Psy 611: Data Analysis I

You are a scientist

Descartes (1637) – doubt everything and then take as axiomatic only that which proved to be indubitable (“I think therefore I am”) and also division between the mind and body. - Descartes was also a physicist and mathematician – developed analytic geometry (Carteisan coordinates named after deCARTES) - hypothesized that that bodies were governed in mechanistic ways – the brain is a complicated system of tubes and valves for shunting animal spirits into specific nerves and therby enacting specific actions (external pressure on sensing device = reflexes) and this work inspired what became modern day neuroscience.

Helmholz (1849) – developed tool to measure reaction time to fractions of seconds (galanometer, “stopwatch”) - also invented ophthalmoscope, which examines the inside of the human eye - also created the Helmholtz resonator, which identifies the various fequencies of complex sounds containing multiple tones - personally replicated experiments of other scientists to ensure accuracy

Goals of this sequence

• Develop the basic quantitative skills necessary to be a research scientist
  • Not all the skills you will need (not the only courses you need)
  • Foundations of statistics, methods, and data science

• Contextualize those skills
  • Building a toolbox, not a cookbook
  • How can I use this? Under what circumstances? When should I not use this?

Goals of this sequence

• PSY 611: Tools of statistics
• PSY 612: Building models
• PSY 613: Taking it to the next level

Goals of this course

• Conceptualize statistics as a method for specifying and testing a model of how the world works

• Execute and understand of the use and limitations of null hypothesis significance testing

• Wrangle, summarize, test, and display data using R

• Independence

  • If you only learn one thing this term, focus on this!

Challenges

• Many basic statistical tests were developed a century ago.

  • Developed for small samples and hand calculations.
  • Example: minimum sample size = 30 per cell. Why?

• This class has variance.

We continue to build off of these tests today. For example, MLM is built off basics of regression and correlation. We need to understand these methods to move forward.

But the world is changing rapidly. Heuristics don’t apply, there are better ways to do things, and we’ve messed up these statistics a lot. We’ll discuss what’s often left out of most statistics courses.

Goal: don’t just do the test, understand what the test does, why you might want to use it, and how you can avoid abusing it

Generative AI (a different kind of challenge)

Why would someone want an employee with a graduate degree?

• Because they are an expert in the role of trait neuroticism in the psychological stress – physiological response – chronic disease progression cycle?

  • (No, not even at a university.)

Generative AI (a different kind of challenge)

• Because they have a wide range of skills.
  • End-to-end research project execution.
  • Data analysis and visualization.
  • Communication of complex and abstract concepts to non-expets. (Read: teaching and conference presentations)
  • Managerial experience (read: mentoring undergraduates)
  • Stakeholder alignment (read: collaboration and service)

Generative AI (a different kind of challenge)

• Because they can figure things out on their own.
  • Independence!
• What does this have to do with AI?

Generative AI (a different kind of challenge)

Generative AI is a great tool for productivity. Think of it like a car: if the goal is to get from point A to point B, a car will get you there fast.

• Coursework is not about getting from point A to point B; it’s about the journey. If you’re training for a marathon, the point isn’t to go 26 miles, it’s to run 26 miles.

Good uses of AI in this class: developing personalized quizzes to test knowledge, explaining concepts that I didn’t explain well, generating questions similar to the ones on the homework for additional practice, etc.

Road map

Where we’re going this term

Week(s)	Topics
1-2	Collecting and summarizing data
3-4	Probability
5	Sampling and making inferences about the unknown
6-9	Making a decision
10	Evaluating that decision

Graduate school vs undergraduate

Undergraduate

• Complete assignment
• Receive grade and (maybe) feedback
• Move on to next assignment

Graduate

• Attempt assignment
• Receive feedback
• Edit/retry/improve work
• Receive feedback
• … rinse, repeat…

My goals (practical)

• Teach you the basics of statistics and R.

• Challenge you – this class is designed to push you to your personal limit and then just beyond.

  • If you feel like you’re sailing through this class, come talk to me.
  • If you feel like you’re drowning, come talk to me.

• Create situations in which you can practice skills that will help throughout your career.

  • learning things on your own, asking for help, saying no, teaching others…

• Give everyone an A.

Format

• Learn
  • Lectures
  • Labs
  • Readings
• Practice
  • Weekly Quizzes (9 total)
  • Homework (3 total)

• Demonstrate mastery
  • Oral exam
• Get support (PSY only)
  • Journals (10 total)

There will be some overlap between the readings and lectures, and also between readings and lab. Sometimes there will be overlap between lab and lecture but not always. The goal of this term is to help give you the foundations you need in theory and in tools. We’ll put them together more in 612.

Quizzes are meant to test lecture and reading. Homework are meant to practice everything.

Talk about quiz policy.

Talk about homework policy.

Materials

• Everything is free and available online

• Materials for this course: uopsych.github.io/psy611

R and RMarkdown

This course will use R and RStudio for all analyses. Why?

• Statistical tool
• Reproducibility
• Workflow
• Marketability

If you want extra help, the library offers R workshops, as well as workshops on reproducibility (data management, database and spreadsheets, version control with Git, etc).

The library also has R consultants (GEs) available during most work hours to help troubleshoot.

https://library.uoregon.edu/research-data-management/training-workshops

Questions?

• Please read the rest of the syllabus on your own.

For the rest of today:

• some terminology
• scales of measurement

Kinds of statistics

• Descriptive
• Inferential

Which is more important? – Neither

Descriptive – about the data Inferential – about the world - assessing risk or uncertainty. How likely?

Kinds of statistics

• Exploratory
• Confirmatory

Exploratory —————————————- Confirmatory

Which is more important? – Neither

Exploratory – I don’t have a hypothesis or theory. I don’t know what’s going to happen Confirmatory – if theory X is true, then data should look like Y

Really more of a continuum

Kinds of research

• Experimental
• Observational

Typically we pair some kinds of statistical tests with experimental work and other kinds of tests with observational work. Examples?

Kinds of research

In reality, most statistical tests can be used with most kinds of research. It’s not so much the kind of research that matters, but which statistic helps to answer your question and what types of variables do you have?

• We’ll discuss the first point throughout the course
• Let’s discuss variables now

Scales of measurement

Four scales

• Nominal
• Ordinal
• Interval
• Ratio

Nominal

• response options are groups
• no specific order
• example: What is your favorite fruit? Blueberries, apples, coconuts

Ordinal

• response options are ordered
• no consistent distance between possible scores
• example: List the following fruits in order of preference: peaches, grapes, raspberries

Interval

• response options are ordered
• distance between response options is the same
• no meaningful 0
• example: On a scale from 0-5, how much do you like bananas?

Ratio

• response options are ordered
• distance between response options is the same
• 0 indicates an absence of something
• example: how many times have you eaten watermelon in the last 6 months?

Breakout groups

Pick one of the following constructs and come up with four ways of measuring it (one for each of the scales: nominal, ordinal, interval, ratio):

• Self-esteem
• Political preference
• Memory
• Dog-enthusiasm
• Anxiety

Scales of measurement

• Nominal

• Ordinal

• Interval

• Ratio

Why does this matter?

• Different amounts of information

• Different mathematical properties

  • Example: If Bobby scores 3 and Megan scores 6, can we say Megan scored twice as high as Bobby?

Scales of measurement

• Nominal

• Ordinal

• Interval

• Ratio

What is this?

• “On a scale from 1 to 6, how much does the word outgoing describe you?”

Scales of measurement

Two types

• Categorical
• Continuous

Scales of measurement

• Nominal

• Ordinal

• Interval

• Ratio

• Categorical
• Continuous

How do these two methods of describing scales match up?

Next time

Measurement validity

Before class, please download and install R and RStudio.