ACADEMIC WRITING GUIDE
How to Present Research Data in Tables and Figures (With Worked Examples)
A research project can contain sound data and careful analysis yet still leave readers confused. Perhaps the table has unexplained abbreviations. Perhaps the chart uses a truncated axis that exaggerates a small difference. Or perhaps the results chapter repeats the same numbers in a paragraph, a table and a graph without explaining why any of them matter.
Good data presentation solves a specific problem: it helps your reader see the evidence accurately, find the quantities that matter and understand what those quantities can—and cannot—support. You do not need elaborate graphics. You need well-chosen displays, honest scales, complete labels and restrained explanatory writing.
This guide shows you how to make those decisions and then walks through three hypothetical research examples. It focuses on presenting analysed data; if you are still checking duplicates, missing observations and inconsistent codes, begin with our research data-cleaning guide.
All numerical examples in this article are invented teaching datasets, not actual study findings. Follow your department’s presentation conventions and required style guide; APA examples are illustrative rather than mandatory for every discipline.
What is data presentation in research?
Data presentation is the organised communication of observations, summaries or analysis results using prose, tables and figures. A table arranges values in rows and columns; a figure may be a chart, diagram, graph, map or other visual. Prose tells readers where to look and what is important. The aim is faithful communication rather than decoration.
The American Psychological Association notes that effective tables and figures help readers understand substantial information efficiently. But a display is only useful when it serves a clear question and is accessible to its intended audience (APA Style, n.d.).
Table, figure or prose: which should you choose?
| Use | Best when | Potential problem |
|---|---|---|
| Prose | Only one or two values matter and their meaning is simple | A long string of numbers becomes hard to compare |
| Table | Readers need exact values, sample sizes, categories or multiple estimates | Too many columns, unexplained units or redundant detail |
| Bar chart | Comparing counts or proportions across discrete groups | Misleading baseline or inconsistent category denominators |
| Line chart | Showing ordered observations over time | Implying trends in irregularly measured data |
| Scatterplot | Showing paired numerical observations and association | Claiming a causal effect from a visual relationship |
| Distribution plot | Showing spread, skew or variation among continuous values | Hiding variability behind a single mean |
A table and a figure can sometimes both be justified—for example, a graphic shows the shape of a pattern while a supplementary table gives exact estimates—but do not duplicate information automatically.
Nine steps for clear, defensible research tables and figures
Step 1: Begin with the research question and the result
Write one sentence describing what the display must communicate before opening Excel, SPSS or another application. For example: “Show how the preferred feedback method differs across three response categories,” or “Show attendance across five monthly observations.” The display should answer that sentence.
A graph prepared only because the software produced it is unlikely to improve the chapter. Our quantitative analysis guide explains the separate task of choosing and interpreting statistical procedures; here the task is communicating their outcomes.
Step 2: Identify the variable type and the comparison
Counts and proportions across categories usually suit a table or bar chart. Numeric measures observed across time may suit a line plot if time points are meaningful. Two paired quantitative measurements may suit a scatterplot. A continuous variable’s distribution may need a histogram, dot plot or boxplot rather than a bar showing only its mean. The chart must match the data, not just the software menu.
Step 3: Verify your denominator and sample size
Percentages are not meaningful without a clear denominator. If 90 students responded to a question, 42 of 90 is 46.7%. If only 75 answered the next question, percentages for that question must use 75 unless there is a justified alternative. Label valid n, exclusions and missing responses where they affect interpretation.
For comparisons among groups, check whether percentages represent proportions of the entire sample, within-group proportions or multiple-response selections that can sum beyond 100%. Do not mix them without explanation.
Step 4: Use informative titles, numbering, and notes
A strong table title identifies what was measured and who or what is represented. “Table 2. Feedback preferences among 90 respondents” is more useful than “Table of findings.” Number tables and figures in the order required by your institution. State the unit, basis of percentages, sample size, and any abbreviations in the title, headings or notes.
APA-style conventions ordinarily place a table number and italicised title above a table; figure arrangements and notes follow the specified style. Check the current manual and departmental rules rather than copying formatting from an unrelated thesis (APA Style, n.d.; Purdue OWL, n.d.).
Step 5: Show the data accurately
Bar lengths normally start at zero because readers interpret their relative heights or lengths. Consider a carefully labelled alternative for specialised graphs, but never shorten an axis silently to make a modest difference appear enormous. Use sensible scales and intervals. For line or scatter plots, a nonzero axis may sometimes be appropriate; make its limits explicit and ensure the display does not distort the claim.
Keep time points in chronological order, show uncertainty when relevant, and do not join observations with lines when a continuous trajectory is unsupported. Research on scientific graphics repeatedly warns that poor graphical choices can mislead even when the numbers themselves are correct (Weissgerber et al., 2019; Midway, 2020).
Step 6: Show variation and uncertainty when they matter
A mean alone may conceal how widely participants differ. For continuous measurements, consider displaying the distribution, group size, median, standard deviation or interquartile range as appropriate. For estimated group comparisons, confidence intervals may communicate uncertainty better than an asterisk for statistical significance alone. Never draw a confidence interval you have not calculated or report inferential claims you have not tested.
Step 7: Design for legibility and accessibility
Use readable labels, consistent number formatting and adequate text size. Do not rely on colour alone; use direct labels, distinct patterns or shapes where useful. Ensure tables remain understandable to someone navigating rows and columns with assistive technology and provide meaningful alternative descriptions for graphical figures in the published format.
Avoid tiny legends, patterned backgrounds, unnecessary 3D effects and dense decorative grid lines. Scientific communication benefits when visual clutter is reduced without removing context needed to interpret the display (Midway, 2020).
Step 8: Write prose that adds interpretation rather than reciting cells
Introduce a display before or near its first appearance, then identify the result that matters. Instead of repeating five rows of counts, write: “Written feedback was the most frequently selected option, representing 46.7% of the 90 respondents (Table 1).” If the chapter keeps Findings and Discussion separate, save broader theoretical interpretation for the appropriate section.
For guidance on organising all these displays into a coherent chapter, see our research findings chapter guide.
Step 9: Perform an independent accuracy check
Check every count and subtotal against the analysis output, verify percentages and rounding, confirm notes and units, open the export at the final page size, and check whether any chart was accidentally distorted while resizing. If a table or chart is adapted from another source, acknowledge that source as required by your citation style and permissions rules.
Worked example 1: Turn survey responses into a clear frequency table
Research question: Which feedback format do students prefer? In this hypothetical survey, 90 respondents each selected exactly one of three formats. Raw choices were counted as follows: written comments = 42, audio feedback = 30, and face-to-face feedback = 18.
Calculate and check the percentages
For each category, percentage = category count ÷ 90 × 100. Written comments: 42 ÷ 90 × 100 = 46.7%; audio: 30 ÷ 90 × 100 = 33.3%; face-to-face: 18 ÷ 90 × 100 = 20.0%. The counts add to 90 and the rounded percentages add to 100.0%.
| Feedback format | Count (n) | Percent of respondents |
|---|---|---|
| Written comments | 42 | 46.7% |
| Audio feedback | 30 | 33.3% |
| Face-to-face feedback | 18 | 20.0% |
| Total | 90 | 100.0% |
Note. Each respondent selected one format; percentages use n = 90 as the denominator. The values are illustrative.
Would a figure improve this result?
A simple bar chart could highlight the ordering of the three categories at a glance, but it is not necessary if the table already gives readers everything they need. If you do use a bar chart, put the formats on one axis, percentages from 0% on the other, and label the values. Avoid a pie chart simply for decorative variety.
Write the accompanying results sentence
Weak: “The table above shows written comments 42, audio 30, and face to face 18.”
Better: “Written comments were selected most often (42/90; 46.7%), followed by audio feedback (30/90; 33.3%) and face-to-face feedback (18/90; 20.0%) (Table 1).”
Why this works: the text highlights the ordering and the main evidence without implying that the preferences represent all university students. The dataset is a hypothetical convenience sample, so no population-level generalisation follows automatically.
Worked example 2: Select a figure for observations across time
Research question: How did recorded library workshop attendance change over five consecutive months? Imagine the monthly counts were January 20, February 23, March 21, April 29 and May 28. These are complete hypothetical administrative counts for the observed workshop sessions—not a random sample of future attendance.
| Month | Attendees |
|---|---|
| January | 20 |
| February | 23 |
| March | 21 |
| April | 29 |
| May | 28 |
Why a line chart may help
Because observations occur in a natural time order, a line chart can show fluctuations and the change from the beginning to the end of the period. Put the months chronologically on the horizontal axis and attendance count on the vertical axis. Use visible points and a clearly marked scale, ordinarily beginning at zero for straightforward count comparisons.
The first recorded value was 20 and the final value 28, a rise of 8 attendees or 40% relative to January. But the series did not rise every month: attendance fell from 23 to 21 in March and from 29 to 28 in May. Saying “attendance rose steadily every month” would misrepresent the data.
A defensible results sentence is: “Workshop attendance was higher in May (28) than in January (20), with month-to-month variation and the highest recorded count in April (29) (Figure 1).”
Do not claim that a new publicity campaign caused the increase unless the research design and evidence permit that conclusion. These five counts alone do not isolate a cause, estimate future attendance reliably or justify a statistical trend claim.
Worked example 3: Show a relationship without implying causation
Research question: Is the number of weekly study hours associated with a self-reported confidence score? Suppose eight students reported the following paired observations. Confidence ranges from 1 (very low) to 5 (very high).
| Student ID | Study hours/week | Confidence (1–5) |
|---|---|---|
| S01 | 2 | 2 |
| S02 | 3 | 3 |
| S03 | 4 | 2 |
| S04 | 5 | 4 |
| S05 | 6 | 3 |
| S06 | 7 | 4 |
| S07 | 8 | 5 |
| S08 | 9 | 4 |
Build the display carefully
If the research question concerns the relationship across participants, a scatterplot is one possible display. Put weekly study hours on the horizontal axis and confidence score on the vertical axis. Show each participant as a point; avoid connecting them with a line because they are different students, not observations across time. Label the confidence scale as ordinal self-report data. Depending on the aims and distribution, a table or distribution-focused presentation may be more appropriate than a scatterplot for only eight observations.
The values suggest that students reporting more study hours sometimes report greater confidence, but the pattern is imperfect: for example, 4 hours corresponds to confidence 2 and 6 hours to confidence 3. Do not invent a correlation coefficient, p-value, confidence interval or causal effect. Any formal analysis must match the measurement scale, sample and study design.
Weak versus defensible reporting
Weak: “The graph proves that studying more causes confidence to increase.”
Better: “In these eight hypothetical observations, confidence scores generally tended to be higher among students reporting more weekly study hours, although individual responses varied. The descriptive display does not establish causation.”
The important teaching point: a chart shows the observations you put into it; it does not automatically confer statistical significance, representativeness or causal identification.
How to present qualitative and mixed-methods evidence
Not every research result is a count or a correlation. In qualitative work, a concise table can help readers navigate themes, definitions and carefully selected evidence—but a tally of how often a word appears does not by itself establish the importance of a theme. If your project uses thematic analysis, our thematic analysis guide explains how codes and themes are developed.
A hypothetical qualitative display might pair the theme “Seeking help felt risky” with a one-sentence definition and an anonymised illustrative extract. The surrounding prose would explain the pattern, differences across accounts and its relevance to the research question. Protect participant confidentiality and follow the approved approach to quotations.
For mixed-methods research, a joint display can place quantitative patterns beside related qualitative explanations and include a third column stating what the combination suggests. Merely putting a survey table next to interview quotations is not meaningful integration. The display should show what each strand adds, where they agree or differ and what remains unresolved.
A practical table-and-figure checklist
-
Can I state the one question or message this display serves?
-
Is this information better shown in prose, a table or a figure?
-
Have I labelled the sample, denominator, units, categories and timeframe?
-
Are all counts, percentages, totals and calculations correct?
-
Is the display numbered and titled according to my institution’s requirements?
-
Can a reader interpret it without guessing what abbreviations mean?
-
Are the axes and scales accurate and not visually misleading?
-
Have I shown distribution or uncertainty where that is important?
-
Can the display be read in greyscale and by readers using assistive technology?
-
Have I referred to it in the body and highlighted its main message?
-
Have I avoided repeating the same data in prose, a table and a figure?
-
Have I removed software output that does not answer the research question?
-
Have I checked sources and permissions for borrowed or adapted figures?
-
Have I avoided overstating causation, generalisability or significance?
-
Have I verified that the exported figure remains clear at the final document size?
Common mistakes and how to fix them
Overcrowded tables
Split or simplify a display when readers cannot readily compare the important values; move secondary statistics to an appendix when justified.
Misleading axes
Check the zero baseline for bars, label any axis truncation and avoid choosing scales to dramatise modest differences.
No denominator
State how many participants or observations contributed to every proportion and why denominators differ.
Unexplained error bars
Define exactly what an interval represents—standard deviation, standard error or confidence interval—and how it was calculated.
Colour-only distinctions
Use labels or shapes so the figure remains interpretable without relying solely on colour.
Excessive precision
Report numerical detail appropriate to measurement and analysis, not every decimal supplied by software.
Missing or vague captions
Write titles and notes that identify the variables, groups, measurement units and sample context.
Treating charts as conclusions
Separate what the graphic shows from what the research design allows you to infer.
Repeating the same evidence
Choose each format for a different communicative purpose; cut redundant displays.
Pasting screenshots from software
Rebuild clean, editable or high-quality figures and tables consistent with the dissertation style.
Frequently asked questions
Should I use both a table and a graph for the same results?
Only when each adds genuine value. Usually one well-designed display plus clear prose is sufficient.
Do I have to follow APA 7?
Only if required by your institution, course, journal or discipline. Other style guides may govern caption placement, numbering and notes.
Can I use Excel or SPSS charts in my dissertation?
Yes, if the resulting figure is accurate, readable, properly labelled and formatted to the relevant requirements. Default software styling is not automatically suitable.
How many figures should a dissertation have?
There is no universal number. Include the displays necessary to answer the research questions clearly and avoid decoration.
Should I put raw data in the findings chapter?
Usually report relevant summaries and analyses; preserve raw data according to ethical, privacy and institutional rules. An appendix or repository may be appropriate only when permitted.
Can I edit a chart after analysis?
You can improve layout and labels without altering the observations or misleading readers. Recheck the plotted values and keep a traceable final version.
Final takeaway
A strong research table or figure is not the most colourful chart or the longest statistics table. It is the simplest faithful representation that lets a reader understand the evidence and evaluate the claim.
Begin with the question, choose the right format, verify the numbers, label every important feature and write a short explanation of the main pattern. Show uncertainty and variability where they matter. Above all, match the strength of your words to the strength of your data.
If you need help planning an appropriate presentation approach or understanding your institution’s reporting requirements, Acadelyra provides academic writing guidance and support. Use our request-help page to describe your assignment brief and the kind of assistance you need. Your research decisions, data and academic responsibilities remain your own.
References and further reading
American Psychological Association. Tables and figures (APA Style).
Purdue Online Writing Lab. APA tables and figures.
Midway, S. R. (2020). Principles of effective data visualization. Patterns.