PSY460: Advanced Quantitative Methods

Week #1: Collaborative Research

Today, we’ll introduce ourselves, discuss the scope of the course, talk about using existing data, and work in teams to begin refining hypotheses.

Introductions

  • What is your name, and what pronouns do you use?
  • What has been your favorite class in Psychology (or SPM/BFB)?
  • Why are you taking this Collab?
  • What is a characteristic of yours that might surprise others? (For example, do you seem extroverted while being an introvert at heart?)
  • If we were stranded on a desert island and decided to vote somebody off the island, what is the most compelling reason you would have to convince us not to vote you off?

Syllabus

  • Today’s quiz:
    • How will assignments be graded in this course?
    • Will your final paper be written individually or in a team?
    • Is anything scheduled outside of class time?
    • Bonus: Can you use Chat GPT?
  • Any questions?

Opportunity for Another Token

Talk by Audun Dahl

Tuesday, Feb. 4 from 11:30-12:30

“Endorsing Force: Our Changing Moral Views On Hitting, Biting, and Kicking Others”

If you attend and email me with something you learned about research design or statistics, you will earn an extra token.

What are Advanced Quantitative Methods?

  • I will define these as any statistical techniques that move beyond the scope of PSY 300, often by adding additional variables to the mix.
  • Examples include mediation analysis, factor analysis, cluster analysis, mixed models, structural equation modeling, and Bayesian statistics.
  • Rather than learning a whole new suite of statistical analyses, we will learn whatever tools are needed for the specific research questions that each team is asking.

Your Objectives for the Semester

  • You will develop a novel hypothesis to test with existing data.
  • You will learn the statistical techniques necessary to test your hypothesis with R.
  • You will convey your findings in a team presentation and an individual paper.

Using Existing Data

  • There are many benefits of using publicly available datasets.
    • The sample sizes are generally enormous, allowing for high-powered analyses, and they have typically not been sufficiently mined to answer many interesting questions.
    • Increasingly, published papers are taking advantage of publicly available data (see the repository that I’ve started on our Canvas site–and feel free to add to it!).
  • Of course, there are also downsides. Some questions (e.g., about how seating arrangements impact learning) will likely not be answerable, though some creative thinking could indicate other avenues to get at the underlying question.

How to Think About Existing Data

  • Coming up with a research question cannot happen entirely without knowing what data are available to test it, but the data cannot generate questions by themselves.
    • You will likely engage in an interative process involving generating possible hypotheses, examining whether data exist to test these hypotheses, and refining the hypotheses on the basis of what data you find and what has already been published in the research literature.

Group Assignments

Topic Group Members
Behavioral economics Josh Kisilevich, Katie Pippenger, & Yuzhe Tian
Health and cognition Ayako Belka, Annaliese Heidelberger, & Yosuke Iguchi
Effects of social media Brianna DeJesus, Hannah Matusiak, & Katherine McGee
Learning and success Gonzalo Gentou, Kyle Samson, & Jake Weiner

Scheduling Weekly Group Meetings

Please take the next five minutes to decide on a team name, and determine which of the following options would reliably work for each of you on Tuesdays:

  • 8:30-9:00
  • 9:00-9:30
  • 9:30-10:00
  • 10:00-10:30
  • 10:30-11:00
  • 11:00-11:30
  • 12:45-1:15
  • 1:15-1:45
  • 1:45-2:15

If none of these options are possible for your group, we can brainstorm about other possibilities.

Initial Hypothesis Generation

  • You all came up with interesting research questions to test! However, nearly everything that was submitted predicted a straightforward relationship between two variables (e.g., does using social media reduce peoples’ satisfaction with their bodies?). While these are worthwhile questions, they could be tested with simple statistical tools (e.g., t-tests or correlations).
  • In order to use more advanced statistical techniques, you will be required to test more complex hypotheses in this course.

Finessing Hypotheses

  • What variables could you add to make your models more sophisticated?
  • How could you operationalize your variables, and what kinds of causal or correlational relationships are you predicting?
    • Is the relationship you are envisioning the only likely relationship between the variables? (For example, could the opposite causal relationship be possible?)
    • Note that not all hypotheses will be about causation. You could instead think about construct validity, non-causal relationships, etc.

Discuss your hypotheses with your teammates.