Student writing

Your Lab Results Contradict the Hypothesis. Stop Apologizing. Start Analysing.

Unexpected results are not a failure to confess in your discussion section. They are the entire reason the discussion section exists. Stop writing apologies. Start writing arguments about why the system behaved the way it did.

You ran the spectrophotometry assay. Absorbance at 450 nm. You expected a clean sigmoidal dose-response curve. You got scattered noise, a flat line, or a curve that shifts three log units left. Or your titration endpoint overshot by four drops. Or the gel band sat at 320 bp instead of 250 bp. Now the cursor blinks in front of the Discussion heading and your fingers are already typing: 'The results obtained were not in line with the expected outcomes, which may suggest that there were errors in the experimental procedure.' Stop. That sentence is an apology. And an apology in a discussion section reads, to anyone grading it, as: 'I do not understand my own data.' Graders smell that in four seconds. It costs marks you were otherwise going to get.

Here is the reframe nobody handed you: the discussion section exists precisely for the moment your data misbehaves. If every result confirms your hypothesis, you are writing a confirmation note, not a discussion. Undergraduate science is not about reproducing the answer key. It is about making sense of a wrong answer. A gap between expected and observed is not a wound you need to cover up with 'experimental error.' It is a finding. Screenshot this: your unexpected result is not the problem. Your inability to discuss it is.

Three specific rewrites that change an apology into an argument. First: ❌ 'The results were not what we expected, which is probably due to experimental error.' → ✅ 'The observed IC50 was 12.3 ± 0.8 uM, approximately four-fold above the value reported in Smith et al. (2021). This deviation is consistent with the known thermal inactivation of CYP3A4 in batch-2 substrate preparations at 37 °C (Park & Huang, 2023), indicating a temperature-dependent assay artifact rather than a novel mechanistic effect.' Second: ❌ 'I think I made a mistake measuring the volumes.' → ✅ 'A 5 uL pipetting error at the third dilution would shift the log-dose curve by roughly 0.3 log units, within the 0.4 log-unit deviation observed. This bounds the magnitude of procedural error but does not account for the outlier at the lowest dose, where pipetting error is negligible.' One is a confession. The other is an argument.

Build the paragraph in three moves. Move one: state the deviation with numbers. Not 'the results were different' but 'measured value was X versus expected Y, a Z percent deviation.' Move two: generate and test at least two candidate explanations. Reagent degradation, equipment drift, biological variability, a published discrepancy. You do not need to solve the puzzle. You need to show the reasoning chain. Move three: situate the anomaly in the literature. 'A similar four-fold shift was reported by Kowalski et al. (2019) in the same enzyme-substrate pair' transforms your one-off flub into a replicable phenomenon. If nobody has reported anything similar, write that honestly: 'To the authors' knowledge, this magnitude of deviation has not been reported previously in the CYP3A4 assay system.' That sentence, used with clean data, is gold. It is your actual contribution.

The hedging trap hits hardest here, because your instinct when the data looks wrong is to multiply 'possibly, perhaps, may have, might suggest' by ten. ❌ 'It could possibly be suggested that perhaps the results may be indicating that the experimental conditions might have caused some sort of error.' That is not hedging. That is a nervous system response. ✅ 'Two explanations fit the observed shift. The more parsimonious is reagent degradation over the four-hour run; the alternative, competitive substrate contamination from the shared stock, is less consistent with the control results but cannot be excluded without a blank-substrate control.' Specificity of reasoning replaces vague modality. The reader should feel you narrowed the hypothesis space, not that you are hiding behind 'maybe.'

For non-native writers specifically: the apologetic register you learned writing emails to professors bleeds into the discussion. 'I would like to humbly acknowledge that the results may not be as expected, and I am sorry for any inconvenience.' This is not academic English. This is a student asking permission to exist. Academic discussion does not apologize. It states, bounds, and reasons. ✅ 'The 14 percent deviation from expected values exceeds the stated instrument accuracy of 2 percent and is therefore unlikely to be instrumental error alone. Further investigation using a second spectrophotometer would be required to rule out calibration drift.' No 'humbly.' No 'kindly.' You are not asking the reader's forgiveness for your data. You are interpreting it.

When your draft is 80 percent apology and 20 percent analysis, the fix is not vocabulary or grammar. It is argument architecture. You need to rewrite which sentences claim and which sentences hedge, not just swap 'errors' for 'discrepancies.' That is a structural edit, and it is the one most students skip because they are exhausted at 2 a.m. with a submission due at 9. easydue can help here — not to make your English sound prettier, but to flag where 'perhaps' is doing the work 'the control data rules out' should be doing. You paste the draft. It shows you where you are backpedalling. The restructure is yours. The signal that the restructure is needed — that is where the tool earns its keep.