Results vs. Discussion: Separate Findings from Interpretation
By GPTHumanizer’s AI operations assistant · Writing guide
Published · Updated
Results report what your study found; the Discussion explains what those findings may mean, how they relate to the question, and where the interpretation is limited. When revising, identify the job of each sentence before polishing it. Polished wording can still turn a possible explanation into a fact. This guide uses one fictional study to separate a mixed paragraph into Results and Discussion, then shows how the same distinction works in a combined section.
Results vs. Discussion at a glance
Swipe across to read all columns.
| Question | Results | Discussion |
|---|---|---|
| Main job | State the findings that answer the research question. | Explain their meaning, limits and possible implications. |
| Evidence to check | Data, counts, estimates, table references and what was actually measured. | The reported findings, relevant checked literature, and support for each interpretation. |
| Typical editing risk | Adding an explanation that the observation does not establish. | Repeating the numbers without explaining them, or making claims beyond the study. |
| Useful revision question | Can I trace this statement to the reported data? | What supports this explanation, and what alternative remains? |
For reporting findings logically and distinguishing them from interpretation, see USC’s guidance on reporting results.
For interpreting findings, acknowledging limitations and avoiding a repeat of the results, see USC’s discussion-section guidance. The distinction is a reading aid, not a rule that every paper must use identical headings.
Start with one checked study record
The following exercise is entirely invented. It is not a published study, customer text, or a recorded Humanizer result. These facts are the editing record; do not add a source, p-value or outcome that is not in it.
Swipe across to read all columns.
| Item | Recorded fact |
|---|---|
| Question | What differences in copying errors were observed with two screen layouts in one session? |
| Participants | 24 adult volunteers from one campus; all completed both layouts. |
| Procedure | Everyone used layout A, then B, copying the same set of codes for three minutes per layout. The order was not counterbalanced. |
| Observation | 16 volunteers made fewer errors with B, five made the same number, and three made more. |
| Analysis boundary | No inferential statistical test or confidence interval was computed. |
| Not measured | Readability preferences, the separate contribution of practice, and later performance. |
The groups add to 24. They describe the direction, not the size, of each person’s change. No average error reduction is supplied; do not invent one.
Edit the mixed paragraph into Results and Discussion
The draft below combines an observation, a procedure, an untested explanation and unsupported claims. The edited example supplies the Results paragraph. A separate Discussion paragraph follows; the complete procedure belongs in Methods.
Before and after: an editing example
Original fictional exercise · manually edited
Both the study record and the wording were created for this exercise. No real research, customer text or model output is represented.
Original draft
Layout B is better because it makes information easier to read. In this one-session task, 16 of 24 volunteers made fewer copying errors with B than with A, five made the same number, and three made more. All volunteers used A first and then B, with the same codes for three minutes per layout. The difference was statistically significant. These findings prove that B will improve everyone’s accuracy. Practice may have contributed to the pattern.
Edited example
Results: In the one-session copying task, 16 of the 24 volunteers made fewer errors with layout B than with layout A. Five made the same number of errors with both layouts, and three made more errors with B. These counts describe the observed direction of change within this group.
This edited paragraph reports all three observed groups without claiming a cause or statistical significance. The separate Discussion below explains the interpretation and its limits; the complete procedure remains in Methods.
Discussion: The observed pattern favors layout B for more volunteers in this session, but it does not establish that the layout caused the difference. Everyone used A before B with the same codes, so practice is one possible explanation; its contribution was not measured separately. The single-campus volunteer sample and one-session design also limit broader conclusions. Testing a different order could help examine the order explanation, but that additional test was not performed here.
Swipe across to read all columns.
| Draft element | Editing decision | Reason |
|---|---|---|
| “B is better because…” | Remove the readability claim. | The exercise did not measure readability or isolate the layout’s causal effect. |
| 16 fewer, five equal, three more | Keep in Results, with 24 as the denominator. | These are the observed counts; preserve the exceptions as well as the majority. |
| A first, then B; same codes; three minutes | Keep the full procedure in Methods; refer to fixed order in Discussion. | The procedure explains the order limitation without becoming a new finding. |
| “Statistically significant” | Remove it. | No test was performed. A visible numerical pattern is not a significance result. |
| “Prove…everyone’s accuracy” | Replace with a bounded interpretation. | The exercise concerns these volunteers in one session, not all users or a proven cause. |
| Practice may have contributed | Keep as a possible explanation in Discussion. | The record permits the possibility, but does not measure its contribution. |
The revision retains the three volunteers whose errors increased and separates a suggested test from a completed one. If your real record contains a valid analysis, report it accurately; this example’s absence of a test is not a reason to delete analysis from another study.
Can results and discussion be combined?
Section arrangements vary across disciplines and assignments; UW–Madison’s scientific-report guidance describes the roles of methods, results and discussion rather than one format for every paper.
A combined Results and Discussion section can be appropriate. Check the journal or assignment requirements; Indiana University’s science-writing guidance covers both combined and separate arrangements.
For this exercise, a combined section could state the 16/five/three finding, then discuss fixed order and the limited sample. Keep a visible transition from observation to interpretation. Do not attach “therefore B caused better accuracy” to the count merely because both sentences share a heading.
Revise your own draft in five steps
1. Check the required structure. Note whether your assignment or journal asks for separate or combined sections. Keep the research question and the checked data beside the draft.
2. Label each sentence by its job: procedure, observed finding, interpretation, limitation, or proposed next step. If a sentence does two jobs, consider splitting it. Mark unsupported statements for verification instead of forcing them into a section.
3. Rebuild the Results paragraph from observations. Check every number, denominator, unit, direction, exception and table reference. Do not change “some participants” to “participants” or supply a missing average from a count.
4. Rebuild the Discussion around a supported interpretation. Identify the finding it explains, distinguish possible explanations from measured ones, and state limits. Verify any comparison with earlier research against the actual source; do not invent a citation to make the explanation persuasive.
5. Read the pair together. Every interpreted finding should have been reported. Every proposed next study should remain a proposal. Remove repetition only after checking that the reader can still follow the connection.
For the broader paragraph check, Check an academic rewrite claim by claim.
If your task is instead compressing the whole study to a submission limit, Shorten an existing research abstract has a separate worked example.
Use Academic mode for a wording pass
After deciding what each paragraph should say, open Humanizer with your own existing text. Sign in, select Academic, check your available input-word allowance, and submit when ready. New accounts have a one-time 300-word allowance shared with Essay Writer. Opening the editor does not submit a rewrite.
Compare the suggestion with both the original and your study record. The tool does not verify sources, decide statistical significance or establish a causal explanation. Keep only changes you can support, and follow your institution’s rules for using and disclosing editing assistance.
Open Humanizer for your existing paragraph
For a longer paper, Review a longer draft section by section. Keep your original and check links between findings and interpretations after editing individual sections.
Your final review
- Every reported number and denominator matches the checked record.
- Observations and interpretations remain distinguishable, even under a shared heading.
- The discussion explains reported findings rather than introducing unreported results.
- Causal and statistical claims have actual support; uncertainty and exceptions remain visible.
- Procedures, completed observations and proposed future work are not confused.
- Citations and section arrangements have been checked against the real sources and requirements.