A strong experiment conclusion does four things in roughly this order: restates the purpose of the experiment, identifies the main findings, notes limitations that affect how the results should be interpreted, and explains what the experiment contributes to the broader understanding of the problem. Most conclusions run one to three paragraphs, and every sentence should connect back to your actual data.
Start With the Purpose and Your Findings
Open your conclusion by briefly restating the question your experiment was designed to answer. You don’t need to repeat your entire introduction. One sentence that reminds the reader of the goal is enough. Follow it immediately with your main finding, stated plainly.
If you had a specific hypothesis, say whether your results supported it or not. Be direct: “The data supported the hypothesis that…” or “The results did not support the prediction that…” If you didn’t start with a formal hypothesis, state whether your results were consistent with what previous studies have suggested. Either way, ground every claim in the data you collected. A conclusion that says “the hypothesis was confirmed” without pointing to the specific measurements or trends that led you there will read as unsupported.
For example, if you hypothesized that plants grown under blue light would be taller than those under red light, you might write: “Plants in the blue-light group averaged 12.4 cm after 14 days compared to 8.1 cm in the red-light group, supporting the hypothesis that blue wavelengths promote greater stem elongation.”
Summarize the Evidence, Don’t Restate the Data
Your results section already contains tables and raw numbers. The conclusion should interpret those numbers, not repeat them. Focus on the patterns, trends, or differences that matter most. If you ran statistical tests, reference the key outcome briefly. For instance, noting that a difference was statistically significant (or was not) gives the reader a clear signal about how confident you can be in the finding.
When reporting these results, pair the statistical outcome with a measure of the actual size of the effect. Saying “the difference was significant” tells the reader the result probably isn’t due to chance, but it doesn’t tell them whether the difference is large enough to matter in practice. A sentence like “the treatment group scored 15% higher on average, and this difference was statistically significant” communicates both pieces.
Address Limitations Honestly
Every experiment has limitations, and naming them isn’t a weakness. It shows you understand what your data can and cannot prove. Focus on limitations that are genuinely relevant to interpreting your results, not a generic list of everything that could theoretically go wrong.
Common limitations worth noting include small sample sizes, uncontrolled variables, measurement precision, or a short time frame. If your experimental design couldn’t separate the effect of time from the effect of your treatment (for example, if there was no control group measured at the same intervals), say so, and present your conclusions as tentative rather than definitive. The goal of any experiment is to provide evidence that something might be true or untrue, not to prove it beyond doubt.
One or two sentences identifying the most important limitations is usually sufficient. You don’t need a full paragraph cataloging every imperfection.
Explain Why the Results Matter
This is the part many students skip, and it’s often what separates a good conclusion from a forgettable one. After stating what you found, explain why someone should care. Connect your findings to the broader problem your experiment relates to.
Ask yourself: what do these results mean beyond this specific experiment? If you tested how caffeine affects reaction time, for instance, what might that imply for people who drink coffee before driving or studying? If you measured bacterial growth under different conditions, how might that relate to food safety or medical treatment? You don’t need to overreach. Even a sentence or two that places your small experiment in a larger context shows the reader you understand the significance of what you did.
If your experiment is part of a class assignment, this section can be brief. If you’re writing a formal research report, you might also suggest a specific direction for follow-up work, such as testing a wider range of concentrations or using a larger sample.
What Your Results Don’t Support the Hypothesis
Negative or inconclusive results deserve the same careful treatment as positive ones. A non-significant result is ambiguous: it could mean there really is no effect, or it could mean your experiment didn’t have enough statistical power to detect one. Don’t write “the hypothesis was wrong” when the more accurate statement is “the data did not provide evidence for the hypothesis.”
When this happens, consider whether your sample size, measurement tools, or experimental conditions could explain the lack of a clear result. Proposing a plausible reason for null findings shows critical thinking. Maybe the treatment duration was too short, or the difference between conditions was smaller than your instruments could reliably measure. State these possibilities without over-speculating.
Mistakes That Weaken a Conclusion
Several common errors can undermine an otherwise solid conclusion:
- Introducing new data. Your conclusion interprets findings already presented in the results section. If a number or observation appears for the first time in the conclusion, it belongs somewhere earlier in the report.
- Overgeneralizing. If you tested 20 college students, don’t claim your findings apply to all humans. Keep your claims proportional to your evidence.
- Using causal language for correlational data. If your experiment measured a correlation (two things that change together), don’t write that one caused the other. Use language like “was associated with” instead of “caused” or “led to,” unless your design genuinely isolated the causal variable.
- Claiming significance from separate comparisons. A common error is noting that one group showed a significant effect while another didn’t, and concluding the two groups are different from each other. That inference requires a direct comparison between the groups, not two separate tests.
- Ignoring contradictory data. If one trial or data point didn’t fit the pattern, acknowledge it rather than pretending it doesn’t exist. Unexplained results are part of science.
Conclusion vs. Discussion: Know the Difference
In many lab reports, especially at the undergraduate level, the conclusion and discussion are combined into a single section. But in longer or more formal papers, they serve different purposes. The discussion section is where you interpret your results in detail, compare them to previous studies, and work through alternative explanations. The conclusion is a shorter, final summary that distills the main takeaway and its broader significance.
Think of it this way: the discussion is where you do the analytical work, and the conclusion is where you step back and tell the reader the single most important thing to walk away with. If your assignment asks for a combined “Discussion and Conclusion,” start with interpretation and analysis, then end with a paragraph that functions as a standalone conclusion. That final paragraph often includes a recommendation for future research or a statement about practical implications.
A Quick Checklist Before You Submit
- Purpose restated? The reader knows what question your experiment was answering.
- Main findings identified? You’ve stated the answer to your research question clearly, with supporting data.
- Hypothesis addressed? You’ve said whether the data supported, partially supported, or did not support your prediction.
- Limitations noted? You’ve identified the most relevant factors that could affect interpretation.
- Broader significance included? You’ve explained what the experiment contributes to understanding the larger problem.
- No new data introduced? Everything in the conclusion already appeared in your results.
- Claims match the evidence? Your language reflects the strength of your data, with no unsupported leaps.