Teaching research to college students#

Teaching research is about helping students move from curiosity to evidence. The course works best when students experience discovery early and often.

Create excitement and momentum#

  • Start with a short mystery or puzzle tied to real data.
  • Use a one class micro study so students see a full cycle.
  • Share quick wins: a chart, a prototype, or a surprising result.
  • Spotlight student work in a weekly “show and tell”.

Make discovery feel possible#

  • Provide a small dataset that is already clean.
  • Give a starter notebook or code template.
  • Use pair work for the first 2 weeks to reduce risk.
  • Frame research as iterative, not perfect on the first try.

Teach research question finding#

  • Have students write 10 questions in 10 minutes.
  • Sort questions into descriptive, comparative, and causal.
  • Use the template: “How does X affect Y under Z?”
  • Ask students to narrow to one measurable outcome.

Teach answer chasing#

  • Identify one baseline method and one improvement.
  • Require a small pilot with results in week 3 or 4.
  • Have students track decisions in a research log.
  • Teach “failure analysis”: what did not work and why.

Build a research interest and area#

  • Use theme weeks: privacy, AI, health, networks, or HCI.
  • Invite a 10 minute guest talk from a local researcher.
  • Give students a menu of data sources aligned to the theme.
  • Ask for a 2 sentence interest statement every two weeks.

Assessment ideas#

  • Evaluate clarity of question and scope, not just results.
  • Grade on iteration: question, pilot, revision, and final.
  • Use a peer review round for research proposals.

Class structures that work well#

  • 15 minutes: mini lecture or demo
  • 20 minutes: lab style activity
  • 15 minutes: group check in and planning
  • 10 minutes: share outs or reflections

A short activity to open the course#

  1. Show a small dataset or system output.
  2. Ask students to write 5 questions about it.
  3. Vote on the most interesting question.
  4. Spend 20 minutes analyzing the data together.
  5. End with a reflection: what else would we need to know?
Backward Research Workflow Working with Ideas Forward