This glossary collects the key concepts, people, and ideas you will meet in this lesson. Use it as a reference while you work through the material, or as a review before assessments. Type in the search box to filter entries.

Introduction and Overview

Welcome to HSCI 341. If you've come from HSCI 230, you spent that course learning to read epidemiological research critically — to evaluate study designs, identify biases, and decide what evidence to trust. HSCI 341 picks up where that left off and asks you to do the work yourself: design valid studies, calculate measures of disease frequency and association, work through screening tests, and conduct surveillance and outbreak investigations. The bias inventory you built in 230 becomes the design checklist you'll use here. Lesson 1 sets up the conceptual foundation for the entire course — what epidemiology is, how scientific inference works (Susser & Susser, 1996), and what we mean when we say one thing “causes” another. Across four content sections we move from the discipline's history (Section 1), through the inferential logic that organizes any study (Section 2), into formal models of causation (Section 3), and finally to the counterfactual framework that underpins modern causal inference (Section 4).

Defining Epidemiology

Epidemiology is fundamentally about understanding the patterns, causes, and effects of health and disease in populations. Historically, epidemiologists have been concerned with identifying the "succession of events which result in the exposure of specific types of individuals to specific types of environment" — that is, the exposures and causal factors that drive disease.

Modern epidemiology aims to improve population health by integrating data from many disciplines and proposing interventions based on scientific evidence. The discipline focuses on identifying exposures — whether demographic factors, infectious agents, nutritional factors, toxins, or lifestyle elements — and evaluating their associations with health outcomes such as disease, quality of life, and mortality.

A Brief History of Causal Thinking

The way we think about what causes disease has shifted dramatically over the centuries (Susser & Susser, 1996a, 1996b). Understanding this history helps us appreciate the complexity of modern causal models. The eight cards below trace this evolution chronologically — from environmental theories in ancient Greece, through miasma and germ theory, to the multifactorial frameworks we use today. As you click through them, watch for one recurring tension: each era pulls between explaining disease at the level of individual mechanism (microbe, gene, biomarker) and explaining it at the level of populations and their environments. Section 2 will pick up that tension when we turn to the logic of scientific inference itself.

Key Historical Milestones

Click each card to learn more:

Introduction and Overview

Section 1 traced the discipline's history. This section moves from history to logic: how do epidemiologists actually reason from observed data to claims about cause and effect? The two forms of reasoning we cover here — induction and deduction — are the philosophical scaffolding underneath every study you'll design later in this course. We'll then map those modes of reasoning onto the concrete components of an epidemiologic study (Figure 1.1) and finish with directed acyclic graphs (DAGs), the modern formal tool for encoding causal assumptions before any data are touched.

Why Scientific Inference Matters

Epidemiology relies primarily on observational studies because many health-related problems cannot be studied under controlled laboratory conditions. Ethical concerns, practical limitations, and the complexity of real-world relationships all demand that we study humans in their natural environments. Drawing valid inferences from these studies requires both inductive and deductive reasoning.

Two Forms of Reasoning

The three tabs below define induction and deduction and add Bayesian thinking, which formalises how prior knowledge enters into our interpretation of any new study. Click through each tab and watch for the unifying point: no single mode of reasoning gives you certainty — epidemiologic claims always rest on a combination of observation, hypothesis testing, and consensus.

Inductive, deductive, and Bayesian reasoning are abstract. The next subsection makes them concrete by walking through the components of an actual study and showing where each form of reasoning enters.

Key Components of Epidemiologic Research

The overall structure of an epidemiologic study involves several interrelated components, each of which must be carefully managed to produce valid results. Read Figure 1.1 below as a roadmap for the rest of HSCI 341 — every box in the diagram corresponds to a topic we'll cover in detail later (sampling in Lesson 3, exposure measurement and questionnaires in Lesson 4, confounding throughout the course, and so on).

Directed Acyclic Graphs (DAGs)

The diagram in Figure 1.1 is informal — it shows boxes and arrows but does not commit to a precise causal meaning. A directed acyclic graph (DAG) is the formal version of that picture (Pearl, 1995; Greenland & Brumback, 2002). It is the working tool that modern epidemiologists use to write down their assumptions about how the world works before they touch the data, and to read off — mechanically — what those assumptions imply for analysis.

The Building Blocks

Every DAG, no matter how large, is built from a small handful of structural pieces. Learning to recognise these by sight is the entire skill:

How to Build One

Drawing a DAG is a substantive exercise, not a statistical one. The arrows come from your subject-matter knowledge of the system, not from p-values. A workable recipe:

1. Name the exposure (X) and outcome (Y). Put them on the page, with the exposure on the left.
2. List every other variable that could plausibly affect either X or Y — demographic, biological, social, environmental. Include unmeasured variables (draw them with dashed nodes); they belong in the diagram even if you cannot put numbers on them.
3. Draw an arrow from each variable to every variable it directly causes. Be ruthless about “direct” — if A → B only by going through C, you draw A → C and C → B, not A → B.
4. Check for cycles. If A causes B and B causes A, you need to add a time index or break the loop with intermediate variables. DAGs forbid feedback loops.
5. Read off the implications. Every “back-door” path from X to Y that does not pass through a collider must be blocked by adjustment. Mediators must be left alone if you want the total effect. Colliders must be left alone, full stop.

What DAGs Are For

A DAG plays four roles in an analysis. The first three are decided before you fit anything; the fourth is what makes it worth the trouble.

• It identifies confounders. Anything on a back-door path is a candidate for adjustment. Anything not on a back-door path is not — even if it is statistically “significant.”
• It flags variables you must not adjust for. Mediators (over-adjustment) and colliders (selection bias) ruin estimates if you control for them. The DAG tells you which is which.
• It makes assumptions criticisable. A reviewer can disagree with an arrow. They can’t disagree with a regression equation in the same way.
• It supports a transparent estimand. “The total effect of X on Y, adjusting for the back-door set {C₁, C₂}” is a precise target you can defend.

Mediation Analysis

A DAG tells you that a variable lies on the pathway from exposure to outcome. Mediation analysis is the next step: it puts numbers on how much of the exposure’s effect runs through the mediator versus around it.

The Classical Baron & Kenny (1986) Approach

Baron and Kenny’s causal-steps procedure is the recipe most students meet first. It runs three regressions and checks four conditions:

1. Step 1 — Total effect. Regress Y on X. The slope (call it c) must be significant. This is the total X → Y effect.
2. Step 2 — X predicts M. Regress M on X. The slope (a) must be significant. If X does not move M, M cannot be a mediator.
3. Step 3 — M predicts Y, controlling for X. Regress Y on both X and M. The coefficient on M (call it b) must be significant.
4. Step 4 — Compare c to c′. The X coefficient in Step 3 (c′) is the direct effect. If c′ is much smaller than c, the gap (c − c′, equivalently a×b) is the indirect effect through M. If c′ is essentially zero, the mediation is “complete”; otherwise it is “partial.”

With this scaffolding in place, the R exercise below puts the DAG side of the story into code. We will return to mediation explicitly in HSCI 410, where you will fit the same kind of model in R.

# One-time install (skip if you have done it before):
# install.packages(c("dagitty", "ggdag"))

library(dagitty)
library(ggdag)

# Smoking -> CHD, with age as a confounder of both.
g <- dagitty("dag {
  smoking -> chd
  age -> smoking
  age -> chd
  smoking [exposure]
  chd     [outcome]
}")

# Which variables do we need to adjust for, and which paths are open?
adjustmentSets(g)        # minimal sufficient adjustment set(s)
paths(g, "smoking", "chd")   # list every path between exposure and outcome

# Tidy plot
ggdag(g, layout = "circle") + theme_dag()

Introduction and Overview

Section 2 gave us the inferential machinery and the DAG vocabulary. This section pushes deeper into what we actually mean when we draw an arrow on a DAG and call it a “causal effect.” Three classical models structure the discussion: the component-cause model (necessary, sufficient, component causes), causal complements (which explain why the same cause has different observed effects in different populations), and the causal-web model. The section closes with the population attributable fraction — a quantity that turns the abstract notion of cause into a number a public-health planner can use.

What Is a "Cause"?

For practical purposes in epidemiology, a cause is any factor that produces a change in the severity or frequency of an outcome. Some causes operate at the biological level within individuals (such as a specific microorganism), while others operate at the group or population level (such as lifestyle, nutrition, or weather).

# Suppose a 2x2 cross-tabulation from a population-based study:
#                Disease+   Disease-
#  Exposed         180       820
#  Unexposed        60      940

a <- 180; b <- 820          # exposed: a = cases, b = non-cases
c <-  60; d <- 940          # unexposed

risk_e <- a / (a + b)             # risk in exposed
risk_u <- c / (c + d)             # risk in unexposed
RR     <- risk_e / risk_u

p_exp  <- (a + b) / (a + b + c + d)   # prevalence of exposure

# AFp = p_e * (RR - 1) / (1 + p_e * (RR - 1))     (Levin, 1953)
AFp    <- p_exp * (RR - 1) / (1 + p_exp * (RR - 1))
round(c(RR = RR, AFp = AFp), 3)