From Raw Data to Chapter 4: A Step-by-Step Guide to Analyzing and Presenting Thesis Results

Completing a research questionnaire and collecting responses are significant milestones in any thesis journey. However, for many students, the real challenge begins after the data have been collected. How should the responses be analyzed? Which statistical tools should be used? How can statistical outputs be transformed into clear tables, meaningful interpretations, and a well-written Chapter 4?

These questions are common among undergraduate students, master’s candidates, doctoral researchers, and even researchers who have already completed several studies.

The good news is that preparing Chapter 4 becomes much more manageable when data analysis follows a systematic process. With the right statistical methods, appropriate software, and a clear understanding of research objectives, researchers can transform raw data into findings that answer their research questions.

This guide explains the essential steps, from preparing your dataset to writing and interpreting your research results.

1. Understand Your Research Questions Before Analyzing the Data

One of the most common mistakes in thesis research is starting statistical analysis without reviewing the study’s research questions and objectives.

Statistical analysis should not be driven solely by the available software or the statistical tests a researcher knows how to perform. Instead, the analysis must be determined by what the study intends to establish.

Before opening SPSS, Jamovi, JASP, or another statistical program, review your statement of the problem.

For example, consider a study examining the relationship between financial literacy and financial well-being among teachers.

The research questions might include:

• What is the level of financial literacy among the respondents?
• What is their level of financial well-being?
• Is financial literacy significantly related to financial well-being?
• Does financial literacy significantly predict financial well-being?

Each question requires a different analytical approach.

Descriptive statistics can summarize the respondents’ levels of financial literacy and financial well-being. Correlation analysis can examine the relationship between the two variables. Regression analysis can determine whether financial literacy statistically predicts financial well-being.

The important principle is simple: choose the statistical method based on the research question, measurement, study design, and assumptions—not merely on software availability.

2. Prepare and Clean Your Dataset

Before conducting statistical tests, ensure that the dataset is accurate, complete enough for the planned analysis, and properly organized.

Poor data preparation can lead to misleading results even when the statistical procedure itself is correct.

Check the following:

Variable names and coding. Use clear, consistent variable names. For example, FL1, FL2, and FL3 might represent financial literacy questionnaire items.

Missing responses. Identify unanswered items and determine how missing values will be handled. The appropriate method depends on the extent and pattern of missingness, the analysis, and the assumptions involved.

Incorrect entries. Check for values outside the permitted range. If a questionnaire uses a five-point Likert scale, a response coded as 8 may indicate a data-entry error.

Reverse-coded items. If negatively worded items are included, reverse-score them when required before computing composite scores or conducting reliability analysis.

Duplicate or questionable records. Investigate possible duplicate submissions and unusual response patterns according to predefined data-quality rules. Do not remove observations without a defensible reason.

Example of a properly organized dataset

Respondent	FL1	FL2	FL3	FW1	FW2
001	4	5	4	3	4
002	3	4	3	4	4
003	5	4	5	5	4

Each row represents one respondent, and each column represents a variable or questionnaire item.

Once the data have been cleaned, save a separate copy of the original dataset. This preserves the raw data and makes it easier to document any subsequent changes.

3. Determine Whether Your Questionnaire Measures the Intended Constructs

For studies using multiple questionnaire items to measure a single construct, researchers should establish that the instrument’s measurement properties are appropriate for the intended use.

Reliability analysis and, where warranted, factor analysis can help assess the quality of the measurement instrument.

Reliability analysis

Cronbach’s alpha is commonly used to evaluate the internal consistency of a multi-item scale. An alpha of approximately .70 is often used as a preliminary guideline, but the appropriate interpretation depends on the research context, number of items, dimensionality, and other evidence.

A high alpha does not automatically prove that a questionnaire is valid or unidimensional.

Factor analysis

Exploratory factor analysis (EFA) may be appropriate when researchers want to explore the underlying factor structure of a set of items.

Confirmatory factor analysis (CFA) is appropriate when researchers want to test a theoretically specified measurement model, provided the design and data support that analysis.

For a straightforward descriptive study using an established questionnaire, a new factor analysis may not always be necessary. The decision should follow the study’s objectives and the evidence available for the instrument.

For researchers using SPSS or Jamovi, these procedures can be conducted through the software’s statistical menus or appropriate modules.

4. Choose the Appropriate Statistical Treatment

The statistical methods used in Chapter 4 should correspond directly to the research questions.

The following table provides a practical guide.

Research objective	Possible statistical method
Describe respondents’ characteristics	Frequency and percentage
Summarize scale scores	Mean and standard deviation, when appropriate
Compare two independent groups	Independent-samples t-test or Mann–Whitney U test
Compare three or more independent groups	ANOVA or Kruskal–Wallis test
Examine the relationship between two variables	Pearson or Spearman correlation
Examine predictive relationships	Linear or logistic regression, depending on the outcome
Test indirect or mediating relationships	Mediation analysis
Test conditional or moderating relationships	Moderation analysis
Evaluate a latent-variable measurement model	Confirmatory factor analysis
Examine relationships among latent constructs	Structural equation modeling

These are possible choices rather than automatic rules. The measurement level, research design, independence of observations, distributional characteristics, sample size, and assumptions must also be considered.

For example, a study investigating whether customer satisfaction predicts customer loyalty may use regression analysis when the outcome and model assumptions support it. A more complex study involving latent constructs, measurement errors, and multiple simultaneous relationships may benefit from structural equation modeling.

Using an advanced statistical technique is not necessarily better. The best method is the one that appropriately answers the research question.

5. Conduct the Analysis Using Statistical Software

Once the dataset is prepared and the statistical methods are identified, conduct the analysis using appropriate software.

SPSS

SPSS is widely used for descriptive statistics, reliability analysis, correlation, regression, group comparisons, and many other procedures. Its graphical interface can be particularly useful for researchers who prefer menu-based analysis.

Jamovi

Jamovi offers an accessible, menu-driven environment for statistical analysis. It supports many common procedures and provides tables that can be copied or adapted for research reporting.

JASP

JASP provides a graphical interface for classical and Bayesian statistical procedures, along with reporting-friendly outputs.

AMOS and other SEM tools

AMOS and other structural equation modeling tools are useful when a study requires confirmatory factor analysis or a structural model involving observed and latent variables. The appropriate software depends on the estimator, model complexity, and research requirements.

Regardless of the software, researchers should understand what the procedure calculates, why it is appropriate, and how its assumptions affect the interpretation.

Remember: software produces statistical output; the researcher remains responsible for the methodological decisions and conclusions.

6. Organize the Results According to Your Research Questions

A well-organized Chapter 4 should make it easy for readers to follow how the findings answer the study’s objectives.

A common structure is:

4.1 Profile of the Respondents

Present relevant demographic or background characteristics using frequency and percentage tables, as appropriate.

4.2 Descriptive Results of the Main Variables

Present the relevant descriptive statistics for the study’s major constructs or variables.

4.3 Reliability and Measurement Results

Include reliability coefficients and measurement-model results when these analyses are required by the research design.

4.4 Results of Hypothesis Testing

Present correlation, regression, group comparisons, mediation, moderation, or SEM results according to the study’s hypotheses.

4.5 Summary of Major Findings

Summarize the principal results in relation to the research questions and hypotheses. Avoid introducing results that have not been presented or analyzed.

This outline should be adapted to the institution’s thesis format and the actual objectives of the study. Not every thesis requires every section.

7. Present Statistical Results in Clear, Accurate Tables

Tables are useful because they allow readers to understand findings without having to navigate lengthy paragraphs.

However, a statistical table should include the information needed to interpret the results correctly.

For example, a correlation analysis might be reported as follows:

Table 1. Relationship Between Financial Literacy and Financial Well-Being

Variables	r	p	Interpretation
Financial literacy and financial well-being	.48	.002	Statistically significant positive relationship

Illustrative values only; replace them with your actual results.

The accompanying discussion might read:

“The results indicate a statistically significant positive relationship between financial literacy and financial well-being, r = .48, p = .002. This suggests that higher financial literacy scores were associated with higher financial well-being scores among the respondents. However, the correlational finding does not establish that financial literacy causes improvements in financial well-being.”

The interpretation should reflect the strength and direction of the relationship, the statistical evidence, and the limits of the research design.

When preparing tables, observe the required institutional or publication format, including appropriate decimal places, statistical notation, table titles, and notes.

8. Interpret the Findings Instead of Merely Repeating the Numbers

One of the most important parts of Chapter 4 is the discussion of the statistical results.

A common weakness in thesis writing is simply restating the values in a table without explaining their meaning.

For example, consider the following statement:

“The mean financial literacy score was 3.85, with a standard deviation of 0.62.”

This reports the result, but it does not fully explain its significance.

A more informative discussion would be:

“The respondents obtained a mean financial literacy score of 3.85 (SD = 0.62). Based on the study’s predefined interpretation scale, this result may indicate a relatively high level of perceived or assessed financial literacy. The finding should be interpreted in relation to the instrument used and the specific dimensions measured.”

Notice that the second statement does not automatically assign a verbal interpretation without reference to the study’s scale.

For inferential statistics, interpretation should address three questions:

1. What did the analysis find? Describe the direction, magnitude, and relevant statistical evidence.
2. What does the finding mean? Explain its relevance to the research question or hypothesis.
3. What can and cannot be concluded? Recognize limitations in the design, measurement, and analysis.

A statistically significant result is not necessarily practically important. Likewise, a nonsignificant result does not prove that no relationship or difference exists. Effect sizes, confidence intervals, sample size, and the precision of estimates can help provide a more complete interpretation.

9. Connect the Findings With Previous Studies

A strong discussion situates the findings within the existing body of knowledge.

After explaining a result, compare it with relevant studies identified in the literature review.

For example:

“The positive relationship between financial literacy and financial well-being is consistent with previous research reporting associations between financial knowledge and financial outcomes. This similarity may indicate that financial literacy is a relevant factor to consider when examining individuals’ financial well-being. However, differences in respondent characteristics, measurement instruments, and research designs should be considered when comparing findings across studies.”

This is an illustrative discussion, not a substitute for citing actual supporting studies.

When writing your thesis, use credible, verifiable academic sources. Explain whether your findings support, differ from, or extend previous research. If the results contradict earlier studies, investigate possible methodological or contextual explanations instead of forcing the findings to agree with the literature.

Avoid making causal claims from correlational evidence and avoid presenting speculative explanations as facts.

10. Use AI Tools Carefully When Writing Chapter 4

AI tools can help researchers understand statistical terminology, improve the clarity of their writing, organize a discussion, and identify questions that require further investigation.

However, AI-generated interpretations must be checked against the actual statistical output.

For example, if an AI tool is given a regression table, it may produce an explanation that sounds convincing but misstates the coefficient, ignores a nonsignificant result, or claims causation when the design does not support it.

To use AI responsibly:

• Provide the actual output and clearly identify the variables.
• Ask for an interpretation of the specific statistics reported.
• Verify all coefficients, significance values, confidence intervals, and conclusions.
• Check the interpretation against the study’s objectives and methodology.
• Verify citations against the original academic publications.
• Follow your institution’s rules concerning AI assistance, research integrity, and disclosure.

Never invent statistical results, fabricate references, or change an interpretation merely to make the findings appear more favorable.

AI should support the researcher’s judgment, not replace it.

11. Review Chapter 4 Before Submission

Before submitting your thesis to your adviser or panel, conduct a final review of the chapter.

Use the following checklist:

• Every research question has a corresponding result or an explanation of why it could not be answered.
• The statistical procedures are consistent with the approved methodology.
• Data have been checked for coding errors, missing values, and relevant assumption violations.
• Reliability and measurement analyses are reported where appropriate.
• Tables are correctly numbered, titled, and formatted.
• Numerical values in the narrative match the statistical output.
• Statistical significance is distinguished from practical importance.
• Interpretations do not claim causation without an appropriate research design.
• Findings are connected to relevant literature using verified sources.
• The actual results support conclusions.
• The chapter follows the institution’s thesis guidelines.

This review can help identify inconsistencies before the final defense, when correcting methodological and reporting problems may become more difficult.

Final Thoughts: Turn Statistical Output Into Meaningful Research Findings

Writing Chapter 4 is not simply a matter of running statistical tests and copying tables into a thesis. It requires a clear connection between research questions, data preparation, statistical methods, results, and interpretation.

Researchers who approach the process systematically are better positioned to produce findings that are accurate, transparent, and useful.

If you are struggling to determine which statistical test to use, interpret SPSS or Jamovi outputs, evaluate a measurement model, or develop a defensible discussion of your findings, seek guidance before finalizing your analysis.

At StatAce, we help students, graduate researchers, faculty members, and professionals navigate research design, questionnaire evaluation, statistical analysis, and the interpretation of research results. Our goal is to help researchers understand their data and present their findings with greater confidence.

Need help with your thesis data analysis? Visit StatAce.com to explore research and statistical consulting support for your academic project.