Measurement, Experimentation, & Causal Inference
Understanding the Tacit Assumptions Under Each Stage of the Research Process
Welcome to our workshop on measurement, hypothesis testing, experimentation, and causal inference. These topics are often taught separately and use different terminology. This workshop brings them together in one framework and shared language.
The workshop is built around a core question that runs through all three modules: how do we draw valid inferences from data? Measurement shapes what we can even ask. Hypothesis testing governs what counts as evidence. Causal reasoning determines what we can conclude.
Participants will see how causal inference depends on the assumptions of hypothesis testing in both experimental and observational settings, and how those assumptions rest on measurement principles from psychometrics. Using a common set of problems, we will connect these core methods in practice.
For most topics, I will provide both an experimental example and a secondary-data example. We will also examine common pitfalls in experimental and non-experimental research through hands-on R tutorials and interactive simulations.
Science combines observable data with assumptions to infer latent or general processes. That makes it essential to understand both the properties of what we measure or manipulate and the explicit and implicit assumptions behind our research. Many research errors arise when these assumptions go unexamined. By the end of the course, participants will be able to identify and evaluate the key assumptions in a published paper, improving how they interpret and generalize findings.
0.1 Modules
0.1.1 Module 1: Measurement
When does your scale measure what you think it measures — and when does it measure something else entirely?
Sections
0.1.2 Module 2: Hypothesis Testing
What does a p-value actually tell you? How does randomization create valid inference, and when does it fail?
Sections
0.1.3 Module 3: Causal Inference
When can we say that X caused Y? What methods go beyond randomized experiments, and what do they assume?
Sections