Regression Discontinuity

Regression discontinuity is a quasi-experimental research design that estimates a policy's causal effect by comparing people just above and below a cutoff. In Intro to Public Policy, it is a common way to evaluate programs like scholarships, benefits, or school interventions.

What is Regression Discontinuity?

Regression discontinuity is a policy evaluation method that uses a cutoff or threshold to estimate whether a program actually caused an outcome change. In Intro to Public Policy, you usually see it when a government policy gives treatment only to people above or below a score, income level, age line, or other rule.

The basic idea is simple: compare people who sit right next to each other on either side of the cutoff. If one group gets the policy and the other does not, and everything else is similar enough near that line, then differences in outcomes can be linked to the policy itself. That makes regression discontinuity a quasi-experimental design, meaning it is not a randomized experiment, but it can still give strong causal evidence.

A common example is a scholarship program that goes to students scoring 80 or higher on an entrance exam. Students with scores of 79 and 80 are probably very similar in background and ability, but only one side receives the scholarship. If the 80-plus group performs better later, the policy analyst can ask whether the scholarship caused the difference.

The strength of the method comes from the cutoff. Near the threshold, assignment looks almost random because tiny changes in the running variable, like a test score or income amount, decide who gets treated. That is why this design is so useful in public policy evaluation, where true randomized trials are often impossible for ethical or political reasons.

The big assumption is that people cannot easily manipulate the cutoff. If families can coach scores, agencies can reclassify cases, or applicants can game their reported income, then the comparison stops being clean. Analysts also need enough cases close to the threshold to make the comparison meaningful, because the design is about what happens near the line, not across the entire population.

Regression discontinuity is not just about finding a difference. It is about using a rule-based policy to make a careful causal claim: when two similar cases land on different sides of a cutoff, did the policy change what happened next?

Why Regression Discontinuity matters in Intro to Public Policy

Regression discontinuity matters in Intro to Public Policy because policy evaluation is only useful when it can separate real effects from good intentions. A program may look successful on paper, but this method lets you ask whether the policy changed outcomes for the people it targeted.

That matters a lot in areas like education, healthcare, and social benefits, where policymakers often use eligibility rules instead of random assignment. If a city awards tutoring support to students below a certain test-score cutoff, regression discontinuity can help show whether the tutoring actually raises later grades or graduation rates.

The method also shows how public policy research turns messy real-world programs into evidence. You do not need a perfect experiment to make a strong argument, but you do need a design that makes comparison groups believable. Regression discontinuity is one of the cleanest ways to do that when a policy already has a strict threshold.

In class, this term usually shows up when you are evaluating whether a policy works, explaining why a design supports causal inference, or comparing it to other evaluation tools. It gives you a way to talk about evidence-based policymaking without pretending every policy can be tested like a lab experiment.

How Regression Discontinuity connects across the course

Quasi-Experimental Design

Regression discontinuity is one type of quasi-experimental design. Both aim to estimate causal effects without random assignment, but regression discontinuity does it by focusing on a cutoff rule. When you see a policy sorted by score, age, or income threshold, that is usually your clue to think about whether a quasi-experimental design is being used.

Causal Inference

This method is all about causal inference, or figuring out whether a policy actually caused an outcome change. Instead of just describing correlation, regression discontinuity tries to make a stronger causal claim by comparing cases right around the threshold. That is why it is used in evaluation sections of public policy courses.

Evidence-Based Policymaking

Regression discontinuity gives policymakers evidence they can use when deciding whether to expand, revise, or end a program. It turns an existing cutoff into a source of evaluation data. In evidence-based policymaking, that kind of analysis matters because it helps leaders justify decisions with results, not just political preferences.

Randomized Controlled Trials

Randomized controlled trials are often treated as the gold standard because they assign treatment randomly. Regression discontinuity is not random assignment, but it can approximate that logic near the cutoff. If a class asks why policy analysts use it, the answer is usually that RCTs are unavailable, expensive, or unethical.

Is Regression Discontinuity on the Intro to Public Policy exam?

A policy evaluation question may give you a program with a cutoff and ask how the analyst should measure its effect. Your job is to identify regression discontinuity, explain that the cases just above and below the threshold are compared, and say why that supports causal inference. If you see a scholarship, tax credit, or benefits program tied to a score or income line, that is your signal.

On short answers or essays, you may need to explain the limitation too. Mention that the design works best when people cannot manipulate the cutoff and when the groups near the threshold are otherwise similar. If the question asks for strengths and weaknesses, pair the strong causal logic with the fact that it only estimates effects near the cutoff, not for everyone in the program.

Regression Discontinuity vs Difference-in-Differences

Both methods are used for policy evaluation without random assignment, but they rely on different logic. Difference-in-differences compares changes over time between groups, while regression discontinuity compares cases on either side of a cutoff at one point in time. If the prompt mentions a threshold, think regression discontinuity. If it mentions before-and-after trends, think difference-in-differences.

Key things to remember about Regression Discontinuity

• Regression discontinuity estimates a policy's effect by comparing cases just above and below a cutoff.

• It is a quasi-experimental design, so it can support causal inference even when random assignment is not available.

• The method works best when the cutoff is strict and people cannot manipulate which side they land on.

• It is most useful for policy evaluation in programs with clear eligibility rules, like scores, income levels, or age thresholds.

• The result usually tells you about the effect near the cutoff, not about every possible participant in the policy.

Frequently asked questions about Regression Discontinuity

What is regression discontinuity in Intro to Public Policy?

It is a policy evaluation method that uses a cutoff rule to compare people just above and just below the threshold. In Intro to Public Policy, it helps analysts estimate whether a program like a scholarship, tax benefit, or intervention actually caused an outcome change.

Why is regression discontinuity considered quasi-experimental?

It is quasi-experimental because it does not randomly assign treatment, but it still creates a strong comparison group near the cutoff. People close to the threshold are often similar enough that the policy assignment looks almost random, which strengthens causal claims.

What is a real example of regression discontinuity in policy?

A common example is a scholarship given only to students scoring 85 or higher on an exam. Analysts can compare students with scores of 84 and 85 to see whether receiving the scholarship changed later academic outcomes.

What is the biggest weakness of regression discontinuity?

The biggest weakness is that it only works well if the cutoff is hard to manipulate. It also tells you about the effect near the threshold, so it may not describe what happens to people far away from the cutoff.