in

American Flags Are Not Useful Political Clues, And Other Lessons From Google Street View

We recently showed Times readers images culled from Google Street View of 10,000 neighborhoods around the United States. Could readers guess, we wondered, how residents in a given place voted in the 2020 presidential election just by eyeballing a typical street scene?

Our neighborhoods were representative of where American voters live, meaning they included about the right number of urban precincts, like this one in brownstone Brooklyn …



Imagery source: Google

… and more rural areas like this one outside Effingham, Ill.:



Imagery source: Google

Not surprisingly, readers nearly aced places like these, accurately guessing that city residents generally voted for Joe Biden and farm-country ones for Donald J. Trump.

But the game was harder than many readers expected, in part because most voters live in between these two extremes.

They live in places like this around Las Vegas …



Imagery source: Google

… or outside Spokane, Wash. …



Imagery source: Google

… or Valparaiso, Ind.:



Imagery source: Google

Places like these confounded our readers. And that may be expected: These precincts were very closely decided in 2020 and don’t fit neatly into any stereotype about Red or Blue America.

Our Street View scenes also have very few people in them, and so they offer few hints about the demographics of local voters. They show, instead, information about driveways, lawns, cars and houses.

Since our quiz was first published a little over a week ago, Times readers have made more than 15 million guesses about the politics of these 10,000 neighborhoods, giving us a better sense of how people perceive partisanship when given just these environmental clues. Most Times readers appeared to recognize the relationship between population density and politics (the denser a community, the more reliably Democratic it is, in general).

But as a group, Times readers did show a subtle bias in their guesses, toward Mr. Trump: When shown a scene from a competitive precinct, they were more likely to guess that Mr. Trump carried it than Mr. Biden.

We also suspect that some readers ascribed too much political meaning to pickup trucks and American flags. And readers looking for socioeconomic signals may have been foiled, too.

Here is some of what we learned from all these guesses — the right ones and the wrong ones. But first, if you haven’t played the game yet, we suggest you do that now.

The easy ones: packed cities and open skies

Below are the scenes that garnered the most accurate guesses from readers (here we’re showing you a small thumbnail of each, but in the game you can pan around every street in more detail). Readers almost universally pegged the neighborhoods at left as voting for Mr. Biden, and those at right as voting for Mr. Trump:



Biden precincts

New York

11498

New York

2263

Brooklyn, N.Y.

6129

Brooklyn, N.Y.

3604

New York

19719

Bronx, N.Y.

15639

New York

17728

New York

16167

New York

13401

Brooklyn, N.Y.

16814

New York

11127

New York

16905

Astoria, N.Y.

3274

Brooklyn, N.Y.

7862

Brooklyn, N.Y.

3086

Corona, N.Y.

9638

Brooklyn, N.Y.

10711

New York

5315

New York

17954

Boston

17257

San Francisco

11752

Brooklyn, N.Y.

14496

New York

15396

Brooklyn, N.Y.

17821

Bronx, N.Y.

8256

New York

16246

New York

4483

New York

2356

Philadelphia

19685

Bronx, N.Y.

8946

Sunnyside, N.Y.

5610

Brooklyn, N.Y.

12585

Brooklyn, N.Y.

9265

New York

16014

San Francisco

16454

Brooklyn, N.Y.

11053

Bronx, N.Y.

9405

Bronx, N.Y.

13288

Brooklyn, N.Y.

9760

Brooklyn, N.Y.

5360

New York

12021

Brooklyn, N.Y.

8979

New York

6259

New York

2800

New York

9217

Brooklyn, N.Y.

7146

New York

9998

Brooklyn, N.Y.

4492

Trump precincts

Augusta, Kan.

16330

Douglass, Kan.

147

Anna, Ohio

4191

Cuero, Texas

8149

La Moille, Ill.

16809

Dodgeville, Wis.

4949

Snyder, Okla.

15184

Grifton, N.C.

5538

Lexington, Neb.

10950

White City, Kan.

14717

Stanley, N.D.

14736

Fredonia, Kan.

17722

Royston, Ga.

16052

Effingham, Ill.

6960

Rush Center, Kan.

6765

Scottsbluff, Neb.

8784

Tuscola, Ill.

2146

Waelder, Texas

5512

Baltimore, Ohio

16186

Oakwood, Ill.

9597

Cleburne, Texas

15179

Norman, Okla.

10854

Roanoke, Ind.

10406

Walla Walla, Wash.

6421

Rolfe, Iowa

13789

Wiggins, Colo.

202

Saint Paul, Ind.

18829

Snyder, Texas

18134

Alamosa, Colo.

11376

Kenyon, Minn.

16129

Tremont, Miss.

17432

Green Bay, Wis.

10663

Conway, S.C.

13957

Madras, Ore.

13714

Council Bluffs, Iowa

19846

Dewey, Okla.

11617

Swanville, Minn.

9033

Granton, Wis.

1899

Oologah, Okla.

13917

Mathis, Texas

14137

Shelby, Ohio

16390

Oakwood, Ohio

10412

Farmer City, Ill.

11087

Greenview, Ill.

992

Ellensburg, Wash.

6728

Fennville, Mich.

14741

Boelus, Neb.

7910

Hartwell, Ga.

7531


Imagery source: Google

Clearly, Times readers have learned well that urban and rural America have different politics — a geographic divide that wasn’t so stark just a few generations ago.

The hardest ones: rural Biden, urban Trump

The scenes readers most frequently got wrong were often ones that broke this general rule. If you tripped over these, we get it (we tripped over them, too).

These images look like the reverse of the photos above: Now we have rural places that backed Mr. Biden on the left and urban neighborhoods that voted for Mr. Trump on the right.



Biden precincts

Santa Rosa, Calif.

11076

Greenwood, Miss.

15937

Hollandale, Wis.

3121

Dudley, N.C.

3441

Hollandale, Wis.

2497

Hephzibah, Ga.

8592

Gunnison, Colo.

10389

Tracy, Calif.

2091

Orangeburg, S.C.

7747

Moscow, Tenn.

2278

Hammonton, N.J.

19403

Houston, Miss.

932

Santa Fe, N.M.

3849

Centreville, Ill.

4286

Norwood, N.Y.

15220

Broomfield, Colo.

11413

Hemingway, S.C.

10321

Olympia, Wash.

8251

Rose Hill, N.C.

2309

Huntsville, Texas

4405

Morganton, N.C.

6182

Waterloo, Iowa

10161

Puyallup, Wash.

18070

Bowman, S.C.

19625

MacOn, Ga.

16601

Mayesville, S.C.

17795

Hartsville, S.C.

19870

Peru, N.Y.

2780

Bishopville, S.C.

17549

Granville, Ohio

7499

Mount Vernon, Wash.

12544

Santa Fe, N.M.

4262

Westminster, Colo.

16636

Canton, Miss.

19721

Fayetteville, Ark.

1034

Osyka, Miss.

18818

Alice, Texas

17559

Waterloo, Iowa

17433

Rewey, Wis.

11330

Shubuta, Miss.

12140

Georgetown, Texas

16367

Winterville, N.C.

7327

Williamston, Mich.

13002

Horn Lake, Miss.

8797

Spartanburg, S.C.

14781

Whiting, Vt.

2012

Byram, Miss.

16028

Chimacum, Wash.

6468

Trump precincts

Brooklyn, N.Y.

7522

Woodhaven, N.Y.

8349

Brooklyn, N.Y.

19785

Brooklyn, N.Y.

18928

Brooklyn, N.Y.

17198

Brooklyn, N.Y.

5673

Brooklyn, N.Y.

15781

Brooklyn, N.Y.

16174

Brooklyn, N.Y.

268

Brooklyn, N.Y.

11879

Brooklyn, N.Y.

3911

Brooklyn, N.Y.

3741

Brooklyn, N.Y.

581

Middle Village, N.Y.

17303

Brooklyn, N.Y.

14197

Brooklyn, N.Y.

16194

Forest Hills, N.Y.

2761

Brooklyn, N.Y.

19655

Ridgewood, N.Y.

6277

Brooklyn, N.Y.

10443

Staten Island, N.Y.

13442

Brooklyn, N.Y.

15202

Brooklyn, N.Y.

4630

Brooklyn, N.Y.

12404

Rego Park, N.Y.

13939

Forest Hills, N.Y.

10244

Brooklyn, N.Y.

2344

Philadelphia

14788

Brooklyn, N.Y.

2109

Charleston, S.C.

11867

Brooklyn, N.Y.

4095

Brooklyn, N.Y.

12430

Yonkers, N.Y.

9355

Brooklyn, N.Y.

2651

Middle Village, N.Y.

8652

Brooklyn, N.Y.

7452

Middle Village, N.Y.

9784

Brooklyn, N.Y.

13417

Philadelphia

10115

Miami

11621

Brooklyn, N.Y.

472

Brooklyn, N.Y.

9928

Brooklyn, N.Y.

14616

Maspeth, N.Y.

12473

Philadelphia

5293

Brooklyn, N.Y.

12938

Brooklyn, N.Y.

7189

Chicago

11300


Imagery source: Google

The single greatest outliers in the whole collection, shown among the pictures above, were in the Brighton Beach neighborhood of Brooklyn (in a precinct that voted for Mr. Trump by 52 points) and in a corner of Santa Rosa, Calif. (where Mr. Biden prevailed by 46 points). Readers almost universally got these two wrong.

The role of income

If you could accurately estimate the income level of a neighborhood, that would be another helpful clue. Both the richest and poorest places voted overwhelmingly for Mr. Biden. But it’s not so easy to determine income — even at the extremes — from a single Street View image.

Among the precincts within the highest-income census tracts in our sample — places where the average household earns more than $150,000 a year — 81 percent voted for Mr. Biden (readers guessed Biden about 61 percent of the time). Below we’re showing you the very richest of those rich places, which tilt even more heavily toward Biden. There are a few gated walls here. But other scenes look almost rural, spoiling our density rule. A lot of money can buy you a very large wooded lot with few neighbors nearby and a house set far back from the road.



Corona Del Mar, Calif.

19819

Plano, Texas

9729

Hewlett, N.Y.

12034

Southlake, Texas

11231

Paradise Valley, Ariz.

15034

Southlake, Texas

15681

Alamo, Calif.

10187

Woodside, Calif.

17207

Lloyd Harbor, N.Y.

7175

Los Altos Hills, Calif.

14044

Chevy Chase, Md.

11180

New York

12615

Potomac, Md.

18704

Ridgewood, N.J.

18307

Woodside, Calif.

19976

Atherton, Calif.

19534

Pinecrest, Fla.

13227

Piedmont, Calif.

619

Atherton, Calif.

4685

Fairway, Kan.

12811

Redwood City, Calif.

16212

Fairfield, Conn.

16700

Syosset, N.Y.

4332

Kenilworth, Ill.

14613

Rye, N.Y.

11181

New York

14965

Villanova, Pa.

8631

Menlo Park, Calif.

10218

Cabin John, Md.

11423

Scarsdale, N.Y.

14522

Yardley, Pa.

9847

Gaithersburg, Md.

15573

Libertyville, Ill.

9240

Medina, Wash.

16201

Pacific Palisades, Calif.

6319

Chatham, N.J.

17647

New York

19609

Danville, Calif.

1519

Dallas

3951

San Francisco

13044

Rye, N.Y.

10524

Rye, N.Y.

1638

Winnetka, Ill.

18823

Menlo Park, Calif.

11947

Menlo Park, Calif.

3309

New York

12021

Woodinville, Wash.

18458

Lake Bluff, Ill.

12721


Imagery source: Google

Among census tracts below $27,000 in median income — close to the poverty line for a family of four — 89 percent of precincts voted for Mr. Biden (readers said 64 percent). But for the most part, those images don’t telegraph clear signals about income, either.

One of them shows a sidewalk homeless encampment on Skid Row in Los Angeles. But others capture a high-rise public housing development in Manhattan, a child care center in Vicksburg, Miss., and a block of handsome single-family homes in Huntington, W.Va.

Tricks readers said they tried: flags, trucks and sidewalks

Readers told us they had varying strategies for navigating this game. Some focused mostly on housing density. Others looked for the make and model of cars, or even the number of cars in a driveway (many cars might mean overcrowded working-class households with more than two working adults).

Some readers treated the presence of sidewalks as a proxy for density: Sidewalks are often missing in communities where there’s not much nearby to walk to. We also heard from readers confident that American flags and pickup trucks were clear indications of a community’s more conservative politics.

To test some of these ideas, we selected hundreds of images from our collection where these features appeared prominently.

Flags

We found that American flags don’t reveal much about a place’s politics. Among scenes we assessed that featured a prominent American flag, those places nearly evenly split 50-50 between Trump and Biden precincts. (Readers guessed the streets were in Trump precincts by about 60 percent to 40 percent.)

Some sample flags from Renton, Wash.:



Imagery source: Google

Philadelphia:



Imagery source: Google

Plainfield, Vt.:



Imagery source: Google

Overland Park, Kan.:



Imagery source: Google

Seneca, S.C.:



Imagery source: Google

Trucks

Pickup trucks were even less useful as a political signal of a Trump area. If you suspected a truck meant a more Republican-leaning precinct, you were more likely to be wrong than right: About 57 percent of places we identified with prominent pickup trucks were in Biden precincts. As a group, readers guessed Trump about 60 percent of the time in places like these.

Below, some of the places where pickup trucks were prominent in our sample:

Jonesboro, Ark.:



Imagery source: Google

Albuquerque:



Imagery source: Google

Manhattan:



Imagery source: Google

Bakersfield, Calif:



Imagery source: Google

Los Angeles:



Imagery source: Google

Sidewalks

Sidewalks were much more common in these scenes than flags or trucks. And they were indeed a decent way to approximate density — and, by extension, to gauge politics. Among the sidewalk scenes we found, about 70 percent were in places that Mr. Biden carried.

Around Tucson:



Imagery source: Google

Toledo, Ohio:



Imagery source: Google

Sonoma, Calif.:



Imagery source: Google

Seekonk, Mass.:



Imagery source: Google

Lewisville, Texas:



Imagery source: Google

The presence of sidewalks might be particularly useful in sorting inner-ring suburbs built before World War II (which lean Democratic) from newer, more car-dependent exurbs (which lean Republican), or in identifying the close-in neighborhoods of smaller towns.

The magic density tipping point

As we suspected, readers tended to struggle with suburban-looking neighborhoods, the sort that are, on average, more evenly divided between Democratic and Republican voters.

Population density is a helpful tool — if you let that be your only guide, you could do pretty well in this game. Where density doesn’t help as much is at the threshold between urban and rural.

If we’re being exact, that threshold is in places that have about 848 voters per square mile. Nearly 75 percent of precincts denser than that backed Mr. Biden, and nearly 75 percent of precincts less dense than that backed Mr. Trump. Understandably, most of us are terrible at picturing population density in the real world — what do places with 848 voters per square mile look like?

Well, they look like this:



Hendersonville, Tenn.

15240

Manhattan, Kan.

4631

Los Lunas, N.M.

18336

Spencerport, N.Y.

3633

Mokena, Ill.

17696

Avondale, Ariz.

3108

Vernon, Conn.

3258

Sharpsville, Pa.

13478

Weymouth, Mass.

13544

Destin, Fla.

2032

Avon, N.Y.

11453

Dayton, Ohio

15187

Rochester, N.Y.

20010

Sumter, S.C.

6784

Summerville, S.C.

4318

Brandon, Miss.

6669

Des Moines

12633

Alton, Ill.

19622

Fayetteville, N.C.

14995

Los Angeles

18325

Keller, Texas

15369

Poughkeepsie, N.Y.

4

Foster City, Calif.

6580

Terrell, Texas

4931

Camarillo, Calif.

8639

Sumter, S.C.

18567

Russellville, Ark.

3303

Greenbelt, Md.

15966

Saint John, Ind.

19382

Littleton, Colo.

7460

Marion, Ind.

14347

Genoa, Ohio

3522

Hudson, N.Y.

886

Tolleson, Ariz.

16203

Plain, Wis.

16191

Houston

9446

Baltimore

760

Harrisburg, Pa.

10368

Shawnee, Kan.

18958

Allentown, Pa.

15386

Mesquite, Texas

19841

Pittsburgh

5259

Hamilton, Ohio

18168

Moon Twp, Pa.

13665

West Liberty, Ohio

12882

Fort Wayne, Ind.

8039

Columbus, Ohio

19195

Madison, Wis.

4238


Imagery source: Google

Main Street(s), U.S.A.

This last set of images is not particularly revelatory about political geography in America today. But we think it’s fun.

When you sample 10,000 addresses from neighborhoods all across the country, you get a lot of homes on Main Street. One hundred and thirteen of them, to be exact, across 30 states, from rural Minnesota to the heart of El Segundo, Calif. Here are all of our Main Street scenes in their cumulative glory. There are about twice as many Trump neighborhoods as Biden ones in this group.



Oak Hill, Ohio

18434

Mocksville, N.C.

10359

Lindsborg, Kan.

16855

Phelps, N.Y.

18853

Schoharie, N.Y.

18546

Bay Shore, N.Y.

17265

Downers Grove, Ill.

14515

Whitehall, Pa.

6418

Cathlamet, Wash.

11391

Brewer, Me.

7018

Frisco, Colo.

2256

Wheaton, Ill.

3239

Madrid, Iowa

107

Cedar City, Utah

13945

Damiansville, Ill.

19414

Napa, Calif.

6407

Cadott, Wis.

7147

Castile, N.Y.

8563

Colo, Iowa

7148

Pierz, Minn.

2791

Tremont, Pa.

17666

Alden, N.Y.

18229

Woodsfield, Ohio

7382

Chillicothe, Ohio

5918

Madison, Ohio

69

Bloomsburg, Pa.

16801

Keedysville, Md.

10239

Bozeman, Mont.

15377

Chatfield, Minn.

7940

Buffalo

2374

Brockton, Mass.

3630

Portage, Ohio

13901

Midland, Ohio

9283

Caledonia, Miss.

2076

Idabel, Okla.

17809

Whitehall, Pa.

4632

Searsmont, Me.

3677

Cazenovia, Wis.

7122

Dickson City, Pa.

8212

Clintonville, Wis.

12810

Corona, N.M.

1095

Arcanum, Ohio

19857

Panama, N.Y.

5110

Callery, Pa.

12174

Waterboro, Me.

1500

Bluffton, Ohio

15941

Bowler, Wis.

9280

Otho, Iowa

3648

Woonsocket, S.D.

6893

Ninety Six, S.C.

14445

Gatesville, Texas

11530

Oshkosh, Wis.

9279

Cornwall, N.Y.

16850

Lincolnville, Me.

2304

Kaysville, Utah

4113

Mendham, N.J.

19729

Brownsville, Pa.

9234

Marlborough, Mass.

6223

Amanda, Ohio

13234

Fremont, Neb.

11542

Burlington, N.C.

5301

Sparta, Ill.

17378

Roswell, N.M.

6871

Shelby, Ohio

2730

Hopkins, Minn.

445

Dunkirk, N.Y.

18472

Darlington, Md.

15083

Concord, Vt.

7432

Clarkdale, Ariz.

1666

Middleton, Tenn.

3363

Napa, Calif.

11288

Chillicothe, Ohio

11911

Circleville, Ohio

7339

MacKinac Island, Mich.

4321

Newark, Ohio

18402

Lawndale, N.C.

11206

Oakfield, N.Y.

12836

El Segundo, Calif.

11707

Kings Park, N.Y.

14950

Blackstone, Mass.

9097

Accident, Md.

9432

Greenwich, Ohio

9541

Hohenwald, Tenn.

5775

Athens, Pa.

9517

Bristol, Vt.

10261

Ney, Ohio

4438

Hempstead, N.Y.

2666

Grantsville, Utah

8528

Tustin, Calif.

756

Creston, Ohio

3744

Lockport, N.Y.

7212

Monongahela, Pa.

672

Houston

5184

Thompsontown, Pa.

12531

Collinsville, Ill.

2820

Dundee, Mich.

17201

Gurdon, Ark.

8026

Batavia, N.Y.

16870

Norwood, N.Y.

15220

Ferndale, Calif.

2677

Royersford, Pa.

15230

Schuylkill Haven, Pa.

6703

Shamrock, Texas

12662

North Andover, Mass.

2819

La Crosse, Wis.

3416

Whitesville, N.Y.

1647

New Holland, Pa.

298

Highland, Kan.

11727

Fountain Inn, S.C.

10100

Ilion, N.Y.

2978

Voorheesville, N.Y.

9501

Whiting, Vt.

2012

Kingston, Mass.

13398


Imagery source: Google


Source: Elections - nytimes.com


Tagcloud:

Biden's infrastructure plan should cover childcare and home care. Here's why | Jamaal Bowman

Katie Hill: Matt Gaetz backed me but he must quit if nude-photo reports are true