Data Analysis for PolicyMaking in Georgia
Module 4 · Disparate Impact Analysis › 4.5 Module 4 checkpoint
Subsection 4.5

Module 4 checkpoint

~3 min

Module 4 Checkpoint

You have completed Module 4: Disparate Impact Analysis. These four questions cover disparate impact concepts, RPV analysis, proximity modeling, and mail ballot rejection rates.

Module 4 Recap — Disparate Impact Analysis Methods Disparate Impact Race-neutral policy → unequal effects 3 steps: ID, quantify, evaluate necessity Post-Callais: evidence of intent in redistricting RPV Analysis Ecological inference (EI) estimates group voting Shows political cohesion (Gingles Factor 2) Cannot reveal why voters chose as they did Proximity Modeling Distance to polling places, drop boxes, wait times Join addresses to shapefiles + demographic data Fulton 2020: 3× farther for majority-Black blocks Mail Ballot Rejection Rejected ÷ Returned × 100 Compare across counties by demographic composition Sig. mismatch = highest racial disparity + addressable GA 2020: Black voters rejected at ~2.7× rate
Diagram 4.5 · Module 4 recap. Four methods for disparate impact analysis in voting rights contexts.

What does it mean to say a policy has 'disparate impact' on a racial group?

Correct! Disparate impact means a facially neutral policy produces unequal effects on a protected group. It does not require proof of intentional discrimination. Not quite. Disparate impact = unequal effects from a race-neutral policy. No proof of intent is required to show disparate impact — but after Callais, intent must be shown to obtain a redistricting remedy.

Ecological inference (EI) is used in RPV analysis to estimate which of the following?

Correct! Ecological inference uses aggregate precinct-level data (racial composition + vote totals) to estimate how voters of each racial group likely voted — since individual ballots are secret. Not quite. EI estimates group voting patterns from aggregate precinct data. It does not predict future turnout or measure compactness.

In a Fulton County polling place proximity analysis, you find that census blocks with majority-Black populations are on average 2.4 miles from the nearest polling place, compared to 0.8 miles for majority-white blocks. What is the correct way to describe this finding?

Correct! The correct framing is descriptive: the blocks are 3× farther on average. That disparity is a signal worth investigating. It does not by itself prove intent or a VRA violation — but it is the kind of evidence that goes into a disparate impact analysis. Not quite. The correct framing is precise and descriptive: 3× farther on average, worth investigating. Avoid overclaiming violation or intent from a proximity calculation alone.

Mail ballot rejection rate research in Georgia found that one rejection reason showed the highest racial disparity and was the most policy-addressable. Which reason was it?

Correct! Signature mismatch showed the highest racial disparity in rejection rates and is most policy-addressable because it involves discretionary judgment by local officials. It is the strongest candidate for targeted policy reform. Not quite. Signature mismatch is the most discretionary reason (a local official makes the call) and showed the highest racial disparity. Late returns and missing signatures are more difficult to address through process reform.