- Blog
- How to Study Figures and Tables in Research Papers
How to Study Figures and Tables in Research Papers
To study a research figure or table, connect the visible result to its caption, measurement, sample, and surrounding argument. Write what the display shows before deciding what it means. Then make a question you can answer without the paper open.
This guide is for university students preparing a seminar explanation or exam answer. The output is a checked evidence card for one display, rather than a summary of the entire paper.
Follow one display through the paper
The University of Minnesota's article-reading tutorial identifies figures, tables, and equations as places where papers present evidence. Use the section headings to move between the display, methods, results, and discussion.
Start with this route:
- Record the paper version and figure or table number, including the panel letter.
- Read its title, caption, legend, and footnotes.
- Identify the groups, variables, units, and comparison.
- Locate the method that explains how the measurement was made.
- Find the results sentence referring to this exact display.
- Separate that observation from the discussion's interpretation.
George Washington University's microbiology writing guide distinguishes reported results from their interpretation and describes figures and tables as presentations of major data. That distinction is useful when reading, too; the guide's discipline-specific writing requirements are not rules for every field.
Make a display identity card
Before interpreting a striking bar or percentage, fill in the identity fields. A panel from another condition or a table using a different population can make a plausible note wrong.
| Field | What to record | If missing |
|---|---|---|
| Anchor | Paper version, printed page, PDF viewer page, display and panel | Find a stable heading or caption locator |
| Population | Who or what each group contains | Mark population unresolved |
| Measurement | Outcome, unit, time window, and how measured | Consult methods |
| Comparison | Groups or time points; paired or separate samples if specified | Do not assume the observations are linked |
| Encoding | Axes, scale, colors, symbols, rows and columns | Read the legend and notes |
| Uncertainty | Exact definition of bars, intervals, missing values, and sample size | Record the ambiguity rather than guessing |
For a multi-panel figure, use one card per panel and a final sentence describing their connection. Do not let the shared figure number erase different outcomes, units, or sample sizes.
Compare values on their own scales
A panel letter identifies a specific display. Before comparing bar height or line slope across panels, record the outcome and read the ticks on each axis. The University of Illinois figures guide explains how units and labeled ticks make plotted values readable. For your evidence card, check the axis range, whether the tick intervals are linear or logarithmic, and whether the scale shows raw measurements, percentages or normalized values. If a table reports the same outcome, use its labeled values to check the pattern you see.
Teaching example with two panel scales
Imagine a fictional reading experiment whose Figure 3 shows mean reading speed for the same two groups in two tasks. The labeled points are:
| Panel | Y-axis shown | Group A | Group B | Difference, A minus B |
|---|---|---|---|---|
| 3A, short excerpts | 180–210 words per minute | 200 | 190 | 10 words per minute |
| 3B, long excerpts | 0–250 words per minute | 180 | 160 | 20 words per minute |
On equally tall linear axes, the gap takes up more vertical space in 3A. The numerical gap is larger in 3B. Put the unit and the panel name in your note so the observation can stand on its own. These are descriptive differences. To evaluate whether the group difference changes across tasks beyond sampling variation, find the analysis that tests that comparison.
Try it from memory. Explain which panel has the larger numerical difference and why the figure can give a different visual impression. Then reconstruct both differences from the table. In your next paper, check for unequal scales before comparing two panels.
Separate observation, interpretation, and unanswered questions
Write three short lines:
- Observed: the visible pattern, with its group and measurement.
- Interpreted: what the authors say it means, with a results or discussion location.
- Unresolved: the missing information needed to assess a stronger claim.
Do not decide statistical significance from the visual overlap of unspecified error bars. First establish what the bars represent and consult the reported analysis. A graph alone does not establish the study design or justify a causal explanation. If the analysis is outside your course knowledge, record a question for your tutor instead of adding a confident verdict.
For more focused checks, use our guides to confidence intervals and correlation versus causation.
A fictional sample-mismatch example
This is an authored teaching example, not a real study, empirical finding, or recorded Notoo output.
Imagine Table 2 reports self-reported weekly study hours:
| Survey wave | Respondents | Mean hours per week |
|---|---|---|
| Beginning of term | 40 | 8 |
| End of term | 25 | 6 |
The fictional table note says responses were anonymous and could not be linked across waves. The paper provides no paired analysis.
Overstated note: “Each student reduced studying by two hours.”
Checked note: “The reported mean was 8 hours among the 40 beginning-of-term respondents and 6 hours among the 25 end-of-term respondents. The unlinked responses do not show each student's change.”
The numerical difference between the two means is two hours. The unsupported step is turning that difference into an individual trajectory. The useful follow-up question is whether the respondent groups differed, not which explanation sounds most persuasive.
Copy this evidence-to-answer template
Paper / version:
Figure, table, panel:
Printed page / PDF viewer page:
Question the display addresses:
Population and sample size for each comparison:
Measurement, unit, time window:
Axis range and scale, legend, footnote, and uncertainty definitions:
Observed pattern:
Authors' interpretation and location:
What this display does not establish:
Unresolved question and next source to check:
Closed-book explanation prompt:
Source-checked correction after recall:
Use exact values only when you can read them reliably. If a plotted point has no exact label, describe an approximate pattern as approximate; do not manufacture precision from a low-resolution image.
Use Notoo for the readable surrounding material
Notoo's PDF-to-notes workflow can draft structured notes from readable research PDFs. Its product guidance asks readers to verify important details, including tables, diagrams, equations, citations, and small text. Keep the original display beside the draft and add source locations yourself.
Check whether a generated sentence belongs to the panel you selected. A completed upload does not prove every legend, footnote, or visual relationship was captured. The card above is a manual checking template; it is not a claimed Notoo export or automatic statistical audit.
Practise the explanation without the display
Close the paper and explain the question, population, measurement, pattern, and limit. Reopen it and check the exact point you missed. For international students, keep the original axis terms and units beside a plain-English explanation; a translation should not silently change the variable.
Your stopping point is a traceable answer with visible limits. Try Notoo on one permitted PDF, select one important display, and complete the evidence card before studying from its summary.
