Quick-start guide to experimental design

PSYC 11: Laboratory in Psychological Science

Jeremy R. Manning
Dartmouth College
Spring 2026

What is the purpose of running an experiment?

  • Understand or explore how something works
  • Distinguish between several potential alternatives
  • Get more information (data!)

Two big design philosophies

  • Classic (maximize control): simplify the phenomenon, carefully manipulate specific factors across conditions, measure behavioral differences
  • Naturalistic (maximize realism): create a rich, realistic scenario, measure as much as possible, mine the data for patterns
  • Most studies fall somewhere on this spectrum -- where does yours?

The key ingredients

  • Independent variable (IV): what you manipulate or compare across groups
  • Dependent variable (DV): what you measure
  • Controls: what you keep the same so you can isolate the effect of your IV
  • Participants: who are you studying, and how many?

Discussion: what's your IV/DV?

  • In your project groups, identify:
    • What is your IV? (What are you manipulating or comparing?)
    • What is your DV? (What are you measuring?)
    • What needs to be controlled?
  • If you are doing an observational study, what are your key variables of interest?
  • Be ready to share with the class

Common pitfalls

  • Confounds: something other than your IV that differs between conditions
  • Demand characteristics: participants guess what you expect and change their behavior
  • Too many variables at once: keep it simple -- one clear comparison is better than five murky ones
  • Forgetting a baseline: what does "normal" look like without your manipulation?

Implementation tools

  • Low-tech: notebooks, audiovisual recordings, Google Forms
  • Mid-tech: slideshows, Qualtrics surveys
  • High-tech: PsychoPy, jsPsych, Google Colaboratory
  • Use what you already know -- simplicity is your friend

Practical advice

  • Simplicity: the art of maximizing the amount of work not done
  • Pilot test early -- run your study on a friend before collecting real data
  • Work together and ask for help
  • You have 3--4 weeks -- scope accordingly!