Map Analytics Export Format

The “Map Analytics” option of the Export dialog in DRA 2020 generates a JSON file that contains the results of analyzing a map. This is an example.

There are five sections in the file. They combine the metrics used in the Analytics view and the Advanced view.¹ To help you map the contents in the file to those metrics, each description notes a location in brackets.

The bias Section

This section contains the following key/value pairs:

• bestS — The number of Democratic seats closest to proportional representation, e.g., 6 [Analytic / Proportionality / in Notes]
• bestSf — The fraction² of the total number of seats that that represents, e.g., 0.4615 [Not shown]
• fptpS — The estimated number of whole Democratic seats using first-past-the-post accounting, e.g., 3 [Not shown]
• estS — The estimated number of Democratic seats using fractional seat probabilities, e.g., 3.5336 [Analytic / Proportionality / in Notes]
• estSf — The fraction of the total number of seats that that represents, e.g., 0.2718 [Not shown]
• bS50 — The seat bias (αₛ) expressed as a fraction of the number of districts, e.g., 0.199 [Advanced / Bias Measures / Seats bias]
• bV50 — The votes bias (αᵥ) expressed as a fraction, e.g., 0.0399 [Advanced / Bias Measures / Votes bias]
• decl — The declination angle (δ) in degrees, e.g., 36.4084; this may be undefined [Advanced / Bias Measures / Declination]
• gSym — The global symmetry (GS) expressed as a fraction, e.g., 0.0579 [Advanced / Bias Measures / Global symmetry]
• gamma — The gamma measure (γ) expressed as a fraction, e.g., 0.2074 [Advanced / Bias Measures / Gamma]
• eG — The efficiency gap (EG) expressed as a fraction, e.g., 0.2046 [Advanced / Bias Measures / Efficiency gap]
• bSV — Partisan bias (β) expressed as a fraction, e.g., 0.19 [Advanced / Bias Measures / Partisan bias]
• prop — The deviation from proportionality (PR) expressed as a fraction, e.g., 0.2164 [Advanced / Bias Measures / Disproportionality]
• mMs — The mean-median difference using the statewide Democratic vote share (mM), e.g., 0.0517 [Advanced / Bias Measures / Mean–median]
• tOf — The turnout bias expressed as a fraction, e.g., 0.0033 [Not shown]
• mMd — The mean-median difference using average the district vote share, e.g., 0.0484 [Not shown]
• LO —Lopsided outcomes (LO) expressed as a fraction, e.g., 0.0756; this may be undefined [Advanced / Bias Measures / Lopsided outcomes]
• deviation — The deviation from the number of Democratic seats closest to proportional representation,³ e.g., 0.1897 [Analytics / Proportionality / Deviation]
• score — The normalized deviation,⁴ e.g., 11 [Analytics / Proportionality / Rating]

The responsiveness Section

This section contains the following key/value pairs:

• littleR — The responsiveness (ρ) which is the slope of the seats–votes curve at the statewide vote share, e.g., 1.7615 [Advanced / Responsiveness Measures / Responsiveness]
• rD — The estimated number of responsive districts using fractional seat probabilities, e.g., 1.8993 [Advanced / Responsiveness Measures / Responsive districts]
• rDf — The estimated number of responsive districts as a fraction of the number of districts, e.g., 0.1461 [Not shown]
• bigR — The overall responsiveness or winner’s bonus (R), e.g., 19.3324 [Advanced / Responsiveness Measures / Overall responsiveness]
• mIR — The minimal inverse responsiveness measure (ζ), e.g., 0.4677; this is not shown with the Advanced metrics [Not shown]
• cSimple — The number of districts with a vote share that fall into the range [45–55%], e.g., 2 [Statistics / in Notes]
• cD — The estimated number of competitive districts using fractional seat probabilities, e.g., 1.8993; the probability function may differ from that used by rD [Not shown]
• cDf — That number expressed as a fraction of the number of districts, e.g., 0.1461 [Analytics / Competititiveness / Competitiveness]
• score — That fraction normalized, e.g., 19 [Analytics / Competitiveness / Rating]

The minority Section

This section contains four key/value pairs:

• score — The number opportunity & coalition districts combined and normalized, e.g., 48 [Analytics / Minority Rights / Rating]
• pivotByDemographic — An object described below that is essentially a pivot table of the minority demographic portion of the table in district Statistics view.
• opportunityDistricts — The estimated number of districts where individual minority groups have the opportunity to elect representatives using fractional seat probabilities, e.g., 0.9909 [Not shown]
• coalitionDistricts — The estimated number of districts where all minorities combined have the opportunity to elect representatives using fractional seat probabilities, e.g., 2.8009 [Not shown]

Pivot Table

Each “row” in the pivot table represents the details for one minority group with these keys:

• minority
• black
• hispanic
• pacific
• asian
• native

Pivot Table Columns

Each “column” in the pivot tables represents details for that minority group with these numeric key/value pairs:

• pct35_40 — The number of districts were the VAP % for the minority group falls into the range [0.35, 0.40), where percentages are expressed as fractions
• pct40_45 — The same for the range [0.40, 0.45)
• pct45_50 — The same for the range [0.45, 0.50)
• pct50_55 — The same for the range [0.50, 0.55)
• pct55_60 — The same for the range [0.55, 0.60)
• pct60Plus — The same for the range [0.60, 1.00]
• vapPct — the statewide VAP % for the minority group
• propSeats — a proportional number of seats for that statewide VAP %

The all Minority row represents potential coalition districts, where the others reflect opportunities for individual minority groups. [Analytics / Minority Rights / <table>]

The compactness Section

This section contains three key/value pairs:

• score — The normalized Reock and Polsby–Popper scores averaged together, e.g., 36 [Analytics / Compactness / Rating]
• reock — The raw sub-property shows the average Reock score for the districts expressed as a fraction, e.g., 0.3373, while the normalized sub-property reports the normalized Reock score, e.g., 35 [Analytics / Compactness / Reock]
• polsby — The raw sub-property reports the average Polsby–Popper score for the districts expressed as a fraction, e.g., 0.2422, while the normalized sub-property shows the normalized Polsby–Popper score, e.g., 36 [Analytics / Compactness / Polsby–Popper]

The splitting Section

This section contains four key/value pairs:

• score — The normalized county- and district-splitting scores averaged together, e.g., 52 [Analytics / Splitting / Rating]
• county — The raw sub-property shows the average county-splitting score for the districts, e.g., 1.1523; the minimum value is 1.0, while the normalized sub-property reports the normalized county-splitting score, e.g., 65 [Analytics / Splitting / County splitting]
• district — The raw sub-property shows the average district-splitting score for the districts, e.g., 1.5185; the minimum value is 1.0, while the normalized sub-property reports the normalized district-splitting score, e.g., 38 [Analytics / Splitting / District splitting]
• details

The details key has five key/value pairs:

• nSplits — The number of times counties are split across districts, e.g., 12 [Analytics / Splitting / in Notes]
• countiesSplitUnexpectedly — A list of the counties that are split even though they have fewer people than a target district [Analytics / Splitting / in Notes]
• unexpectedAffected — The fraction of the total population that is affected by the counties splitting unexpectedly, e.g., 0.3097 [Analytics / Splitting / in Notes]
• nSplitVTDs — The number of VTDs that are split.
• splitVTDs — The GEOIDs of the VTDs that are split.

See Measuring County & District Splitting for details on how the county- and district-splitting metrics are calculated.

Footnotes

1. See Advanced Measures of Bias & Responsiveness for details on what the advanced metrics mean.
2. All fractions are in the range [0.0–1.0]. Multiply fractions by 100 to get the corresponding percentages.
3. The deviation and prop values in the bias section can differ, because they use different baselines. The latter uses the seat share that equals the vote share which, in general, does not correspond to a possible first-past-the-post result. The former uses the seat share that corresponds to the number of Democratic seats closest to proportional representation.
4. All scores/ratings are normalized to the range [0–100] such that bigger is better.