Analyzing Your Dissertation Data

You have collected your data. Now comes the work of making sense of it. Data analysis is where the intellectual core of your dissertation takes shape – where raw numbers or transcripts are transformed into findings that answer your research questions. For many students, this stage feels daunting because it demands both technical skill and interpretive judgment. The good news is that if you planned your methodology carefully, you already know what analyses to run. The challenge now is executing them rigorously and interpreting them honestly.

This guide covers the practical dimensions of data analysis: preparing your data, selecting and conducting your analyses, interpreting your results, and presenting your findings. Whether your approach is quantitative, qualitative, or mixed methods, the principles here will help you move from raw data to a defensible results chapter.

Data Cleaning and Preparation

Before you analyze anything, you need to ensure your data is ready. Rushing into analysis with messy data is a recipe for errors that may not surface until your committee review – or worse, after publication.

Quantitative Data Preparation

For quantitative data, cleaning involves several steps. First, examine your dataset for missing data. How much is missing, and is it missing at random or systematically? A few missing responses scattered across variables is normal. A pattern where certain participants skipped entire sections, or where a particular question has a high non-response rate, signals a problem worth investigating.

Decide how you will handle missing data. Common approaches include listwise deletion (excluding any case with missing data), pairwise deletion (using all available data for each analysis), and imputation (estimating missing values based on other data). Each approach has assumptions and trade-offs. Consult your methodology literature and your advisor to choose the approach that is most defensible for your situation.

Next, check for data entry errors and outliers. Run frequency distributions and descriptive statistics for every variable. Are there values outside the possible range (a score of 6 on a 5-point scale)? Are there extreme outliers that could distort your results? Document every correction you make.

Recode variables as needed. If you need to reverse-score certain survey items, create composite variables, or transform variables to meet statistical assumptions, do this systematically and document each step. Create a codebook that defines every variable, its coding scheme, and any transformations applied.

Qualitative Data Preparation

For qualitative data, preparation means organizing your data for analysis. Transcribe all interviews verbatim if you have not already done so. Verify transcripts against recordings – even the best transcription services make errors, and automated transcription tools require significant cleanup.

Format your transcripts consistently. Include participant identifiers (codes, not names), timestamps or line numbers for reference, and clear marking of interviewer versus participant speech. Import your transcripts into your analysis software and create a filing system that allows you to locate any segment quickly.

If you collected field notes, observation records, or documents, organize these in the same system. Qualitative analysis often involves moving back and forth between different data sources, and having everything accessible in one place saves considerable time.

Choosing Your Analysis Methods

Your proposal should have specified your analysis methods, but the reality of your data may require adjustments. The key principle is alignment: your analysis methods must match your research questions and the nature of your data.

Quantitative Analysis Selection

For quantitative studies, your research questions and the type of data you collected determine which statistical tests are appropriate. Some common alignments:

Comparing group means on a continuous outcome calls for t-tests (two groups) or ANOVA (three or more groups). Examining relationships between continuous variables calls for correlation or regression. Predicting a categorical outcome from one or more predictors calls for logistic regression. Examining change over time calls for repeated measures ANOVA or growth curve modeling.

Before running your primary analyses, check the assumptions of each test. Normality, homogeneity of variance, linearity, independence of observations – each test has assumptions that must be met (or reasonably approximated) for the results to be valid. If assumptions are violated, you may need to use non-parametric alternatives or data transformations.

The Effect Size Calculator can help you compute and interpret the practical significance of your statistical results, which is essential for understanding whether your findings matter beyond statistical significance alone.

Qualitative Analysis Selection

Qualitative analysis methods are tied to your research tradition. Thematic analysis identifies patterns of meaning across a dataset and is flexible enough to work within multiple traditions. Grounded theory analysis uses iterative coding to build theory from data. Phenomenological analysis focuses on the essence of lived experience. Narrative analysis examines how participants construct meaning through storytelling.

Whichever approach you use, be explicit about your process. Qualitative analysis is sometimes perceived as subjective or unsystematic. Counter this perception by documenting every step: how you developed codes, how codes evolved into categories and themes, how you ensured your interpretations were grounded in the data rather than imposed upon it.

Mixed Methods Analysis

If you are using a mixed methods design, you need to analyze each strand (quantitative and qualitative) using appropriate methods, and then integrate the findings. The integration strategy should be defined in your methodology: Are you using the qualitative data to explain quantitative results (explanatory sequential)? Using quantitative data to test qualitative findings (exploratory sequential)? Collecting both simultaneously and comparing (convergent)?

The integration phase is where mixed methods studies produce their unique contribution. Do not treat your quantitative and qualitative analyses as two separate studies that happen to be in the same document. Show how the findings from each strand inform, confirm, contradict, or extend the other.

Software Options

Choose your analysis software based on your methods, your existing skills, and the resources available at your institution.

Quantitative Software

SPSS remains the most widely used software in education, psychology, and the social sciences. Its menu-driven interface makes it accessible for students without programming experience, and most university statistics courses teach it. It handles the analyses required by most dissertations comfortably.

R is a free, open-source statistical environment with enormous flexibility. If your analyses require advanced techniques (multilevel modeling, structural equation modeling, Bayesian analysis), R likely has a package for it. The learning curve is steeper than SPSS, but the investment pays off for students planning research careers.

Stata is popular in economics, political science, and public health. It excels at handling large datasets and complex survey designs.

Choose the software you can use competently or learn within your timeline. This is not the moment to teach yourself a new platform unless the analysis demands it.

Qualitative Software

NVivo is the most widely used qualitative data analysis software. It supports coding, memoing, query building, and visualization. It handles multiple data types (text, audio, video, images) and is well-suited for thematic analysis, grounded theory, and other common qualitative approaches.

Atlas.ti offers similar functionality with a somewhat different interface. Some researchers prefer its visual mapping tools for exploring relationships between codes and themes.

MAXQDA is another strong option, particularly popular in mixed methods research because it supports both qualitative and quantitative data within a single project.

Dedoose is a web-based alternative that works well for collaborative projects and is less expensive than desktop software.

You can also conduct qualitative analysis manually using word processing documents and spreadsheets. This approach is viable for smaller datasets but becomes unwieldy with more than a dozen or so transcripts.

Conducting Your Analysis

Quantitative Analysis Process

Run your analyses in a logical sequence. Start with descriptive statistics: means, standard deviations, frequencies, and distributions for all variables. These give you an overview of your data and help you check assumptions before running inferential tests.

Next, run your primary analyses – the tests that directly address your research questions. For each research question, run the specified analysis and record the results completely: test statistics, degrees of freedom, p-values, confidence intervals, and effect sizes.

Effect sizes deserve special attention. Statistical significance tells you whether a result is likely due to chance; effect sizes tell you whether the result is meaningful in practical terms. A statistically significant finding with a trivial effect size may not be worth much. Report both, and interpret your findings in light of both.

Finally, run any supplementary analyses your design requires: assumption checks, post-hoc tests, sensitivity analyses, or exploratory analyses. Clearly distinguish between confirmatory analyses (those specified in your proposal) and exploratory analyses (those conducted after seeing the data).

Qualitative Analysis Process

Qualitative analysis is iterative, not linear. You will cycle through your data multiple times, each pass deepening your understanding.

Initial coding: Read through your data and assign codes – short labels that capture the meaning of each segment. In your first pass, code broadly. Do not try to force data into predetermined categories; let the codes emerge from the data.

Focused coding: After initial coding, review your codes and begin grouping related codes into broader categories. Some codes will merge. Others will split. Some will prove irrelevant to your research questions and can be set aside.

Theme development: From your categories, identify overarching themes – patterns of meaning that address your research questions. A good theme is more than a topic; it captures something specific about what the data reveals. “Participants discussed mentoring” is a topic. “Mentoring relationships served as a bridge between institutional expectations and participants’ cultural identities” is a theme.

Verification: Check your themes against the data. Does the evidence support each theme? Are there disconfirming cases – instances where the data contradicts the theme? Honest engagement with disconfirming evidence strengthens your analysis and your credibility.

Document your coding process thoroughly. Save iterations of your codebook so you can show how your analysis evolved. Write analytical memos throughout the process – these become the raw material for your results chapter.

Interpreting Your Results

Interpretation is where analytical skill meets scholarly judgment. Your results are not self-explanatory – you must make sense of them.

Quantitative Interpretation

For each finding, ask: What does this mean in the context of my research questions? A significant positive correlation between variables X and Y means they tend to increase together – but does that align with what theory predicts? Does it match or contradict prior research?

Be cautious about overclaiming. Correlation does not imply causation (unless your design supports causal inference). Non-significant results are results – they mean you did not find evidence of the hypothesized relationship, not that the relationship does not exist. Report non-significant findings honestly. They are part of your contribution to the field.

For students working in health-related disciplines, interpreting statistical findings within the context of clinical significance and theoretical frameworks adds an important layer of meaning beyond the numbers alone.

Qualitative Interpretation

Qualitative interpretation involves moving from description (what participants said or did) to analysis (what it means). Your themes should not merely describe the data – they should offer an interpretation that connects to your theoretical framework and to the broader literature.

Use participants’ own words to ground your interpretations. Direct quotations provide evidence for your themes and give voice to your participants. Select quotations that are vivid, representative, and illustrative of the theme. Avoid over-quoting – the bulk of your results chapter should be your analytical narrative, not blocks of transcript text.

Presenting Your Findings

Tables and Figures

Well-designed tables and figures communicate findings more efficiently than text. For quantitative studies, present descriptive statistics in a summary table, create tables for each major analysis (with proper APA or discipline-specific formatting), and use figures to illustrate key relationships or patterns.

For qualitative studies, consider using a thematic map (a visual representation of how your themes relate to each other), a table summarizing themes with representative quotations, and diagrams that illustrate the processes or relationships your analysis revealed.

Follow your program’s formatting guidelines precisely. Table and figure formatting is one of the most common reasons final documents are returned for correction.

Writing the Results Chapter

Your results chapter should be organized by research question. For each question, present the relevant findings clearly and completely. In a quantitative study, report the statistical results and what they mean. In a qualitative study, present each theme with supporting evidence. In a mixed methods study, present the findings from each strand and then the integrated results.

Save interpretation and connection to the literature for your discussion chapter. The results chapter presents what you found; the discussion chapter explains what it means.

Common Pitfalls

P-Hacking and Data Dredging

Running numerous analyses and reporting only the significant ones is a serious ethical violation. Specify your analyses in advance (as you did in your proposal) and report all results, significant or not.

Ignoring Assumptions

Every statistical test has assumptions. Violating them can produce misleading results. Check assumptions before running each analysis, and report what you found. If assumptions are violated, use appropriate alternatives and explain your decision.

Thin Qualitative Analysis

A qualitative analysis that stays at the surface – identifying topics rather than themes, describing rather than interpreting – fails to realize the potential of your data. Push beyond the obvious. What patterns connect your participants’ experiences? What underlying processes explain the surface-level observations?

Over-Interpreting

The flip side of thin analysis is reading too much into your data. Let your findings speak proportionally to the evidence. If a theme emerged from only two of your fifteen participants, note that. If your effect size is small, say so. Intellectual honesty is your most important quality as a researcher.

Moving Forward

With your analysis complete and your results chapter drafted, you have the substance of your dissertation in hand. You know what you found. The next stage – writing and revising your dissertation chapters – is where you weave those findings into a complete scholarly narrative, connecting your results to the literature, exploring their implications, and acknowledging their limitations.