Synthesizing across studies

PSYC 11: Laboratory in Psychological Science

Jeremy R. Manning
Dartmouth College
Spring 2026

A synthesis is not (just) a summary

  • Summary: "Smith et al. (2018) found X. Jones et al. (2020) found Y. Lee et al. (2022) found Z."
  • Synthesis: "Together, these studies suggest X is true under condition A but not condition B — consistent with the broader idea that..."

The reader can already see that you read the papers. What they want from you is the connective tissue — how the findings fit together (or don't), and what they collectively imply.

When studies seem to disagree, first ask: do they?

  • Different methods. A self-report study and an fMRI study might both be "right" but measure different things
  • Different operationalizations. "Memory" could mean recall, recognition, or working memory — pick a different one and you'll get a different result
  • Different populations. College undergrads vs. children vs. older adults vs. clinical samples
  • Different effect sizes, not different signs. Both papers show r>0r > 0, but one rounds to "no relationship" and the other to "strong relationship"
  • Sampling noise. With small NN, two studies of the same effect can look very different just by chance

A genuine disagreement is when both studies measured the same thing the same way in comparable samples and got incompatible results. That's interesting — and worth saying so.

Example: do these papers agree or disagree?

  • Paper A (Smith, 2019; N = 40 undergrads): Caffeine improves recall on a word list (d=0.6d = 0.6, p=0.02p = 0.02).
  • Paper B (Jones, 2021; N = 200 adults aged 18–75): No effect of caffeine on a recognition task (d=0.05d = 0.05, p=0.51p = 0.51).
  • Paper C (Lee, 2023; N = 80 sleep-deprived medical residents): Caffeine strongly improves working memory (d=1.1d = 1.1, p<0.001p < 0.001).
  • Do these three papers agree, disagree, or address different questions?
  • Which differences (sample, task, measure) might explain the pattern?
  • If you had to write one sentence synthesizing all three, what would it say?

Building the narrative thread

  • Frame the question first, before introducing any paper. Readers need to know what they're trying to learn before they hear about the evidence
  • Group papers by what they tell you, not by author or date. ("Three studies suggest X..." is better than "Smith found X. Then Jones found X. Then Lee found X.")
  • Name the tension explicitly. "These studies disagree on whether X causes Y. The disagreement appears to come from..." is much stronger than ignoring it
  • Take a position. What do you think the evidence actually shows? Hedge where appropriate, but don't refuse to commit

Being honest about uncertainty

  • Be specific about what is uncertain: the direction of an effect? Its size? Whether it generalizes? The underlying mechanism?
  • Distinguish what the evidence supports from your speculation. "These results are consistent with..." vs. "We speculate that..."
  • Identify testable predictions. "If our speculation is correct, then we would expect to see X in a future study."

"More research is needed" is true of literally every result ever published. If you write it, follow it with what kind of research, and why that research would resolve the specific uncertainty you've named.

Apply it to this week's lab!

  • Continue working on your literature review
  • Focus on synthesizing across your sources, not just summarizing each one
  • Draft the narrative thread that connects your papers
  • For each pair of papers that seem to disagree: ask whether the disagreement is apparent or real

Questions? Want to chat more?

📧 Email me
💬 Join our Slack
💁 Come to office hours
  • Today: Finish finding papers, summarize, and synthesize
  • Friday: Discuss and brainstorm, submit your Data Sleuthing Lab writeup by 11:59pm!