Exploring and understanding data

PSYC 11: Laboratory in Psychological Science

Jeremy R. Manning
Dartmouth College
Spring 2026

Truth and data

  • The "universe" produces data
  • Math doesn't lie— but analyses involve choices
  • Different analyses can lead to different conclusions
  • The same dataset can tell very different stories

What patterns should you look for?

  • Shape: How many observations? How many features? Any missing data?
  • Distributions: Are values clustered? Spread out? Skewed?
  • Relationships: Do any features move together? In opposite directions?
  • Outliers: Are there values that seem "wrong" or surprising?

The power of visualization

  • Tables of numbers hide patterns; plots reveal them
  • Always look at your data before running statistics
  • Anscombe's quartet (below): four datasets with identical means, variances, correlations, and regression lines — but completely different stories

Discussion: what does this graph tell you?

A psychology professor surveyed 60 of her PSYC 6 students about their study habits and recorded their midterm exam scores. She fit a regression line to the data:

  • What story does this plot tell?
  • What is it not telling you? What would you want to know before believing the "more studying → higher scores" story?
  • What follow-up plot or analysis would you want to see next?

Discussion: how about this one?

The number of xkcd comics published about literature is strongly positively correlated with the robbery rate in Vermont (2007–2022).

  • What story does this plot tell?
  • What is it not telling you? What would you want to know before believing the "more xkcd literature comics → more robberies" in Vermont story?
  • What follow-up plot or analysis would you want to see next?

Discussion: how about this one?

The number of xkcd comics published about literature is strongly positively correlated with the robbery rate in Vermont (2007–2022). Source

"As the number of xkcd comics about literature increased, so did the number of book enthusiasts flocking to Vermont. These visitors, caught up in the intrigue of literary references, found themselves drawn to the bustling criminal underworld. In their fervor for all things book-related, they inadvertently sparked a wave of daring heists, leaving the authorities scratching their heads as to why Shakespeare and Hemingway seemed to be the unusual inspiration behind the sudden rise in robberies. The increase in xkcd comics published about literature may be positively influencing robberies in Vermont through the unintended consequence of attracting more tourists to the state. Tourists who are coming to Vermont after being drawn in by the literary theme in the comics. More tourists can sometimes lead to more opportunities for theft, which could explain the correlation."

Analytic flexibility

  • There are typically many ways to analyze data
  • Different choices (which subset, which test, which visualization) can lead to different conclusions
  • This is why transparency about your analysis choices matters

What's in your toolkit?

  • Observation, intuition, and logic
  • Simple summaries (mean, standard deviation, sorting)
  • Traditional statistical tests (t-tests, correlations, ANOVAs)
  • Fancier methods and simulations

Getting help

  • Teaching staff (instructor + TAs)
  • Other students
  • Slack (#stats-stuff, #data-sleuthing-lab)
  • Google, Stack Exchange, Wikipedia, ChatGPT/Claude/chat.dartmouth

Questions? Want to chat more?

📧 Email me
💬 Join our Slack
💁 Come to office hours
  • Friday: Part 3 of the lab (more analysis, discussion)