What Does It Mean When Your Objectives Promise More Than Your Design Can Deliver?

Can an Objective Be Perfectly Clear and Still Be Methodologically Impossible?

Your objective says "to determine the effect of generative AI use on student achievement." Your design is a one-time survey asking students how often they use generative AI and recording their current grades.

The objective sounds specific. The variables are identifiable. The data may even reveal an association.

But can that design deliver the "effect" the objective promises?

Research objectives do more than make a proposal sound purposeful. They define what the study intends to accomplish and therefore imply what evidence the design must be capable of producing. A mismatch occurs when the objective requires a type, strength, scope, or precision of conclusion that the planned design cannot support.

The Objective Is Making a Claim the Evidence Cannot Support

An Objective Is a Commitment About What the Study Will Accomplish

Research objectives should be closely related to the research question and sufficiently specific and achievable to guide the study. Guidance based on SMART criteria similarly emphasizes that objectives should be achievable or realistic within the constraints of the design, resources, and research context.

This means that an objective cannot be evaluated solely as a sentence. You must ask what evidence would be required to fulfill it.

An objective such as "to compare," "to estimate," "to explore," "to predict," or "to evaluate an effect" points toward different evidentiary requirements. The appropriate design depends on the actual research problem, but the basic principle remains: the study should be capable of doing what the objective says it will do.

The clearest warning sign is a change in the level of inference

Consider:

Research question: Is frequency of generative AI use associated with academic writing performance?

Objective: To determine the effect of generative AI use on academic writing performance.

The question asks about association. The objective promises an effect. That is not merely a stylistic variation.

Depending on the context, "effect" is commonly interpreted causally. Establishing a causal effect requires more than observing that two variables vary together. Researchers must consider alternative explanations, temporality, confounding, selection processes, measurement, and the design features used to support causal inference.

If the planned design can only estimate an association, a more coherent objective might be:

To examine the association between frequency of generative AI use and academic writing performance.

Changing the objective does not weaken the study. It makes the promised inference consistent with the evidence the design is intended to produce.

Cross-sectional data cannot directly demonstrate change over time

Another common mismatch appears when the objective contains words such as "increase," "decrease," "improve," "develop," or "change," but the study measures participants only once.

Suppose the objective is:

To determine whether students' AI literacy improves during university education.

A cross-sectional comparison of first-year and fourth-year students might reveal that the groups differ in AI literacy. But because different students are being compared at one point in time, the design does not directly observe the same students improving as they progress through university.

A more defensible objective for that design might be:

To compare AI literacy between first-year and fourth-year students.

If the researcher genuinely wants to study within-person change, a longitudinal design or another design capable of addressing change may be required.

Measuring perceptions cannot automatically establish actual outcomes

Researchers sometimes use self-report measures to answer questions that concern objectively different constructs.

For example:

Objective: To determine whether generative AI improves students' academic writing performance.

Data collected: Students' agreement with the statement, "Generative AI improves the quality of my academic writing."

The data can provide evidence about students' perceptions of improvement. They do not directly measure writing performance.

A corresponding objective might instead be:

To examine students' perceptions of how generative AI influences their academic writing.

Alternatively, if actual writing performance is the intended outcome, the researcher needs an appropriate performance measure and a design capable of addressing the intended comparison or effect.

The population in the objective must match the population the design can support

Suppose a study recruits undergraduate students from one degree program at one university but states:

Objective: To determine university students' attitudes toward generative AI in higher education.

The objective appears to encompass a much broader population than the sample. Whether broader generalization is defensible depends on the sampling strategy, target population, context, and inferential approach.

A narrower objective might specify the actual target population represented by the design rather than silently treating a convenience sample as though it represents all university students.

The objective can require a variable the study never actually measures

A proposal may be logically aligned at the wording level but fail at operationalization.

Consider:

Objective: To examine the relationship between AI literacy and responsible generative AI use.

If the questionnaire measures factual AI knowledge and frequency of AI use, neither measure necessarily operationalizes the constructs named in the objective.

The researcher must either select measures that adequately represent AI literacy and responsible AI use or revise the objective to match what the study actually measures. This is why consistent variables and terminology must extend into operationalization.

An objective can demand more precision than the sample provides

Study feasibility is also statistical.

An objective may propose comparisons across numerous disciplines, academic ranks, age groups, institution types, and levels of AI experience. Even if the overall sample appears large, some subgroups may contain too few observations to estimate differences with useful precision or to support the intended models.

The issue is not solved by writing every subgroup into the objective. The sample-size rationale and analysis plan need to support the comparisons being promised.

A design can answer a different question very well

A mismatch does not necessarily mean that the study design is poor. Sometimes the design is entirely appropriate, just not for the objective that has been attached to it.

The verb can expose the problem, but replacing the verb is not always enough

Words such as "determine," "establish," "prove," "cause," "impact," and "effect" deserve scrutiny because researchers may use them to imply stronger conclusions than their designs support.

However, simply replacing "determine the effect" with "examine the relationship" does not automatically repair the study. The underlying research question, hypothesis, measures, analysis, and eventual interpretation must also be compatible with the revised objective.

Conversely, causal language is not prohibited merely because it is strong. If the study is explicitly designed for causal inference and the assumptions and evidence warrant that interpretation, causal objectives can be appropriate. The issue is evidentiary fit, not a blacklist of verbs.

The objective may be achievable only with a different design

When an objective exceeds the design, researchers sometimes assume the objective must be weakened. Not necessarily.

If the objective represents the central scientific question and is important enough to justify the additional work, strengthening the design may be the better choice.

For example, a researcher genuinely interested in whether an educational intervention improves AI literacy could move from a one-time observational survey toward an appropriate intervention design with pre-specified outcomes and measurements over time. The exact design would depend on the question, context, feasibility, ethical considerations, and inferential goal.

The choice is therefore between changing what you promise and changing what you do.

The problem can originate in the research question

Objectives do not exist independently. An overpromising objective may simply be faithfully translating an overambitious question.

If the question asks, "How does generative AI cause changes in students' critical thinking?" while the planned study consists of a cross-sectional self-report survey, rewriting only the objective leaves the original mismatch intact.

Trace the issue back through the correspondence among the research question, objective, and hypothesis. The study should be coherent from the question through the evidence and conclusion.

Feasibility can also make an objective unattainable

An objective may be theoretically compatible with a design yet unrealistic under the actual research conditions.

Perhaps the required sample cannot be recruited. The necessary follow-up period exceeds the project timeline. A validated measure is unavailable in the required context. The intervention cannot be implemented consistently. The planned analysis requires expertise or data that the project does not have.

Research-question guidance commonly treats feasibility as including available participants, technical expertise, resources, time, and manageable scope. Objectives should likewise be achievable within the study's practical constraints.

This is why a project with too many questions, objectives, or hypotheses can eventually produce objectives that no longer fit what the study can realistically deliver.

From an Overpromising Objective to a Defensible One

Common Mistakes When Objectives and Designs Do Not Fit

Choose Between Narrowing the Promise and Strengthening the Evidence

When an objective exceeds the study design, there are usually two broad routes: revise the objective to match what the planned study can validly answer, or modify the design so that it can address the intended objective.

After making the change, return to the beginning of the study and trace the consequences. A revised objective may require corresponding changes to the research question, hypothesis, variable definitions, analysis plan, title, or intended conclusions.

If that feels inconvenient, it is still considerably less inconvenient than discovering during the defense or peer review that the central objective asks the study to establish something its design never had the capacity to show.

Check Whether Your Design Can Fulfill Each Objective

Questions About Matching Objectives to Research Design

Your Objective and Design Must Make the Same Evidentiary Promise

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.