Academic writing

Your Results Contradict Themselves? Your Tone Is the Real Problem

Mixed findings don't need your apology. They need to be stated as a finding, not defended like a verdict.

You stare at Table 3 for twenty minutes. The p-value sits at .054. Your hypothesis is half supported, half dead. The first thought that hits isn't "this is interesting" — it's "this is a disaster." So you open Word and type: "However, the results of this study may indicate that the relationship between X and Y is not straightforward, and further investigation is warranted." You read it back. You're not discussing results. You're reading your own plea bargain.

Mixed findings are not a death sentence. Here's the uncomfortable truth: in actual research, clean results are the minority. Your supervisor doesn't look at contradictory data and think "this student is a failure." They think "this student is a failure" because your writing makes it sound like you already think so. The problem was never "my data isn't good enough." The problem is "my tone is apologizing on behalf of my data." You don't need cleaner results. You need a less nervous voice.

There are three ways international students kill their discussion section when results are mixed. First: hedging overload. "It may be that... This could suggest... It is possible that..." Four hedges in one paragraph, and the reader doesn't think "this student is rigorous." They think "this student doesn't believe their own conclusion." Second: burying. You front-load the supportive findings and tuck the contradictory ones in the last two paragraphs. The reader reaches page 12 and discovers "oh, there was also this result." That's not honesty. That's a slow retreat. Third: pretending. You write "the results supported the hypothesis" and the contradictory data simply... evaporates. Your supervisor sees through it in one pass.

The fix is one sentence long: treat the mixed result as a finding. Not "our results may suggest that X operates under certain conditions" — but "X operated under Condition A (p = .02) and did not operate under Condition B (p = .14), and this divergence is the central finding of this study." See the difference? ❌ The results may suggest that the intervention had a limited effect, and further research is needed. ✅ The intervention reduced scores in Group A (p = .02) but not in Group B (p = .14), a divergence that points to prior knowledge as a boundary condition. The second sentence doesn't apologize. It doesn't retreat. It converts inconsistency into information.

One more. Limitations are where students lose the most composure: ❌ It should be noted that the sample size was relatively small, and the findings may not be generalizable. You've read this sentence in eight hundred papers. It carries zero information. It's just "I know this isn't perfect." Replace it: ✅ The sample (n = 47) limits power for detecting small effects; however, the effect in Group A (d = 0.61) exceeds the threshold for which this sample is adequately powered. You acknowledged the limitation. You also told the reader your key finding survives it. That's not bragging. That's precision.

And the transition sentence — where most people fumble the hardest. Moving from supportive results to mixed ones, 90% of drafts say: "However, the results were not entirely consistent with our expectations." The problem with "not entirely consistent" is that it's fudging for the data, implying you expected consistency and it let you down. Cut it: ❌ However, the results were not entirely consistent with our expectations. ✅ The pattern diverges from Hypothesis 2 in a way that is theoretically informative. You don't owe the reader your expectations. You owe them the meaning of the divergence.

Practical move: after your first draft of the discussion, count every instance of "may," "might," "could," "possible," and "suggest" across the whole section. If you're over eight, you're hedging out of fear, not rigor. Cut half. Keep the rest where genuine caution applies. I dump my first draft through easydue to flip the register from "nervous" to "declarative" — the time I save goes into thinking about mechanisms and boundary conditions, which is where the actual marks live. Your data can contradict itself. Your writing can't.