Student writing

Your Results Contradicted Theory. Write the Discussion That Wins.

Unexpected results do not weaken a lab report. A lazy 'possible errors' paragraph does. Your job in the Discussion is not to apologize for the data — it is to interrogate it like a scientist.

You look at the output. The predicted absorbance peak is 520 nm. Your peak is at 498. The theoretical yield was 87%. You got 41%. You stare at the screen at 1am, delete the paragraph, start again, delete it. You type 'Possible sources of error include human error and equipment inaccuracy' because that is the safe, empty, professor-skimming sentence. Stop. That sentence is not a discussion. It is a surrender. The whole point of the Discussion section is the part where reality disagrees with the textbook. That is the part your marker has been waiting for. Unexpected data is not a mark of failure. It is the single most analytical thing you can write about all semester. Use it.

Here is the structural move nobody teaches clearly. A Discussion built on contradictory results runs in four beats, not one: state the deviation precisely, localize where it occurs, propose a mechanism or measurement cause, then design the one next experiment that would confirm or kill your explanation. Four beats. Not a paragraph. Not an apology. Your first beat is not 'the results were unexpected.' It is: 'The measured rate constant was 2.1 x 10^-3 s^-1, 38% above the Arrhenius-predicted value at 25°C.' Precise number. Named model. Direction of deviation. That sentence alone puts you in a different league than 90% of submissions.

Now the ❌→✅ pairs. ❌ 'The results did not match the expected outcomes due to experimental error.' ✅ 'The 38% overestimate clusters in the first three trials (t = 0-15 min) and drops below 4% after the cuvette was replaced. This pattern rules out random scatter and points to a scratched cuvette introducing a scattering artefact at the 520 nm detection wavelength.' ❌ 'It is possible that some errors occurred during the experiment.' ✅ 'The 2.3% pH deviation at equivalence points correlates with the burette not being flushed for the first three titrations; subsequent runs show <0.4% deviation, indicating a correctable systematic offset rather than stochastic noise.' See the difference? One names a mechanism and a pattern. The other names nothing and apologizes.

The third beat — mechanism and localization — is where non-native English speakers and AI drafts both fail in the same way: flat, hedged, every clause weighted equally. 'It could possibly be that the reaction may not have proceeded as expected, which might be due to various factors such as impurities or temperature fluctuations.' This is word-soup. It says nothing. A sharp Discussion says: 'The plateau above 40°C is inconsistent with a single-pathway mechanism but matches a competing parallel-reaction model in which the secondary product precipitates at higher temperatures, removing it from the equilibrium calculation.' One sentence. A mechanism. A falsifiable reason. If your draft reads like the first version, run it through easydue before you submit. Not to 'make it sound AI-free.' To strip the hedging fat off the analytical muscle so your actual reasoning hits the page clean.

The fourth beat is the one students skip entirely: the next experiment. Not 'further research is needed.' Not 'more trials should be conducted.' Specific. 'Repeating the titration with a dry burette and a calibrated pH probe at three equivalence points would isolate the systematic offset; if the deviation persists below 0.5%, the source is in the buffer preparation, not the apparatus.' You are not asking permission to do more work. You are showing the marker that you think like a researcher who is done writing and ready to pick up glassware.

Two things will tank this section even if your analysis is good. First, the tone slide into apology: 'We attempted to carry out the experiment carefully, but unfortunately the results were not in line with expectations.' Drop every 'unfortunately,' every 'despite our best efforts,' every 'it is worth noting that errors may have been present.' Your data is the data. Reporting it is not a confession. Second, the vague-bucket trap: 'human error,' 'equipment error,' 'environmental factors' as a three-item list is a paragraph that costs you marks because it proves you did not actually think about your own setup. If you cannot name the specific screw you left loose, you did not investigate your own experiment.

Last thing. The emotional core of this assignment is not English. It is fear. You were taught that science means the answer matches the book. When it does not, you feel like you broke something, and you write a Discussion that whispers 'sorry' instead of one that says 'here is what the data actually show, and here is what it means.' Your professor does not want a corrected result. They want to watch you think honestly with messy numbers. That is the skill. The phrasing is the packaging. easydue exists for the packaging — taking a draft where you know what you mean but the English flattens it into mush — but the thinking has to be yours first. You built the model. You ran the trial. You saw the peak at 498, not 520. Now write the paragraph that treats that gap as your finding, not your failure.