Using Election Technology to Make Better Decisions: The Case of Precinct Wait Times

Transcription

1 Using Election Technology to Make Better Decisions: The Case of Precinct Wait Times Charles Stewart III MIT NCSL Conference on The Future of Elections: Technology Policy and Funding June 15, 2017

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25 And by the way, we have to fix that

26

27 Avg. minutes waiting 50 FL 40 DC Nationwide avg.: 30 MD 13 min. 20 SC VA MI LA GA OK UT SC DC IN OK GA MD 10 0 VT VT

28 Wait times by mode of voting

29 Wait times by mode of voting Early voting: 11 min.

30 Wait times by mode of voting Early voting: 11 min. Election-Day voting: 14 min.

31 Wait times by race

32 Wait times by race Whites: 11 min.

33 Wait times by race Whites: 11 min. Blacks: 20 min.

34 Wait times by residential density Density level Type of location Wait Lowest quartile Very rural 6 min. Second quartile Rural/small town 10 min. Third quartile Suburbs 14 min. Fourth quartile Cities 19 min.

35 Wait times by residential density Density level Type of location Wait Lowest quartile Very rural 6 min. Second quartile Rural/small town 10 min. Third quartile Suburbs 14 min. Fourth quartile Cities 19 min.

36 Wait times by residential density Density level Type of location Wait Lowest quartile Very rural 6 min. Second quartile Rural/small town 10 min. Third quartile Suburbs 14 min. Fourth quartile Cities 19 min.

37 Wait times by residential density Density level Type of location Wait Lowest quartile Very rural 6 min. Second quartile Rural/small town 10 min. Third quartile Suburbs 14 min. Fourth quartile Cities 19 min.

38 Wait times by residential density Density level Type of location Wait Lowest quartile Very rural 6 min. Second quartile Rural/small town 10 min. Third quartile Suburbs 14 min. Fourth quartile Cities 19 min.

39 Avg. minutes waiting FL 40 DC Nationwide avg.: 30 MD 13 min. 20 SC VA MI LA GA OK UT SC DC IN OK GA MD 10 0 VT VT

40 Avg. minutes waiting FL 40 DC Nationwide avg.: 30 MD Nationwide avg.: 13 min. SC VA 11 min. UT 20 MI LA GA OK SC DC IN OK GA MD 10 VT VT 0

41

42

43 PCEA Report, p. 39

44 The Planning Cycle of Elections Plan Election Implementation Assess

45 The Planning Cycle of Elections Plan Election Implementation Assess

46 The Planning Cycle of Elections Plan Election Implementation Assess

47 The Planning Cycle of Elections Plan Election Implementation Assess

48 Planning Plan Election Implementation Assess

49 Planning What you need to know, in order to implement the most basic model in queuing theory (the M/M/c model)

50 Planning What you need to know, in order to implement the most basic model in queuing theory (the M/M/c model) Arrival rate of voters

51 Planning What you need to know, in order to implement the most basic model in queuing theory (the M/M/c model) Arrival rate of voters Number of service locations (poll books, scanners, voting machines, etc.)

52 Planning What you need to know, in order to implement the most basic model in queuing theory (the M/M/c model) Arrival rate of voters Number of service locations (poll books, scanners, voting machines, etc.) Service times (how long it takes to check in, scan a ballot, cast a ballot, etc.)

53 Planning web.mit.edu/vtp/

54 Planning web.mit.edu/vtp/

55 Planning web.mit.edu/vtp/

56 Planning web.mit.edu/vtp/?

57

58

59 Assessing: Measuring Wait Times Post-Election Plan Election Implementation Assess

60 Little s Law (Long term average) Length of queue (Long term) Average wait time L = λw (Long term) Arrival rate

61

62

63 Little s Law (Long term average) Length of queue (Long term) Average wait time L = λw (Long term) Arrival rate

64 Little s Law Average queue length Average wait time W = L λ Average arrival rate

65 Little s Law Average queue length Average wait time W = L λ In-person votes for the day hours the polls are open Average arrival rate

66 Little s Law This needs to be measured Average queue length Average wait time W = L λ Average arrival rate

67

68

69

70

71 Normalized check ins Early6am 8am 10am Noon 2pm 4pm 6pm 8pm Hour Data Average by hour

72 2 Normalized in line Early6am 8am 10am Noon 2pm 4pm 6pm 8pm Hour Data Average by hour

73 3 Normalized arrivals Early6am 8am 10am Noon 2pm 4pm 6pm 8pm Hour Data Average by hour

74 Final thoughts Polling places are the hardest part of the election experience to optimize Improving the polling place experience requires election officials to find ways to extract management data from polling places Extracting the data requires rethinking some management practices and leveraging technology Improvements in wait times demonstrates that small changes to management practices and managing-by-data can work

75 Final thoughts Polling places are the hardest part of the election experience to optimize Improving the polling place experience requires election officials to find ways to extract management data from polling places Extracting the data requires rethinking some management practices and leveraging technology Improvements in wait times demonstrates that small changes to management practices and managing-by-data can work

76 Final thoughts Polling places are the hardest part of the election experience to optimize Improving the polling place experience requires election officials to find ways to extract management data from polling places Extracting the data requires rethinking some management practices and leveraging technology Improvements in wait times demonstrates that small changes to management practices and managing-by-data can work

77 Final thoughts Polling places are the hardest part of the election experience to optimize Improving the polling place experience requires election officials to find ways to extract management data from polling places Extracting the data requires rethinking some management practices and leveraging technology Improvements in wait times demonstrates that small changes to management practices and managing-by-data can work

78 Final thoughts Polling places are the hardest part of the election experience to optimize Improving the polling place experience requires election officials to find ways to extract management data from polling places Extracting the data requires rethinking some management practices and leveraging technology Improvements in wait times demonstrates that small changes to management practices and managing-by-data can work

79 Charles Stewart Election Updates electionupdates.caltech.edu