STAT 6160 — Experimental Design

About the course

The principles in this course were worked out for agricultural field trials a century ago, and these techniques turned out to be unusually portable! Today, the same problems appear in scientific experiments, clinical studies, online A/B tests and product experimentation platforms, engineering and process optimization, benchmarking studies in machine learning, and policy evaluation (in addition to many other settings).

Throughout the course, we will emphasize not only how to analyze an experiment after the data is collected, but also the more important question: how should we design the experiment before collecting the data?


Course Information

Course STAT 6160 — Experimental Design (Fall 2026)
Class Tuesdays & Thursdays, 9:30–10:45 AM
Physics Building 217
Instructor Siddharth Vishwanath
Email sidv@virginia.edu
Please start the email subject with “[STAT 6160]”
Office Hours Tuesdays, 3:30–5:00 PM
Prerequisite Graduate level exposure to statistical inference, linear algebra, and linear models


Important Dates

Weekly Summaries Every Sunday night @ 11:59 PM
HWs On the deadline date @ 11:59 PM
Project Proposal TBD
Final Project TBD

Lectures

  1. Introduction
  2. Refresher
  3. Randomization, contrasts, and one-way ANOVA
  4. One-way ANOVA (extra)
  5. Blocking
  6. Full and fractional factorial designs

Learning Objectives

Design for credible comparisons
Use randomization, replication, blocking, and careful assignment mechanisms to separate treatment effects from unwanted variation.
Understand variation
Not all experimental noise is interchangeable. We will learn to identify meaningful sources of variation, decide which should be controlled through design, and understand how different designs change both the analysis and the inference.
Discover interactions
Factorial experiments let us ask not only whether a treatment works, but when, where, and under what conditions it works. We will develop ways to design and interpret experiments in which the effect of one factor depends on another.
Work computationally
We will use Python/R and other computational tools to design experiments, visualize results, check assumptions, explore uncertainty, and communicate conclusions.


Textbooks

  1. Experiments : Planning, Analysis, and Optimization by Wu and Hamada
    available online through UVA library: [link]
  2. A First Course in Experimental Design: Lecture Notes by Art Owen
    available online on Art Owen’s website: [link]
  3. Statistics for Experimenters by Box, Hunter, and Hunter
    physical copy available in the UVA library: [link]
  4. Design and Analysis of Experiments by Douglas Montgomery
    available online through UVA library: [link]

Course Project

will be updated soon

STAT 6160 · Experimental Design
Siddharth Vishwanath · Department of Statistics · University of Virginia