Using Preliminary Data to Refine Your Research Question

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-27

Quick Answer

• Use descriptive statistics and exploratory plots from preliminary data to identify new angles, feasibility constraints, and measurement issues before finalizing your research question.
• Document every step of your preliminary analysis in a lab notebook or electronic record so you can distinguish hypothesis-generating exploration from confirmatory testing.
• Preliminary data can sharpen your question, but it cannot prove your hypothesis, and overusing the same dataset for both discovery and confirmation risks biased conclusions.

Why Preliminary Data Matters in Research Design

Preliminary data serves a distinct purpose in the research lifecycle. It is not a miniature version of your final study. Instead, it provides the empirical grounding you need to ask a better question, design a feasible protocol, and write a competitive grant application. Researchers who skip this step often find themselves committing to questions that are too broad, too narrow, or impossible to answer with the resources available.

The National Library of Medicine hosts a collection of research methods resources that cover the full arc of study design, from formulating questions to interpreting results. These resources emphasize that the quality of your research question determines the quality of your entire project. A vague question produces vague methods, and vague methods produce results that are difficult to interpret or publish.

When you have preliminary data in hand, you have an opportunity to move from a general interest area to a specific, answerable question. This refinement process involves looking at what your data actually show, what they do not show, and what they cannot show given how they were collected. Each of these observations can reshape your question in a different way.

The Difference Between Exploration and Confirmation

One of the most important distinctions in research design is the line between exploratory analysis and confirmatory analysis. Exploratory analysis is what you do with preliminary data. You look for patterns, generate hypotheses, and identify variables that seem worth measuring. Confirmatory analysis is what you do with your main study data. You test a pre-specified hypothesis using methods that were chosen before you saw the results.

The Committee on Publication Ethics core practices address the responsibilities that come with this distinction. Authors must be transparent about how data were collected and analyzed, and they must avoid practices that misrepresent the research process. If you use the same data to generate and test a hypothesis without acknowledging that process, you create a problem for yourself at the publication stage.

Your preliminary data can help you refine your question, but you need to be honest about the fact that the refinement happened after you saw the data. When you write your final paper, you should describe your hypothesis as one that emerged from preliminary exploration, or you should collect new data to test it. This transparency protects you from accusations of p-hacking or hypothesis switching after the fact.

How Preliminary Data Sharpens Your Question

Identifying New Angles From Observed Patterns

The most direct way preliminary data refines your research question is by revealing patterns you did not anticipate. Suppose you are studying growth rates in dairy calves and your preliminary data show an unexpected difference between seasons. That observation can lead you to refine your question from "What factors affect calf growth?" to "Does seasonal temperature variation affect calf growth through changes in feed intake?"

This refinement works because the data give you a specific direction to pursue. You are no longer casting a wide net. You are focusing on a mechanism that your preliminary data suggest is important. The National Institutes of Health grant resources describe how preliminary data strengthen applications by showing reviewers that the proposed work is grounded in actual observations instead of speculation.

Testing Feasibility Constraints

Preliminary data also reveal whether your planned study is feasible. You might discover that the measurement you planned to use is too variable, the animals you planned to study are harder to recruit than expected, or the laboratory assay you planned to run produces inconsistent results. Each of these findings should push you back to your research question.

If your preliminary data show that a key measurement is highly variable, you have three options. You can refine your question to focus on a more reliable outcome, you can increase your sample size to account for the variability, or you can change your measurement approach. The choice depends on which option best serves your underlying scientific interest.

Refining Population and Comparison Groups

Preliminary data often reveal that the population you planned to study is more heterogeneous than you assumed. This observation can refine your question by narrowing your population definition or by adding a comparison group you had not considered.

For example, if your preliminary data show that response to a treatment differs by age group, you might refine your question to focus on a specific age range or to include age as a stratification variable. This refinement makes your question more precise and your eventual answer more useful to farmers or clinicians who need to know which animals will benefit from a given intervention.

The Role of Descriptive Statistics

Descriptive statistics are the foundation of preliminary data analysis. They summarize your data in ways that let you see patterns, assess variability, and identify problems. The key descriptive statistics for refining a research question include measures of central tendency, measures of spread, and frequency distributions.

Measures of central tendency, such as the mean and median, tell you where your data are centered. Measures of spread, such as the standard deviation and interquartile range, tell you how much your data vary. Frequency distributions tell you how often each value occurs. Together, these statistics give you a picture of what your data look like before you attempt any formal hypothesis testing.

The EQUATOR Network provides reporting guidelines that emphasize the importance of describing your data accurately and completely. When you eventually write your paper, reviewers will expect to see descriptive statistics for all key variables. Using them now, during the refinement phase, helps you identify which variables deserve a central place in your research question and which are too noisy or too sparse to support a meaningful study.

Using Exploratory Plots to See Your Data

Exploratory plots are often more informative than tables of numbers, especially in the early stages of refining a research question. A well-constructed plot can reveal patterns, outliers, and relationships that summary statistics miss.

Scatter Plots for Relationships

A scatter plot shows the relationship between two continuous variables. If your preliminary data include measurements of two variables that you suspect are related, a scatter plot will show you whether that relationship exists, whether it is linear or curved, and whether it is driven by a few extreme points.

This information can refine your research question in several ways. If the relationship is linear, you can ask a question about the strength and direction of that relationship. If the relationship is curved, you might need to ask a question about the nature of that curvature. If the relationship is driven by outliers, you need to decide whether those outliers represent real biological variation or measurement error.

Box Plots for Group Comparisons

A box plot shows the distribution of a continuous variable within groups. If your preliminary data include measurements from multiple groups, such as different treatment groups or different farms, box plots let you compare their centers and spreads at a glance.

Box plots are particularly useful for refining questions about group differences. They show you also whether groups differ in their central values but also whether they differ in their variability. A research question about group differences is more complex when the groups have different variances, because that difference itself may be biologically meaningful.

Time Series Plots for Trends

If your preliminary data were collected over time, a time series plot shows you how your variable of interest changes across the study period. This is especially relevant in animal farming contexts, where growth, milk production, and disease incidence all have temporal patterns.

A time series plot can reveal seasonal trends, sudden changes, or gradual drifts that should be incorporated into your research question. If your preliminary data show a strong seasonal pattern, your research question should account for that pattern, either by focusing on a specific season or by including season as a covariate.

A Practical Workflow for Refining Your Question

Step 1: Organize Your Preliminary Data

Before you can learn anything from your preliminary data, you need to know what you have. Organize your data in a spreadsheet or database with clear variable names, consistent units, and documented missing values. The NIH Data Management and Sharing Policy describes expectations for data organization and documentation that apply to NIH-funded research, and the same principles are good practice for any research project.

Your data organization should include a data dictionary that defines each variable, its units, and its coding. This documentation will be essential when you write your methods section, and it will help you avoid confusion as you explore the data.

Step 2: Calculate Descriptive Statistics

For each variable that might be relevant to your research question, calculate the mean, median, standard deviation, and range. Look for variables with extreme variability, because those variables may be difficult to study without large sample sizes. Look for variables with many missing values, because those variables may not be feasible to measure in your main study.

Step 3: Create Exploratory Plots

Create scatter plots, box plots, and time series plots for the relationships that seem most relevant to your research interest. Spend time looking at these plots. Ask yourself what patterns you see, whether those patterns make biological sense, and what new questions they raise.

Step 4: Write Down Candidate Questions

Based on what you have seen, write down three to five candidate research questions. These questions should be specific enough that you could design a study to answer them, but they should still be open to revision.

Step 5: Evaluate Each Candidate Question

Evaluate each candidate question against three criteria. First, is it answerable with the methods and resources available to you? Second, is it specific enough to guide a focused study? Third, does it address a question that matters to your field or to the farmers, clinicians, or other stakeholders who will use your results?

Step 6: Select and Refine Your Final Question

Select the candidate question that best meets these criteria and refine it further. Make sure it specifies the population, the exposure or intervention, the comparison, and the outcome. This level of specificity will guide your study design and make your eventual paper easier to write.

At a Glance

Common Failure Patterns When Using Preliminary Data

Overinterpreting Small Samples

Preliminary data are usually collected from small samples, and small samples produce unstable estimates. A pattern that looks clear in a sample of ten animals may disappear in a sample of one hundred. Researchers who overinterpret preliminary data often commit to research questions that are not supported by the evidence.

The remedy is to treat preliminary data as suggestive instead of conclusive. Use the patterns you see to generate questions, but do not assume those patterns will hold in your main study. Your main study is the test, not the preliminary data.

Confirming on the Same Data

Using the same data to generate and test a hypothesis is a common but problematic practice. When you refine your research question based on what you see in your preliminary data, you are generating a hypothesis. Testing that hypothesis requires new data.

The Committee on Publication Ethics core practices emphasize the importance of accurate reporting of how research was conducted. If you test your hypothesis on the same data that generated it, you must disclose that process. Better yet, design your main study to collect new data that can provide a genuine test.

Ignoring Measurement Problems

Preliminary data often reveal measurement problems that are invisible in a research proposal. A scale that drifts, an assay that produces inconsistent results, or a survey question that respondents misunderstand can all be detected in preliminary data if you look carefully.

Ignoring these problems and proceeding with your original research question is a recipe for a failed study. The refinement process should include a hard look at your measurement methods. If your preliminary data suggest that a measurement is unreliable, your research question needs to change, or your measurement method needs to change.

Letting Data Drive Without Biological Reasoning

Preliminary data can refine your research question, but they should not replace biological reasoning. A pattern in your data that has no plausible biological explanation may be a statistical artifact or a measurement error. Committing to a research question based on such a pattern is risky.

The best research questions emerge from an interaction between data and theory. Your preliminary data can point you toward new angles, but you should evaluate those angles using your knowledge of the biological system you are studying.

Records and Measurements You Should Keep

Good record keeping during the preliminary phase protects you at every later stage of your research. The NIH Data Management and Sharing Policy describes expectations for data management that apply to NIH-funded research, and the same principles are good practice for any project.

Your records should include the raw preliminary data, the code or commands used to analyze them, the descriptive statistics and plots you generated, and notes about what you observed and how those observations shaped your research question. This documentation serves several purposes. It helps you write an accurate methods section. It provides evidence of your research process if questions arise. And it allows you or others to revisit the preliminary analysis if new questions emerge.

The ORCID for Researchers resource describes how maintaining a persistent researcher identifier helps you connect your work across publications and projects. While ORCID is primarily about publication identity, the underlying principle applies to your research process as well. Keeping clear records of what you did and why you did it makes your work more transparent and more trustworthy.

How Preliminary Data Strengthens Grant Applications

The National Institutes of Health grants portal describes the application and review process for NIH funding. Reviewers evaluate grant applications based on significance, innovation, and approach. Preliminary data can strengthen your application on all three dimensions.

Preliminary data demonstrate significance by showing that the problem you propose to study is real and measurable. They demonstrate innovation by showing that your approach is grounded in actual observations. And they demonstrate the strength of your approach by showing that your methods are feasible and your measurements are reliable.

When you include preliminary data in a grant application, you should present them clearly and honestly. Describe what you did, what you found, and how those findings shaped your proposed research question. Do not overstate what your preliminary data show. Reviewers are experienced at distinguishing genuine preliminary data from cherry-picked results.

Reporting Guidelines and Transparency

The EQUATOR Network provides a comprehensive collection of reporting guidelines for health research. These guidelines help researchers report their methods and results completely and transparently. While the specific guideline you use will depend on your study design, the underlying principle applies to all research.

When you eventually publish your study, you should describe how your research question was developed, including the role of preliminary data. This transparency helps readers understand the context of your findings and evaluate their validity. It also protects you from concerns about selective reporting or hypothesis switching.

The Committee on Publication Ethics core practices address the responsibilities of authors, reviewers, and editors in maintaining the integrity of the research record. These practices include accurate reporting of how research was conducted, transparent handling of data, and honest disclosure of conflicts of interest. Using preliminary data to refine your research question is a legitimate and valuable practice, but it must be reported honestly.

Welfare and Safety Context

In animal research, the refinement of your research question has direct implications for animal welfare. A well-refined question allows you to design a study that uses the minimum number of animals needed to answer the question reliably. A poorly refined question may lead to a study that is underpowered, requiring more animals than necessary, or that fails to answer the question at all, wasting the animals that were used.

Preliminary data can help you estimate the variability you will encounter and the sample size you will need. This information is essential for designing a study that is both scientifically valid and ethically responsible. If your preliminary data show that a planned outcome is highly variable, you need to either increase your sample size, refine your outcome measure, or reconsider your question.

Safety considerations also apply. If your preliminary data suggest that a planned procedure is riskier than expected, or that a treatment has unexpected side effects, you need to incorporate those observations into your research question and study design. The refinement process is also about scientific precision. It is also about protecting the animals and people involved in your research.

Professional Escalation Criteria

Some situations call for consultation with a statistician, a research mentor, or an institutional official. You should seek professional input if your preliminary data show patterns that you do not understand, if you are uncertain about the appropriate statistical methods, or if your preliminary data suggest that your planned study is not feasible.

You should also seek input if you are considering a major change in your research direction based on preliminary data. A statistician can help you evaluate whether the patterns you see are likely to be real. A research mentor can help you assess whether the new direction is worth pursuing. An institutional official can help you understand the regulatory and ethical implications of any changes.

The NIH grants portal describes the resources available to researchers at each stage of the grant lifecycle. If you are preparing a grant application, your program officer can provide guidance on whether your preliminary data are sufficient and how to present them effectively.

Frequently Asked Questions

How much preliminary data do I need to refine my research question?

The amount of preliminary data you need depends on the complexity of your question and the variability of your measurements. A small dataset can be enough to reveal feasibility problems or generate new hypotheses, but it is rarely enough to establish that a pattern is real. Treat preliminary data as a starting point for refinement, not as proof of an effect.

Can I use the same data to generate and test my hypothesis?

Using the same data for both generation and testing is problematic because the test is not independent. If you refine your question based on what you see in your data, then test that question on the same data, your results will be overly optimistic. Collect new data for the confirmatory test, or clearly describe your analysis as exploratory.

What should I do if my preliminary data contradict my original hypothesis?

Contradictory preliminary data are valuable information. They may indicate that your original hypothesis is wrong, that your measurement methods are flawed, or that your study population differs from what you expected. Use the contradiction to refine your question, but investigate the source of the contradiction before committing to a new direction.

How do I present preliminary data in a grant application?

Present preliminary data clearly and honestly. Describe what you did, what you found, and how those findings shaped your proposed research question. Include descriptive statistics and exploratory plots that reviewers can evaluate. Do not overstate what your preliminary data show, and acknowledge their limitations.

What is the difference between a research question and a hypothesis?

A research question is the broad question your study aims to answer. A hypothesis is a specific, testable prediction about the answer to that question. Preliminary data can help you refine both. They can narrow your research question and help you formulate a hypothesis that is specific enough to test.

How do I know if my research question is too broad or too narrow?

A research question is too broad if it cannot be answered within the scope of a single study. It is too narrow if it addresses a question that is too trivial to matter. Preliminary data can help you find the right balance by showing you which variables are measurable and which effects are large enough to detect.

What records should I keep from my preliminary analysis?

Keep your raw data, your analysis code, your descriptive statistics, your exploratory plots, and notes about what you observed and how those observations shaped your research question. This documentation will help you write your methods, respond to reviewer comments, and maintain the integrity of your research process.

When should I consult a statistician about my preliminary data?

Consult a statistician early in the refinement process, especially if you are uncertain about how to interpret your preliminary data or how to design your main study. A statistician can help you evaluate whether the patterns you see are likely to be real, estimate the sample size you will need, and choose appropriate analysis methods.

Using the Evidence

Related Articles

• The Role of the Research Ethics Committee: What Researchers Need to Know
• Qualitative Data Examples: How to Recognize and Use Non-Numerical Evidence
• Cross-Sectional vs. Longitudinal Studies: Choosing the Right Design for Your Research Question
• Protein Synthesis Step By Step
• Welch's t-Test: When and How to Use It

References and Further Reading

• Research Methods Resources. National Library of Medicine.
• EQUATOR Network. EQUATOR Network.
• Core Practices. Committee on Publication Ethics.
• NIH Grants and Funding. National Institutes of Health.
• ORCID for Researchers. ORCID.
• Data Management and Sharing Policy. National Institutes of Health.
• NCBI Literature Resources. National Center for Biotechnology Information.
• PubMed. National Library of Medicine.

This article is educational and does not replace institutional policy, professional advice, or applicable safety and regulatory requirements.