Analysing your research results involves checking data, choosing an analytical approach, performing calculations or analysing participant accounts, presenting findings and linking them to the research questions. In a bachelor's thesis or master's thesis, both correct numbers and the limits of interpretation are important. The description should not promise more than the collected materials allow you to establish.
In this guide, you will see how to process survey results for a thesis: from raw responses and calculations in Excel to an organised description. You will find examples of calculating percentages, a commentary on survey results, how to work with an interview and a checklist before sending the chapter to your supervisor. All sets of numbers and sample statements presented below are fictional educational materials, not the results of Student Support clients' research.
Where should you start when analysing research results?
Start by relating the data to your thesis aim and research questions. For each question, note which responses, interview extracts or documents will help answer it. Then determine which comparisons or calculations you will need.
For example, the question 'What difficulties do students face when preparing a bibliography?' may require an analysis of the response options selected about reference formatting, footnotes and citation rules. A question about differences between students on two degree programmes also requires information about their programme and a justified method of comparing the groups.
If it is unclear what was actually meant to be investigated, return to the guide on the aim, problems and hypotheses in the methodology of your own research. When processing results, you are implementing a previously adopted concept. Additional analyses can be performed, but they must be clearly distinguished from analyses planned before seeing the data.
Prepare in one place:
- the original file with responses or interview recordings;
- the questionnaire, interview guide or other instrument used;
- the aim, research questions and hypotheses, if adopted;
- a description of participant selection and the process of data collection;
- university guidelines and the supervisor's comments.
Do not start by choosing chart colours. First establish what you will show and why it is necessary for the thesis.
How to organise data before analysis?
Keep the original and work on a copy
Save the raw survey export without changes, and perform cleaning and calculations in a separate version. This allows you to return to the source material and check where specific values came from.
In a simple survey, one row in the spreadsheet usually corresponds to one person, and columns store answers to individual questions. Add a record identifier and a description of the coding, e.g. '1 = yes, 2 = no, 3 = hard to say'. These codes should not later be automatically treated as numerical values suitable for calculating an average.
Check the completeness and logical consistency of responses
Determine how many people started the survey, how many met the participation criteria and how many records you are using in a given analysis. Check for empty fields, out-of-range values, the functioning of filter questions and possible duplicate submissions.
Similar responses from two people are not yet proof of a duplicate. Do not delete records solely because they spoil the expected result or seem unusual. Data exclusion rules should be justified and described. If you make a decision only after seeing the data, note this fact.
Distinguish between a missing response and a 'no' answer
An empty field may indicate a skipped question, a technical error or a correct transition to another part of the survey. It is not automatically a negative answer. 'Hard to say' is also a separate response, not missing data.
For a question directed only at working people, the basis of analysis may be only the participants to whom that question applied. This limitation should not be hidden in the description.
If the problem stems from the form's structure, compare the applied solutions with the guide on preparing a survey for your own research. However, after data collection is finished, you cannot change the content of the asked question as if respondents had answered its corrected version.
How to calculate percentages in survey results?
In a simple summary without weights, you calculate the percentage as the number of people who selected an answer, divided by the chosen denominator and multiplied by 100. The base can be the entire analysed group or the number of people who gave a valid answer to the given question. You must clearly indicate which one was adopted.
If the project requires research weights, the calculations and reporting method will be different. The examples below concern unweighted data from simple survey summaries.
Example: single-choice question
Assume that 30 people answered the question: 'Did preparing a bibliography cause you difficulty?'. Each person chose one answer:
- Yes: 18 people, i.e. 60.0%.
- No: 7 people, i.e. 23.3%.
- Hard to say: 5 people, i.e. 16.7%.
The calculation for the first answer is: 18 / 30 × 100 = 60%. The counts sum to 30, and the shares to 100%. Use one rounding method within the summary. After rounding, the sum may sometimes be e.g. 99.9% or 100.1%; do not manually adjust a single value just to get exactly 100%.
Example: some people skipped the question
30 people participated in the study, but only 27 answered a specific question. 18 people marked 'yes'. This is 66.7% of the 27 respondents or 60% of the entire group of 30 people. Each result uses a different denominator and answers a slightly different question.
Possible wording:
27 out of 30 participants answered the question. 18 people indicated 'yes', i.e. 66.7% of respondents who provided an answer. Three people skipped this question.
Do not present 66.7% as the percentage of all participants. For subsequent questions, the base may change, so it is not enough to write once at the beginning of the chapter that the study covered 30 people.
Example: a question allowing multiple responses
In a question about difficulties with thesis writing, 30 people could select several options. Bibliography was selected by 18 people, footnotes by 15 and formatting by 12.
If you calculate the proportion of respondents selecting each option, the results are 60%, 50% and 40% respectively. They total 150% because one person could appear in several categories.
Add the following note below the summary:
Respondents could select more than one answer. Percentages were calculated relative to the 30 people who answered the question, so their sum may exceed 100%.
Another method is the share of a specific answer in all selections. In this example, there are 45 selections, so the bibliography accounts for 18 / 45 × 100 = 40% of selections. 40% of selections does not mean 40% of respondents. Choose the base appropriate for the purpose of the analysis and describe it consistently.
Practise the calculations with a worked example
The workbook contains 30 fictional responses: 18 'Yes', 7 'No' and 5 'Hard to say'. Change the responses in the Data tab, then check the recalculated counts, percentages and chart in the Results tab. Missing data are shown separately.
Percentages refer to valid responses to this question. After changing the data, adjust the comment on the results as well. You can find a sample description below.
How to process survey results in Excel step by step?
The following steps use the same fictitious frequencies as the previous example: 18 “yes” responses, 7 “no”, and 5 “hard to say”. Perform your own calculations on a copy of the data. The cell layout is an example to be reproduced independently, not a description of all form exports.
1. Organise responses and the code dictionary
Place data for one person in one row and the answer to one question in one column. Assume that responses to the question about difficulties with the bibliography are in cells B2:B31. Standardise the notation of the same answer, e.g. remove unnecessary spaces, while preserving the original export. Add a separate note about changes and the meaning of codes. A category number, e.g. 1 for “yes”, is a label, not a measure of the difficulty level.
2. Count each category
In C2, C3 and C4, enter the Polish example categories in order: tak (yes), nie (no), trudno powiedzieć (hard to say). In D2, enter =LICZ.JEŻELI($B$2:$B$31;C2) and copy the formula to D4. You will get 18, 7, and 5. This is the syntax for the Polish version of Excel; in the English version, the function is called COUNTIF, and the argument separator depends on settings. Check that every non-empty response belongs to one of these categories. Unknown entries must be explained rather than omitted without comment.
3. Establish the percentage base
For this question, the number of valid responses is the sum of the counts in three mutually exclusive categories. In D5, enter the following formula, using the Polish function name SUMA (SUM in English): =SUMA(D2:D4). In E2, enter =D2/$D$5, copy to E4, and set the percentage format with one decimal place. You will get 60.0%, 23.3%, and 16.7%. Do not multiply by 100 additionally if the cell has a percentage format. If the number of valid responses is zero, do not calculate percentages.
4. Separate missing data from negative responses
If the range covers exactly 30 participants, the number of missing responses is 30 minus the number of valid responses, provided you have first resolved invalid entries. If three people skipped the question and 27 valid responses remain, 18 'yes' responses represent 66.7% of those who answered. Also state that three people did not answer. Spreadsheet filters help you check the data; LICZ.JEŻELI (COUNTIF in English) does not, by itself, restrict calculations to rows visible after filtering.
Use separate columns for options in multiple-response questions
Prepare a separate variable for each option: 1 indicates selection, 0 indicates no selection by a person who answered the question. Separately mark people to whom the question did not apply or who skipped it. Do not automatically change them to zeros. With 30 respondents and 18 selections for bibliography, 15 for footnotes, and 12 for formatting, the respondent shares are 60%, 50%, and 40%. Their sum is 150% because one person could select several options. Do not use the total of 45 selections as the denominator if you are describing the percentage of people.
Transfer the summary to your chapter and check the commentary
Transfer category names, frequencies, and percentages to the table in the thesis, without formulas or participant identifiers. Add a number, title, percentage base, and information about missing data. You can use a bar chart to compare categories. Then check the consistency of the data with the comment shown in the next part of the guide. After changing the data, correct the description as well: the sheet will recalculate the numbers, but not the justification of conclusions.
How to write up survey results: example and outline
The comment should indicate the most important finding, provide its basis, and answer the research question. It does not need to repeat every value from the chart or always start with the words “from the above table it follows”.
For the bibliography example, you can write:
18 out of 30 surveyed students (60%) reported difficulties in preparing the bibliography. Seven people (23.3%) did not report such difficulties, and five (16.7%) chose the answer “hard to say”. This means that in the surveyed group, reported difficulties were more frequent than their absence. The question alone does not allow determining whether the problem concerned searching for sources, citation rules, or bibliographic formatting.
The final sentence sets the limits of interpretation. If the question only asked whether difficulties occurred, it cannot establish their cause.
For the next summary, use this outline:
- Scope: What does the question concern and which group was included?
- Result: Which result is most important for the research problem?
- Comparison: What other answers or differences are worth noting?
- Meaning: what can you establish, and what do these data leave unresolved?
The largest category is not always a majority
If 14 out of 30 people select an answer, that is 46.7%. It may be the largest category, but it is not a majority of the whole group. A majority means more than half.
Suppose also that 14 people selected 'probably yes' and 8 selected 'definitely yes'. Together, that is 22 out of 30 people, or 73.3%. Calculate the result from the sum of counts, not by adding already rounded percentages. Combining categories makes sense when both relate to the same direction of response; also indicate which categories were combined.
Tables and charts in a bachelor's or master's thesis – how to choose them?
A table is effective for precise numbers and comparing several values. A bar chart facilitates the comparison of categories. A line chart can show change over time if you have suitable measurements. Choose the presentation to fit the data type and the question you want to explain.
A pie chart is only appropriate for mutually exclusive parts of one whole. It is not suitable for percentages of respondents in a multiple-response question. A bar chart is often clearer when many proportions are similar.
Each summary should have a number, title, clear labels and an explanation of the calculation basis. Check your faculty's guidelines for the required position of the title and caption. One example title is 'Reported difficulties preparing a bibliography in the study group'; a note below the summary might read: 'Author's analysis based on survey research; n = 30; single-choice question'.
Pay attention to the details:
- for percentages, state the number of people they were calculated against;
- for several groups, show the size of each group;
- on the value axis of a bar chart, generally keep the start at zero to avoid exaggerating differences;
- avoid 3D effects, unnecessary legends, and too many colours;
- do not replace labels with colour alone;
- check readability after moving the chart into the document and exporting to PDF.
You do not need to present the same data simultaneously in a table, a chart, and a long paragraph. Choose a presentation that helps the reader, and explain in the text the finding that is significant for the thesis.
How to process open-ended responses and interview results?
Descriptive responses cannot be reduced to counting the most frequent words. You must analyse the meaning of the statements, the context, and the similarities and differences between participants. The choice of a qualitative approach should correspond to the research question and the adopted methodology, in line with the distinctions discussed by Creswell and Creswell in the textbook Research Design.
From statements to analytical categories
Start by reading the entire material. Then highlight fragments referring to the research problem and assign them working codes. Compare them across statements, organise them into categories, and check whether they accurately reflect the content of the material. The specific procedure depends on the chosen analysis method; not every project requires the same coding approach.
In an educational example of a study on difficulties with references, the statement “I know which book I used, but I do not know how to cite a chapter by another author” can be coded as a difficulty with describing a chapter in an edited volume. The statement “I cannot find the year of publication” may relate to missing source data. Both codes can belong to a broader category of problems with bibliographic recording, if such a division is justified in the analysis.
A quote illustrates the analysis but does not replace it
First present the category and explain its meaning, then illustrate it with a relevant fragment of a statement. Labels such as R1 or R2 help distinguish participants without giving their names. Before publication, also check details in the statement that might reveal the identity of the interviewee.
Do not select only quotes that fit the expected image. Include also different experiences and cases that complicate the interpretation. In qualitative research, an important theme does not have to be the most frequently mentioned one.
If you provide numbers, specify the unit of counting
'Ten accounts of difficulties' and 'ten people reporting difficulties' are different pieces of information. One person may return to an issue several times. When reporting category frequencies, explain whether you count people, responses or extracts from the material. Do not treat percentages from a small, purposively selected group as representative of the entire population.
Results, discussion, and conclusions – what to include in each part?
The results present what was established in the analysis. For example: 18 out of 30 respondents reported difficulties with references.
The discussion explains the significance of the findings, compares them with the literature, and considers limitations. You can check how your result relates to other studies, but first compare the groups, questions, and measurement methods. A similar percentage value alone does not prove that the results are directly comparable.
Once the data description is ready, proceed to the guide on writing the discussion of results. It shows how to combine your own findings with the literature and the study limitations.
The conclusions answer the research questions concisely. In the presented example, it is justified to state that a majority of the surveyed group declared difficulties. It does not follow from this that all students have this problem or that its cause is a specific way of conducting classes.
Separating these functions does not mean there is one mandatory document structure. On some courses, results and discussion are presented separately, while on others they are partially combined. Agree on the organisation method with your supervisor. The guide on the logical structure of a thesis.
How to relate results to hypotheses and when is statistics needed?
If your questions require group comparisons or assessment of associations, see the guide Statistical analysis in a master's thesis: choosing tests and describing results. You will find there an example of reporting a test, effect size, and confidence interval.
If you have hypotheses, link each one to specific data and the planned method of assessment. Do not write 'the hypothesis was confirmed' simply because one percentage is higher than another. Frequencies may answer a descriptive question, but assessing associations or differences may require additional analysis.
Test selection depends on the variable types, measurement method, independence of observations, group sizes and analytical assumptions, among other things. Two independent groups are treated differently from measurements of the same people before and after an event. No single test suits every master's thesis.
Interpret the analysis alongside effect size, uncertainty and study context. The p-value does not indicate how large or practically important an effect is. Nor is it the probability that a hypothesis is true. These limitations are addressed in the American Statistical Association statement described by Wasserstein and Lazar (2016).
Insufficient grounds to reject the null hypothesis do not prove that the relationship being studied definitely does not exist. A detected association likewise does not automatically imply causality. The conclusions must fit the study design, measurement quality and sampling method.
New questions arising after you see the results can be investigated through exploratory analyses. Make clear that they emerged at that stage. Do not present hypotheses added after the analysis as predictions made before the study.
Do you have quantitative data but need help comparing groups or investigating associations? As part of the service “Statistical analysis for bachelor’s and master’s theses” I will help select tests, perform calculations, and interpret the results. I will tailor the scope to your research questions and data quality; not every thesis requires hypothesis testing.
What to prepare if the work is to cover the entire research section?
- The original file with responses or other actual material, along with information about missing data and coding.
- The version of the questionnaire, interview guide or data sheet actually used in the study.
- The objective, research questions, any hypotheses, and a description of how the material was collected.
- Your current methodology chapter, agreed thesis structure and supervisor's comments.
The service Analysis and presentation of research results for your thesis may include processing the results, their discussion, and conclusions in the agreed chapter. If you need to combine this material with methodology, tool description, and a coherent discussion, check the service “Thesis research section: methodology, results and discussion”. The scope will be determined based on the documents; missing responses are not filled in with invented data.
How should you structure your research results chapter?
A practical starting point is the order of the research questions. First, briefly describe the material used in the analysis, and then show the results corresponding to the successive problems. This will not always be the same order in which the questions appeared in the questionnaire.
An example structure for a chapter concerning difficulties with bibliography:
- Characteristics of the analysed group and data completeness.
- Declared occurrence of difficulties in preparing the bibliography.
- Types of reported difficulties.
- Comparisons resulting from the research questions, if planned in the project.
- Summary of answers to the research questions.
- Limitations of the analysis and discussion, if the adopted structure places them in this chapter.
For each subsection, identify the data summaries you need and briefly note what the subsection should explain. If the data cannot answer a question, acknowledge the gap. Do not replace a missing measurement with your own assumption.
If you are still preparing the overall concept for your document, see the guide to thesis planning and preparation for the meeting with the supervisor. At the results stage, the plan can be refined while maintaining transparency of changes and consistency with the conducted study.
If the results are to go into a standalone report for a class, rather than a chapter of the thesis, check the service “Research projects”. The scope of the work will be tailored to the assignment and the actual materials.
Most common errors in processing results
Unclear basis for percentage calculation. The reader does not know whether the percentage refers to all participants, valid responses or the total number of selections. State the denominator with each summary.
Mixing units of analysis. The numbers of selections, people and participant accounts are not interchangeable. This applies both to multiple-response survey questions and to interview analysis.
Treating the most numerous category as the majority. An answer chosen by 40% of respondents may be the most frequent, but it does not cover more than half of the group.
Unchecked automated charts. Export from a form may hide missing responses, shorten labels, or use a different basis for percentage calculation than adopted in the text. Check the consistency of the chart with the source data.
Interpretation going beyond the question. A report of a difficulty alone does not establish its cause. A respondent's self-report is not the same as an independent measurement of behaviour.
Omitting results inconsistent with expectations. The description should include results significant to the problem, even if they do not support the author's predictions.
Changing hypotheses to fit the data. New observations can be discussed as exploratory. It should not be hidden when they were created.
Checklist before sending the chapter to the supervisor
- Do you describe the correct number of participants and explain the exclusion of records?
- Does each data summary clearly state the denominator used for percentages?
- Are the counts, percentages, charts and comments consistent with each other?
- Are questions allowing multiple responses clearly marked?
- Is the rounding method applied consistently?
- Do the comments address the research questions?
- Do you distinguish between findings from the data and possible explanations?
- Does the analysis of responses include differing viewpoints?
- Were the hypotheses evaluated appropriately, and are the limitations described?
- Are participants unidentifiable based on quotes and details?
- Do the titles, numbering and captions comply with the university guidelines?
Check the chapter after exporting to PDF as well. A chart that looks good in the spreadsheet may become unreadable when scaled down in the document. Large tables can be placed in the appendices, while the text should retain only those necessary to understand the key findings.
Questions on the analysis and description of research results
Is Excel enough for analysing survey results?
For organising data, simple summaries, percentages, and charts, a spreadsheet may be enough. More complex analysis requires appropriate tools and knowledge of their assumptions. The choice of software should be based on the task. The software itself does not assess whether the analysis method fits the research design.
Does every survey or thesis require statistical tests?
No. The scope of the analysis depends on the aim of the work and the research questions. Sometimes counts, percentages, and descriptive statistics are enough. If the study concerns differences or relationships, I check which methods are appropriate for the data and whether the conditions for their application are met.
A thesis based on interpreting regulations, a literature review, qualitative analysis of interviews, or an individual case study may not require testing statistical hypotheses. In qualitative research, the analysis may focus on the meaning of the statements. In that case, the appropriate support may be methodology, material analysis, or text editing. I do not add calculations just to make the thesis look more scientific or advanced.
Can the sum of percentages exceed 100%?
Yes, for example when calculating the percentage of people selecting each option in a question that allows multiple answers. The calculation method must be explained. For a single-choice question with mutually exclusive categories and the same denominator, the percentages should add up to about 100%, with a possible small difference due to rounding.
Is a description needed for every chart?
The result should be discussed in the text, but the commentary does not need to repeat every value. Highlight the key finding and its connection to the research question. Avoid including a series of charts without explaining why they are shown.
What to do if the hypothesis could not be tested?
Write what prevented its evaluation: lack of necessary measurement, incomplete data, or other limitations. Distinguish the inability to evaluate from a result that does not support the hypothesis. Agree on the next steps with the supervisor, without adding answers or changing the data.
Do you have the data but find it difficult to process the results?
Send the spreadsheet or interview material along with the research questions. I will help you determine how to organise the data, present the findings and describe them without drawing overly broad conclusions.
Analysis and presentation of research results for your thesis covers data summaries, tables, charts and commentary on the data. If you need to compare groups or assess associations, also see the service “Statistical analysis for bachelor’s and master’s theses”.
- Include the data without names, email addresses or other unnecessary participant identifiers.
- Provide the questionnaire or interview guide, as well as the objective and research questions.
- Indicate the supervisor's comments, any completed sections of the chapter and the deadline.
References
Creswell, J. W., Creswell, J. D. Research Design: Qualitative, Quantitative, and Mixed Methods Approaches. 6th ed. SAGE.
Wasserstein, R. L., Lazar, N. A. (2016). The ASA Statement on p-Values: Context, Process, and Purpose. The American Statistician, 70(2), 129-133. DOI: 10.1080/00031305.2016.1154108.
