Missing candidates in precinct returns

[kuriwaki]

From our improved classification in #319 I found 14 counties where the issue is: there is no corresponding R/D/L entry in the Baltz et al. data. I took a quick look and all of them look like counties that would be releasable if we just fixed that error in the precinct returns data.

Here are the candidates and the code to reproduce this:

state	county_name	color2_h	office	district	party_detailed	candidate_c	votes_c
FLORIDA	DUVAL	no entry in Baltz	STATE HOUSE	014	DEMOCRAT	ANGIE NIXON	62151
GEORGIA	DOUGLAS	0 difference	US HOUSE	013	LIBERTARIAN	MARTIN COWEN	21
IDAHO	BONNER	not collected	STATE HOUSE	01A	DEMOCRAT	GAIL BOLIN	7860
IDAHO	BONNER	not collected	STATE HOUSE	01A	REPUBLICAN	HEATHER SCOTT	14911
IDAHO	BONNER	not collected	STATE HOUSE	01B	DEMOCRAT	STEPHEN F HOWLETT	6859
IDAHO	BONNER	not collected	STATE HOUSE	01B	REPUBLICAN	SAGE G DIXON	15422
IDAHO	BONNER	not collected	STATE HOUSE	07A	REPUBLICAN	PRICILLA GIDDINGS	2876
IDAHO	BONNER	not collected	STATE HOUSE	07B	REPUBLICAN	CHARLIE SHEPHERD	2801
NEVADA	LINCOLN	no entry in Baltz	STATE HOUSE	036	REPUBLICAN	GREGORY T HAFEN II	701
NEVADA	NYE	no entry in Baltz	STATE HOUSE	036	REPUBLICAN	GREGORY T HAFEN II	19852
NEW JERSEY	CAMDEN	red	US HOUSE	001	DEMOCRAT	PROGRESSIVE	1
NEW JERSEY	ESSEX	any < 1% mismatch	US PRESIDENT	FEDERAL	REPUBLICAN	KAREN BASS	1
NEW JERSEY	ESSEX	any < 1% mismatch	US HOUSE	010	DEMOCRAT	DONALD PAYNE JR	1
NEW JERSEY	HUDSON	any < 1% mismatch	US HOUSE	009	LIBERTARIAN	JOHN MIRRIONE	1
OHIO	HANCOCK	red	STATE HOUSE	083	DEMOCRAT	MARY E HARSHFIELD	1317
TENNESSEE	LOUDON	not collected	US PRESIDENT	FEDERAL	LIBERTARIAN	JO JORGENSEN	270
TENNESSEE	PICKETT	not collected	US PRESIDENT	FEDERAL	LIBERTARIAN	JO JORGENSEN	11
TENNESSEE	SEVIER	not collected	US PRESIDENT	FEDERAL	LIBERTARIAN	JO JORGENSEN	402
TENNESSEE	WILLIAMSON	any < 1% mismatch	US PRESIDENT	FEDERAL	LIBERTARIAN	JO JORGENSEN	1494

library(tidyverse)
library(readxl)

# counties to check
county_check <- read_excel("combined/compare.xlsx", sheet = 2) |> 
  filter(color2_c == "no entry in Baltz") |> 
  select(state, county_name, matches("color2"))

# which candidates are the problem
read_excel("combined/compare.xlsx", sheet = "by-cand-coalesced") |> 
  inner_join(county_check, by = c("state", "county_name")) |> 
  filter(party_detailed %in% c("DEMOCRAT", "REPUBLICAN", "LIBERTARIAN"), 
         state != "DISTRICT OF COLUMBIA") |> 
  filter(is.na(votes_v)) |> 
  select(state, county_name, 
         color2_h, 
         office:party_detailed,
         candidate_c, votes_c, votes_v)

• [x] In the NJ and OH states, the candidates should be listed as write-ins on the Reece data (medsl).

• [x] But in the others, it seems like the Baltz data is simply missing the candidate from the record (returns/by-county) (I quickly took a few county websites). We can create a mini csv as we did for #39 and merge it in.

• [x] A similar thing should be done for the 4 Maryland counties in #324. For those, the returns data should be changed to NA and writein == 1.

• [x] #83 (see comment there to change baltz)

• [x] Michigan: Eaton #276, Mason #277, Washtenaw #99

• [x] California: Shasta: no issue filed by Aleksandra; MEDSL cvr missing a state house race

• [x] California, Lake: MEDSLcvr missing a state house race

• [x] California, Orange: This looks close. No issue filed by Jeff. MEDSL cvr missing party metadata, but also seems slightly more off than Harvard. Shiro will look in to this.

If we can release these, we'd get two new Republican states.. TN/ID.

[mreece13]

I resolved the write-in candidates in NJ and OH. Pending the next build.

[kuriwaki]

"I resolved the write-in candidates in NJ and OH. Pending the next build."

These don't seem to be reflected yet in the medsl on Dropbox right now, but if it's on its way, no worries.

The NJ case is annoying to deal with .. in this example, we basically want to turn the last two rows into WRITEINs. You can see how the second one is missing his middle name.

> open_dataset("medsl") |> filter(state == "NEW JERSEY", county_name == "ESSEX", office == "US HOUSE", district == "010") |> count(office, candidate, party_detailed) |> collect() |> filter(str_detect(candidate, "PAYNE"))
# A tibble: 3 × 4
  office   candidate         party_detailed      n
  <chr>    <chr>             <chr>           <int>
1 US HOUSE DONALD M PAYNE JR DEMOCRAT       146100
2 US HOUSE DONALD PAYNE JR   DEMOCRAT            1
3 US HOUSE DONALD PAYNE JR   WRITEIN             1

I linked Kenosha and Maryland to this issue too

[mreece13]

Lake and Shasta resolved.

[mreece13]

Some notes on the missing candidates:

• [x] In Duval, Florida, the missing candidate from State House 014 appears in the CVR but not in the official election results because the general election was not held (candidate ran unopposed). I propose we remove this candidate from the CVR since voters would not have actually selected them (I think).
• [x] in Douglas, GA, Martin Cowen the Libertarian candidate was a qualified write-in. I will mark them as such.
• [x] Idaho is pending build
• [x] Nevada is pending build
• [ ] Essex, New Jersey is pending build
• [x] Mary E. Harshfield in Hancock, Ohio is a qualified writein. Pending build
• [x] Tennessee is pending build