You've completed your survey on pressure ulcer prevention. Most respondents rated their knowledge positively, but their answers to the knowledge questions were more varied. In a paper you found for the discussion chapter, there is another percentage, obtained using the PZ-PUKT test. How can you discuss these findings in the same master's thesis? Start by adding a note to each figure saying what or who was counted and which instrument produced the result. The percentage of people who rated their knowledge positively, the percentage who achieved a particular score and the percentage of correct answers can all concern the same subject while answering three different questions. Show where the evidence supports a comparison between studies and how far that comparison goes.
This guide uses a fictional survey for the degree programme “Nursing” to take you from a spreadsheet of responses to a results section and a discussion that draws on Polish studies. You'll see what can be established when respondents both rate their own knowledge and answer questions with an answer key, but some skip one of those sections. The figures from the fictional survey are presented separately from the published findings.
What do you measure when you ask about knowledge of pressure ulcer prevention?
Start with your questionnaire. Check the exact wording of the questions and response options, and look back at what you agreed with your supervisor. Asking people how they rate their own knowledge of pressure ulcer prevention gives you their perspective. A question with one correct option according to an answer key lets you assess the accuracy of a response on a specific topic. Asking whether respondents use a risk assessment tool at work gives you a report of what they say they do. Each question describes a different aspect of the subject, even if the questions appear next to each other in the survey.
- Self-rated knowledge. The unit is one person's answer to a question about their own knowledge. The response options from 1 to 5 form ordered categories. In your results chapter, you can give the count and percentage for each one. If you combine options 4 and 5, keep the original distribution available and explain what the combined category represents. Even for the same person, ratings of readiness to act, familiarity with a procedure and overall knowledge can differ.
- Scores on researcher-developed knowledge questions. The unit can be a participant with a total score or an answer to an individual question. The interpretation depends on what the questions cover, how the response options are designed and what the key counts as correct. Ten questions on selected aspects of prevention tell you how participants performed on those ten questions. If most of the questions concern recognizing risk factors, a high total score primarily represents performance on that material.
- Scores from a published test. Along with the name, identify the version, the language and the particular score you are reading. The Polish adaptation of the 2021 PZ-PUKT covers prevention, stages of pressure ulcers and wound description [1]. If your survey focuses on prevention alone, the prevention subscale is the relevant part to look up. A subscale is the part of an instrument that assesses a particular area. The total score also includes answers from the other areas.
- Reported and recorded practice. A survey response saying that someone carries out an activity is a self-report. If you observe work or examine records, those are separate sources of data with their own units of analysis. A good test score by itself gives you no information about how participants act on shift or their patients' care outcomes.
Check who the question refers to as well. A respondent's rating of their own knowledge differs from their opinion of the knowledge of staff across a ward. Likewise, recognizing the name of a risk assessment tool and recognizing its purpose from a scenario are different tasks. This matters in the discussion even when both papers report the results under a heading such as “Knowledge.”
Try writing a short sentence next to each result without using the general phrase “level of knowledge.” For example, write “47 people selected one of the top two responses to the question about their own knowledge,” “43 people gave the answer in the key for item B” or “31 people scored at least 8 out of 10.” If a figure is difficult to describe in a concrete sentence, go back to the questionnaire and your calculation rules. Keep the same terms in your aim, results, discussion, abstract and figure captions.
How can you tell if a paper is suitable for comparison?
Open the full text and find the sections that describe the participants, the instrument and the outcome you're interested in. A title or abstract helps you identify a relevant paper. You will usually need the full text to find out how incomplete questionnaires were handled, what a subscale contains and how categories were defined. Note the page, table or figure where you found each detail. If the same result appears in several places, check that the counts and the denominator match. Record any remaining uncertainties before drafting a sentence that relies on the number.
Start with the numerator and denominator
For a percentage, write out the fraction. The denominator might be all participants, people with a valid answer, people who completed the test or the total number of scored item responses. If the authors report a mean percentage score, check how they calculated it. A value of 60% may describe the average proportion of the possible points that people earned. In another paper, 60% may be the proportion of people in a category defined by a range of scores. In the first case, the percentage describes points. In the second, it describes the share of people in a group.
A useful check is to turn the number back into an ordinary sentence. “117 of 203 people were in the 16-to-22-point category” explains a result from the study by Szymański, Porębska and Sipak-Szmigiel [2]. The numerator and denominator are both visible. “57.64% of answers were correct” would change the meaning even though it uses the same figure. If a paper leaves the basis for the percentage unspecified, record that gap and use the source to discuss the scope of the topic or how it was measured.
Check the instrument and the conditions of measurement
Two papers may use similar instrument names or the same abbreviation. Check whether they used the same version and whether the result describes the total score, a subscale or an individual item. Record the wording, language, response format and scoring rules. Shortening a test, changing the response options or dropping some items can affect how a score should be interpreted. For example, one study may offer an “I don't know” option, while another requires respondents to choose one of the substantive answers.
It can be tempting to convert two scales to a range of 0 to 100%. A result of 7 out of 10 can be expressed as 70% of the available points on that particular set of questions. Converting the scores on a longer instrument to a percentage also changes how they are displayed. The content, difficulty, coverage and scoring rules in each instrument still need to be checked. Identical percentages alone do not establish that a score of 7 out of 10 means the same thing as 70% on a broader test.
Look at the participants and the timing of the study
Record whether the study recruited students, practising professionals or a group that combined study and work. For students, the stage of the programme and the timing of the survey in relation to classes may matter. For professionals, useful information includes where they work, the type of work they do, whether they attended training and how they were recruited. If your sample comes from a single year group and a paper covers several year groups across two degree cycles, even a detailed result for the full published sample cannot tell you how your particular group would score. Look for relevant subgroup results. If there are none, identify how the samples differ and leave open whether that caused the observed difference.
Look at how the data were collected. People completing a questionnaire independently at home may be in different conditions from people doing so at a training session. A survey conducted immediately after classes and one conducted later may capture different points in students' learning. Describe the differences that are documented in the study methods. Ideas about motivation or access to materials can be questions for further investigation.
Match your wording to what the available evidence supports
If the outcome and calculations are the same and the groups are clearly described, you can report the values from both studies under the same precise label. Keep a descriptive comparison separate from statistical inference. A statement that a difference between samples is statistically significant requires an appropriate analysis and data that support it. Two published percentages will usually let you calculate an arithmetic difference in percentage points if both represent the same proportion. That calculation alone provides no estimate of uncertainty, leaves recruitment differences unresolved and cannot attribute the difference to a particular cause.
If the questions differ but the studies address a similar association, discuss its direction. You might say that higher self-ratings tended to accompany better answers in both studies, then describe how each study reached that finding. Different techniques can address related questions. A rank correlation describes a tendency across ordered results. The proportions who score at or above a chosen cut-off in two subgroups show how many fall on one side of that threshold. Both observations can support a comparison of direction, while the magnitudes remain specific to each analysis.
Some papers with different measures are best used for context. A paper on test properties helps explain the domains covered and how the test was documented. A study of self-assessment in long-term care can illustrate how another group rated its own knowledge. Giving each source an explicit role helps avoid putting several percentages in one paragraph and leaving the reader to look up what each one means.
How can you use three published Polish studies?
Rogowska and Rychlewska – describing self-rated knowledge
Katarzyna Rogowska and Monika Rychlewska studied 50 nurses in one long-term nursing and care facility. They collected data using a researcher-developed questionnaire. The prose accompanying Figure 6 expressly identifies the result as self-rated knowledge of pressure ulcer prevention. In the sample, 36% rated their knowledge as very good, 46% as average and 18% as low [3]. The caption refers to the level of knowledge, and the chart labels the 36% category “high”. In the accompanying text, the authors describe the results as self-ratings and call the highest category “very good.” Use the explanation in the text when interpreting the result. The paper also presents results from other questions, so record the particular result when you cite it.
If the figure of 58.8% in your study combines “good” and “very good,” the published figure of 36% covers only the highest category as defined by the authors. Subtracting one from the other applies arithmetic to two differently defined categories. You can say that both studies included self-ratings and then report the actual wording and circumstances in each. The worked example uses five response options. The published figure shows three categories, and the paper does not reproduce the question or the response options.
Szymański, Porębska and Sipak-Szmigiel – the direction of the association among students
Sławomir Szymański, Emanuela Porębska and Olimpia Sipak-Szmigiel studied 203 full-time nursing students in first- and second-cycle programmes at the Pomeranian Medical University in Szczecin. Their questionnaire contained 27 questions on the development, prevention and treatment of pressure ulcers. The authors reported a maximum knowledge score of 33 points. Their “sufficient” category corresponded to 16 to 22 points and included 117 of 203 participants, or 57.64% [2]. This is an example of how to read a percentage of people falling within a category defined by the researchers.
In their study, higher self-ratings tended to accompany higher knowledge scores [2]. You can compare that direction with your own observations if you have both measures for the same participants. In the fictional example, two groups are formed by combining self-rating categories, then compared using a score threshold. The published study reports an association based on correlation. Explain how both analyses worked and mention the wider scope of the published questionnaire and the inclusion of students from different years and degree cycles. An association between two ordered measures leaves open whether each person accurately estimated their knowledge on a common scale.
Przybek-Mita and colleagues – the scope of the 2021 Polish PZ-PUKT
Joanna Przybek-Mita and colleagues reported on the Polish adaptation and psychometric assessment of the 2021 PZ-PUKT. The study included 82 nurses practising in primary care and taking part in postgraduate training. The test contained 72 items. The authors included 31 on prevention, 21 on stages of pressure ulcers and 20 on wound description [1]. The paper presents findings for the instrument as a whole and for individual domains. If your interest is prevention, start by finding the row or passage that covers this domain and noting how the measure is defined.
In a thesis using your own questionnaire, you can describe the topics you share with the PZ-PUKT prevention subscale. Check the content and demands of the questions before comparing them. Asking people to identify the purpose of a risk assessment tool, for example, covers a narrower aspect of knowledge than the full prevention subscale. When discussing coverage, state which domains in the published instrument fall outside your study aim. This helps explain why the published percentage offers no cut-off for your own questions.
If you used the PZ-PUKT in your study, record the authors, version, language, domains used, conditions of use and the source of the scoring rules. Report relevant information on the measurement properties. Przybek-Mita and colleagues estimated the internal consistency of the full test for that instrument in their sample [1]. Applying that figure to a shorter questionnaire you developed gives no evidence about the properties of your questions. If you did change an instrument, agree with your supervisor how to describe the change and what comparisons the modified version can support.
What should you note down from a paper before writing the discussion?
One short note per source is enough. Record where you verified the result in the full text and the sentence it can support. For Rogowska and Rychlewska, the note would be about self-rated knowledge in a long-term nursing and care facility, as presented with Figure 6. For Szymański and colleagues, record separately the proportion in the 16-to-22-point category and the direction of the association between self-rating and score. For the PZ-PUKT paper, note the version of the instrument, the scope of the prevention subscale and the participant group. You can then trace each comparison back to the relevant part of the paper and the instrument that produced it.
When a study refers to a “sufficient level,” check who defined the category and on what basis. It may be a rule from an instrument manual, a division chosen by the researchers or a category that participants selected directly when rating themselves. In the fictional example, “at least 8 out of 10 points” is a descriptive category. If your own survey did not specify a threshold, the score distribution will often answer the question about performance without adding a new division after you have seen the results.
The fictional survey – from 84 questionnaires to 76 pairs of responses
In this example, a master's student collected 84 questionnaires from a single year group in a second-cycle nursing programme. Participation was voluntary, and data were collected once after a teaching block had ended. The questionnaire included one self-rating question and 10 researcher-developed questions on selected aspects of pressure ulcer prevention. Some concerned recognizing risk factors, understanding the purpose of risk assessment and understanding why it is documented. You can follow the analysis without knowing the clinical content of the individual questions. All example responses and counts are fictional.
The main research question is how students rate their knowledge and how they perform on the set of questions. A secondary question asks how self-ratings relate to scores in the same questionnaires. This distinction determines how many responses can enter each analysis. A valid answer to one question is enough to describe self-ratings. Calculating the total score requires a response to every item. To compare the two measures, you need a self-rating and a complete score from the same person.
Define what counts as a response
Self-ratings were recorded in five categories, from 1 “very poor,” 2 “poor” and 3 “average” to 4 “good” and 5 “very good.” For the set of ten questions, each answer that matched the key earned one point. Both an incorrect answer and a selected “I don't know” received zero points. A blank field was recorded as a missing response. A total score from 0 to 10 was calculated only for questionnaires with a response to all ten questions. For an individual item, partially completed questionnaires could be included if they contained a response to that item.
This rule was set before the example analysis. The original answers were retained, and variables for calculation were created from them. The distinction matters when a spreadsheet shows a zero. In the score column, zero means an incorrect answer or an explicit “I don't know.” A blank in the original response column means that the participant selected nothing. Keeping the raw response allows you to count “I don't know” separately later and check whether particular questions were skipped more often than others.
Count how many questionnaires enter each analysis
In the example, 76 people gave a self-rating and answered all the knowledge questions. Four gave a self-rating but left the knowledge section incomplete. Two completed the knowledge section but skipped the self-rating. A further two skipped the self-rating and left part of the knowledge section incomplete. The four counts add up to 76 + 4 + 2 + 2 = 84. There were 80 people in the analysis of self-ratings, 78 in the analysis of total scores and 76 in the analysis of paired results. For individual questions, the denominator was the number of valid responses to that specific item.
In practice, start by marking whether each questionnaire has a valid self-rating and the full set of responses needed to calculate a score. Two columns with “yes” or “no” are enough at this stage. If you store the two sections separately, match their data using the same questionnaire identifier where you can do so without linking responses to a name. If all you have are two separate lists of summary percentages, you cannot work out who gave a particular self-rating and how many points they earned. You can still report two valid distributions. The question about their association remains open.
Present the results that answer your main research question
The self-rating analysis used 80 valid responses. There were 8, 12, 13, 33 and 14 people in categories 1 to 5 respectively. A total of 47 out of 80 people chose one of the two highest options, or 58.8%. Dividing 47 by 84 would give the share of all questionnaires with a category 4 or 5 response. To describe the distribution of valid self-ratings, the appropriate denominator is 80. Report the four missing responses alongside that figure so the reader can reproduce the calculation.
Among 78 complete sets of answers, the median score was 7 out of 10. The median indicates the middle of the ordered scores. The first and third quartiles were 6 and 8 points respectively. They mark the limits of the middle part of the distribution. The observed scores ranged from 2 to 10. A total of 31 of 78 people scored at least 8 points, or 39.7%. In the example, this additional figure shows the upper part of the distribution. The threshold gives no classification of professional competence. Comparing 58.8% with 39.7% alone says nothing about how individual self-ratings related to the scores of the same people.
For item B, on the purpose of a risk assessment tool, 43 of 81 valid responses matched the key, or 53.1%. Twelve respondents selected “I don't know,” 26 gave an incorrect answer and three left the field blank. Item C, which was about identifying the purpose of documenting risk assessment, had 83 valid responses, of which 66 matched the key, or 79.5%. One person skipped it. The percentage for item C describes responses to that item and can include a questionnaire with no response to item B. State this rule once in your report so the reader can understand why one analysis uses 78 and the other two use 81 and 83.
When interpreting the scores, take the design of both questions into account. If item C has a shorter description or more distinct response options, its format could partly explain the higher proportion of correct responses. If risk was covered in more detail in class, this may help you plan further research. You would need to confirm what was taught and how. You can describe the distribution of answers on particular items precisely while leaving the explanation of that pattern open.
Compare the two responses in questionnaires with complete pairs
Of the 76 people with a pair of results, 45 selected category 4 or 5 and 31 selected category 1, 2 or 3. In the first group, 22 people scored at least 8 points, or 48.9%. In the second group, 8 people achieved that score, or 25.8%. The difference was 23.1 percentage points. These percentages were calculated within each self-rating group. Dividing 22 and 8 by 76 would instead give each combination's share of the entire paired sample. That is a different question from how often people in each self-rating group scored at least 8.
The other combinations also tell you something. Of the 45 people who chose 4 or 5, 23 scored below 8. Of the 31 who chose 1, 2 or 3, 23 scored below 8. The two score distributions overlap. The four paired counts, 22, 23, 8 and 23, add up to 76. There were 31 people with a score of at least 8 in the analysis of all complete scores and 30 in the analysis of pairs. Of the two people with complete scores and no self-rating, one scored 8 and the other scored 6. Both are counted in the overall distribution of scores and neither belongs to a self-rating group.
Combining responses into two groups makes the result easier to describe and hides differences between individual categories. In the example, the group from 1 to 3 covers responses ranging from “very poor” to “average.” If your research question asks whether scores change as self-ratings increase, you can present score distributions for all five categories and consider an analysis that preserves their order, in line with your research plan. The direction of the association can be studied using individual data and a method suited to the measures, such as a rank correlation. The counts for two combined groups are insufficient to reconstruct a coefficient for the five original response categories and the full scores.
Check where missing data might matter
Scores are known for 78 of the 84 people. A simple check can show how the six incomplete sets might affect the proportion with at least 8 points in the full sample of 84. If none of those six people would have reached the threshold, 31 of 84 people would have scored at least 8, or 36.9%. If all six could have reached it, the upper bound would be 37 of 84, or 44.0%. This second assumption deliberately allows a broad upper bound. If you have partial responses, check whether anyone would still fall short of 8 even with correct answers to all omitted questions. Both scenarios are bounds based on explicit assumptions. The observed result for complete sets is still 31/78.
Where the data allow, check missing responses in the two self-rating groups. In the example, four people gave a self-rating but did not complete the knowledge questions. Two chose category 4, one chose category 1 and one chose category 2. This gives two incomplete sets among the 47 who rated themselves 4 or 5 and two among the 33 who chose 1, 2 or 3. The counts are small, and the description alone leaves the reasons for missing responses unknown. It does show whether the results needed for the comparison were available for similar proportions of both groups. Substantial missing data may call for a fuller assessment and further discussion of the analysis plan.
Sample master's thesis passages based on the example
The Faculty of Health Sciences at the Medical University of Warsaw gives separate roles to the methods, results and discussion chapters in its guidance for second-cycle nursing theses. The discussion relates the findings to the aim and literature and considers similarities and differences [4]. This is published guidance from one university. Agree on your own thesis structure with your supervisor using the requirements of your programme. For help reporting a cross-sectional study, see the STROBE checklist items about missing data, participant numbers at each stage and limitations [5].
Sample methods passage
The aim was to describe how students in a second-cycle nursing programme rated their own knowledge of pressure ulcer prevention and how they answered ten researcher-developed knowledge questions. The analysis also examined how often participants in different self-rating groups scored at least 8 points. The study was cross-sectional and included 84 people from one year group. Self-ratings were recorded in five ordered categories from “very poor” to “very good.”
A response that matched the answer key received 1 point. Incorrect answers and “I don't know” responses received 0 points, and omitted items were recorded as missing. Total scores from 0 to 10 were calculated for complete sets of responses. Percentages for individual items were calculated using the number of valid responses to that item. The analysis of paired measurements included questionnaires with a valid self-rating and a complete set of responses to the knowledge questions. The results for people who rated themselves 4 or 5 were reported separately from those for people who rated themselves from 1 to 3. The threshold of 8 points was used to describe the upper part of the distribution of scores on the researcher-developed questions.
In an actual thesis, describe how you recruited participants, when you collected responses, which inclusion criteria you used, what sources informed the answer key, and how and why you handled missing data as you did. If specialists reviewed the questions, you ran a trial completion or you conducted a pilot study, say what you checked and what changed before data collection. This helps readers assess how the questions were developed. Use your own study records to describe what you did and what is known about the instrument.
Sample results passage
A total of 84 questionnaires were returned. Eighty people gave a valid response to the self-rating question. Eight rated their knowledge as very poor, 12 as poor, 13 as average, 33 as good and 14 as very good. A total of 47 of 80 people (58.8%) selected “good” or “very good.”
A total score for the knowledge questions was calculated for 78 participants. The median was 7 out of 10 points, and the first and third quartiles were 6 and 8 points respectively. The observed scores ranged from 2 to 10. Thirty-one participants achieved at least 8 points (39.7% of complete sets). For item B, 43 of 81 people (53.1%) gave an answer that matched the key. Twelve selected “I don't know,” and three questionnaires had no response to this item. For item C, 66 of 83 people (79.5%) gave the answer in the key.
The comparison of self-ratings and scores included 76 participants. Among people who selected a self-rating of 4 or 5, 22 of 45 (48.9%) scored at least 8 points. Among people with a self-rating from 1 to 3, 8 of 31 (25.8%) achieved that score. The descriptive difference was 23.1 percentage points.
Here the reader can check where the two percentages came from and whether the denominators reflect the same stage of analysis. State how you report quartiles or dispersion throughout the chapter, especially if you use more than one piece of software. Programs may use different conventions for calculating quartiles, which can matter when groups are small and values repeat. Reporting the convention makes the calculations easier to reproduce. The results of the fictional example can be understood from the text without a graph.
Sample discussion passage
In this sample, 47 of 80 people described their knowledge as good or very good. Among people with both measures, a score of at least 8 was more frequent with these self-ratings than with ratings from 1 to 3. Both groups included scores below 8, and some of those who rated themselves lower also scored at least 8. The two measures capture different aspects of the subject, self-rated knowledge and performance on a set of knowledge questions.
This pattern is similar in direction to the finding by Szymański, Porębska and Sipak-Szmigiel [2], who described an association between higher self-ratings and higher knowledge scores in a sample of nursing students. They examined students in several years of both degree cycles and used an instrument that also covered the development and treatment of pressure ulcers. The present comparison concerns a single year group in a second-cycle programme, ten questions about selected aspects of prevention and groups based on self-ratings. The studies can be compared in terms of the direction of the observation. The size of the association is specific to each analysis.
Rogowska and Rychlewska [3] provide a point of reference for self-rated knowledge. The text accompanying Figure 6 states that 36% of the participants employed at the long-term nursing and care facility rated their own knowledge of prevention as very good. The figure of 58.8% from the present study combines “good” and “very good” from a different set of response categories. Both studies show how people perceived their knowledge in their own settings. Differences in response categories and sample composition limit the use of a single measure to quantify a difference between self-ratings.
The Polish adaptation of the 2021 PZ-PUKT [1] illustrates which items concern prevention and how the full test covers a wider range of topics. The present study used a shorter, researcher-developed set that addressed selected aspects of prevention. Differences in scope and structure between the instruments limit the comparison to the content covered by the measurements. The PZ-PUKT data also came from practising primary care nurses who attended postgraduate training. Comparing percentages alone would leave open whether characteristics of the sample, the questions or the measurement conditions contributed to any difference.
The study was limited to one year group and one measurement after a teaching block had ended. The results describe this group's responses. With no baseline, they provide no estimate of change during the programme. The analysis of self-ratings and scores included 76 of 84 people. Those with incomplete responses may have differed from those included in the paired analysis. Further research could use the same instrument under comparable conditions across several year groups to see whether the pattern persists.
You can summarise the answer in a single conclusion paragraph. For example, “In this year group, good or very good self-ratings were more common than all other responses combined. Among participants with both measures, a score of at least 8 out of 10 was more common among people who rated their knowledge 4 or 5. The association described here applies to the researcher-developed questions and the students in this year group.” If you write your conclusions as a list, connect each one to a research question. Repeating every result can make the answers to those questions harder to see.
How to revise your discussion after a supervisor's feedback
“Please compare your findings with the literature. This reads like a description.”
Identify which result the comment refers to. If you're describing an association within your sample, start with a paper that also examined both measures in the same participants. State what the observations have in common, then explain the most important difference in the instruments or the samples. In the example, that paper is the study by Szymański and colleagues. Rogowska and Rychlewska help explain self-ratings in a different group. Przybek-Mita and colleagues help you discuss what an instrument covers. Three papers can serve three different purposes in the same discussion even if all their titles concern knowledge about pressure ulcers.
When revising the chapter, select a claim and check it against the full text. “My group achieved higher scores than the participants in the 2020 study” requires a shared score or another justified common basis. “In both samples, higher self-ratings tended to accompany higher scores” requires you to check that each study actually related the two measures. With the wording revised, you can also report the counts behind the comparison in your own sample so the reader can see where the observation comes from.
“Can I say the respondents overestimated their knowledge?”
That claim requires linking self-ratings to a well-justified reference level. Explain what participants were expected to mean by “good” and how a test result could correspond to the same knowledge category. This is a different research question from whether the groups differed in how often they achieved a particular score. The fictional example shows overlapping distributions. You can report that 23 of the 45 people who rated themselves 4 or 5 scored below 8. That result leaves open how accurate their self-ratings were.
If your survey contains only a self-rating question, limit your conclusion to the distribution of self-ratings. If you have only knowledge questions, describe the results of those questions. If you have both measures but cannot match the responses, report each distribution separately and explain that the data cannot be paired. With complete pairs, start by describing what the data show, including any overlap. A stronger claim requires its own measurement strategy and interpretation rules.
“Another paper reached the opposite conclusion. Why might that be?”
First establish whether the studies concern the same association. A lower total test score in one study and a weak association between self-ratings and scores in another are different kinds of results. If you're comparing the direction of the same association, report the findings and explain the limits of the comparison. Identify a difference in the tools or conditions if you can verify it in both methods sections. Leave open which differences mattered unless the data show their role. With small groups, variation across estimates can also reflect measurement uncertainty and chance.
Apply similar care when discussing training. A measurement taken once after classes shows the responses at that time. If you compare people who report attending training with those who did not, the groups may also differ in experience or working conditions. To assess change over time, you need measurements at different points. Attributing that change to training also requires a justified study design and analysis. In a cross-sectional study without those data, report an association with reported training attendance if you actually examined it.
“I couldn't find a study identical to mine.”
Look for studies that share one specific feature with yours. Combine search terms for the population, subject and instrument. For example, try the Polish terms “studenci pielęgniarstwa,” “profilaktyka odleżyn” and “samoocena,” or the English terms “nursing students,” “pressure injury prevention” and “knowledge.” If you are looking for results from a particular instrument, search for its full name and version. Record whether a source helps you interpret the group, instrument, direction of association or study conditions. A comparison focused on one feature tells readers more than saying that all the papers found a similar “level of knowledge.”
If you need to structure the rest of your thesis and link it to the discussion, see the guide to methodology, analysis and discussion in a nursing master's thesis.
Review your chapter before submitting it
- Find the numerator and denominator for every percentage. Make sure readers can tell whether each figure represents a person, a response to one question or points.
- Check which participants contributed to the analysis of self-ratings and scores together. To describe an association at the individual level, include only questionnaires that can be matched and have both measures available under your analysis plan.
- Check how you handled “I don't know,” incorrect answers and blank fields. Make sure missing data are treated consistently in the spreadsheet, the results chapter and the abstract.
- Locate the exact passage supporting each finding you cite. Note the instrument version, scope, categories and population. Check whether the source discusses prevention, the total score or a self-rating.
- Mark each sentence in the discussion that starts with words such as “higher,” “lower,” “similar” or “influenced.” Check which results it connects, what data support it and whether the design can support a claim about causes.
- Answer the research questions in your conclusion. For researcher-developed questions, state what they covered. A survey may suggest another research question about clinical practice or teaching outcomes if your study did not observe them.
Questions about comparing survey results with the literature
What should a research instrument description in a thesis include?
State the instrument’s name and origin, purpose, language version, structure, response options and scoring rules. Explain its conditions of use, checks performed and limitations. Describe the questionnaire actually used rather than a revised version created afterwards.
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.
How can I connect a limitation to its effect on the conclusions?
Identify a specific feature of the procedure, the finding or claim it affects, and the consequence. Then explain what you actually did to address it and what remains uncertain. A generic list mentioning sample size and time pressure is not enough.
How does preparing the research chapter differ from statistical analysis?
The research chapter combines a description of the actual study, presentation of results, interpretation and answers to the thesis research questions. Statistical analysis concerns the choice of methods and calculations on quantitative data. You can request both types of support together, but I agree any new tests and verification of calculations separately after reviewing the material.
Can an already prepared research chapter be improved?
Yes. I can check the consistency of tables, charts, calculations, comments, and conclusions, and adjust the layout to the supervisor's comments. To verify the accuracy of the results, I will also need the source data and the research instrument.
What should I send for a quote?
Provide your topic, subject area, type of thesis, stage of preparation and required deadline. Describe what needs to be prepared, checked or revised. Attach the current document, university guidelines and any comments from your supervisor.
For research support, include your research outline, survey or interview guide and the data you have collected. For editing that addresses text similarity, attach your JSA report if you have one. You don't need every document before asking about the scope of support.
Sources
- [1] Przybek-Mita J., Bazaliński D., Małek E., Kozieł D., Kachaniuk J., Kózka M., Szewczyk M.T. (2024). Psychometric properties of the Pressure Ulcer Knowledge Test (PZ-PUKT) among nurses. Medical Studies, 40(3), 248-262. DOI
10.5114/ms.2024.141694.https://www.termedia.pl/doi_ft/10.5114/ms.2024.141694 - [2] Szymański S., Porębska E., Sipak-Szmigiel O. (2020). Wiedza studentów pielęgniarstwa na temat profilaktyki i leczenia odleżyn. Co wiemy na temat odleżyn? Polski Przegląd Chirurgiczny, 92(3), 22-25. DOI
10.5604/01.3001.0014.0508.https://bibliotekanauki.pl/articles/1391704.pdf - [3] Rogowska K., Rychlewska M. (2019). Wpływ kształcenia podyplomowego na rozwój kompetencji zawodowych pielęgniarek opieki długoterminowej. Innowacje w Pielęgniarstwie i Naukach o Zdrowiu, 4(3), 43-58. Figure 6 and the accompanying text, pp. 54-55. DOI
10.21784/IwP.2019.015.https://wyd.edu.pl/images/Czasopisma_naukowe/PPA/IwP2018/IwP_2019_1_4/IwP_3_4_2019/IwP_2019_3_calosc.pdf - [4] Medical University of Warsaw, Faculty of Health Sciences. Jak napisać pracę magisterską? Zalecenia dotyczące pisania pracy magisterskiej. Guidance for second-cycle nursing students. Part IV, chapters on methods, results and discussion.
https://wnoz.wum.edu.pl/sites/wnoz.wum.edu.pl/files/pielegniarstwo-studia-ii-stopnia-zalecenia-dotyczace-pisania-pracy-magisterskiej.pdf - [5] STROBE Initiative. STROBE Statement. Checklist of items that should be included in reports of cross-sectional studies. Items 12(c), 13, 14(b), 19 and 20.
https://www.strobe-statement.org/download/strobe-checklist-cross-sectional-studies-pdf
Need help writing the results and discussion for a nursing thesis?
Send your topic, research questions, supervisor's requirements, questionnaire and a summary of responses with identifying data removed. Include the papers you want to compare with your results and your draft discussion if you have one. Tell us whether you need help organising the survey analysis, describing missing data or preparing a sample passage for the research section.
Contact StudentSupport


