survey might provide both population descriptions and analyses of relationships among measured variables.
Whether the same dataset can support both purposes depends on the claims being made.
A dataset adequate for description may not be adequate for a strong causal explanation. Cross-sectional data can describe distributions and associations but may provide limited information about temporal ordering. Conversely, a randomized experiment designed for causal inference can also generate descriptive information about participants and outcomes.
The important question is not whether the data are reused. It is whether the structure of those data supports each intended inference.
The two purposes can also use different forms of evidence
Sometimes the explanatory component requires evidence that the descriptive component does not provide.
Suppose a large survey identifies an unexpected pattern: students who report frequent AI-assisted studying perform better in one discipline but not another. Researchers might follow the survey with interviews or observations to investigate how students in the two disciplines actually use AI and how course expectations shape those practices.
This could justify a mixed methods design if the qualitative component is intentionally used to explain or contextualize the quantitative pattern. Explanatory sequential mixed methods designs are specifically structured around an initial quantitative phase followed by qualitative inquiry that helps explain or elaborate the quantitative findings.
The choice among quantitative, qualitative, and mixed methods approaches should therefore follow the evidence required by the descriptive and explanatory questions rather than the desire to attach multiple methodological labels.
Multiple purposes do not automatically mean multiple research designs
If one study contains descriptive and explanatory questions, it does not follow that the study must have two completely separate research designs.
A single coherent design may support both. A longitudinal cohort study, for example, could describe trajectories and investigate factors associated with changes in those trajectories. An experiment could describe outcome distributions while primarily addressing an explanatory causal question.
In other cases, different components may have distinct design features. A mixed methods study might contain a quantitative observational component followed by a qualitative explanatory component.
Whether that should be described as one design with several components or more than one research design within a study depends partly on methodological convention and how substantively distinct the components are.
Do not confuse multiple purposes with an unfocused study
A study can pursue several purposes and remain coherent. The relevant test is whether the questions belong together.
If the descriptive component establishes a phenomenon and the explanatory component investigates that same phenomenon, their relationship is easy to justify.
Problems emerge when a project accumulates loosely related questions simply because data are available. A survey describes student AI use, interviews investigate faculty job satisfaction, institutional records examine graduation rates, and suddenly everything is said to be part of one “comprehensive” study because it happens at the same university.
Topical proximity is not methodological integration.
One project can contain multiple questions, methods, phases, or populations while remaining a coherent study when the components are intentionally connected to a common higher-order question. When clusters of questions become largely independent investigations, however, the project may be better understood as several studies within a broader research program.
State which claims are descriptive and which are explanatory
Researchers do not necessarily need to label every sentence as descriptive or explanatory. Still, the distinction should be visible in the logic of the paper.
The research questions should indicate what is being described and what is being explained. The methods should show how each question is addressed. The results should distinguish observed patterns from analyses intended to account for those patterns. The discussion should avoid converting description into explanation through stronger wording alone.
This is particularly important when describing the overall purpose of the study. Multiple purposes are acceptable when the relationship among them is explicit.
