# Comments on Group II, Improving Registration Procedures (PowerPoint presentation)

Group II:
Improving Voter Registration
Procedures
(Allison Horst, Stephen Larson, Alex Tahk)
Do voters in states with statewide
registration systems report fewer
problems voting than voters in states
without them?
Basic logic
• Find measure of “voter registration problems”
and see whether voters in states with statewide
voter registration systems report fewer problems
– Define “statewide voter registration system”
• Control for confounding effects, because states
with statewide registration systems might also
have other characteristics that minimize
problems
• Individual or aggregate level analysis
What the group did
• Data
– Measure of statewide registration systems:
NCSL report
– Registration problems: CPS Voting and
Registration Supplement
– Demographic data: Census
Table 1: Voters in states with
centralized systems vs. voters in
states without
Table 2: State-level multivariate
analysis
• Dependent variable: pct. of registered voters with
reported registration problems
• Independent variable: laundry list of
demographics, state-level characteristics, etc.
– Problem: not much works, and there’s collinearity
– Solution: drop most variables
• Table 2
Individual results
• General
– Overall strategy a very good one: simple test
to a more complex test
– The aggregate analysis is redundant and
unnecessary
– It’s good to have a summary table of variables
1.0%
1.0%
Centralized Voter Registration
System
Pct. with problems
N
No
Yes
1.0
1.0
26793
23729
Question: what is the sample? (All voters? All registered voters?)
What was the main reason (you/name) did not
vote?
Main thing: What is the dependent
Variable measuring?
•Denominator (all eligible voters or
all non-voters?
•Numerator (only reg. problems or
all polling place problems?)
-9 No Response
-3 Refused
-2 Don't know
-1 Not in universe
1 Illness or disability (own or family's)
2 Out of town or away from home
3 Forgot to vote (or send in absentee ballot)
4 Not interested, felt my vote wouldn't make a
difference
5 Too busy, conflicting work or school schedule
6 Transportation problems
7 Didn't like candidates or campaign issues
8 Registration problems
10 Inconvenient polling place or hours or lines
too long
11 Other
Alternate regressions
Numerator:
Only registration problems
Registration problems or lines
Eligible
Non-voters
Eligible
Nonvoters
(days)
0.000053
(0.00048)
0.0076
(0.0046)
0.000074
(0.000057)
0.0012
(0.0006)
Centralized voter
reg. (0,1)
0.00039
(0.00041)
-0.0022
(0.0041)
0.00099
(0.00048)
0.0014
(0.0047)
Same-day
registration (0,1)
-0.0049
(0.0012)
-0.042
(0.012)
-0.0060
(0.0014)
-0.051
(0.015)
Education (years
[if done right])
0.00029
(0.00012)
0.0059
(0.0014)
0.00029
(0.00014)
0.0077
(0.0016)
Log of income (\$)
-0.00082
(0.00042)
0.0038
(0.0038)
-0.0012
(0.0005)
0.0044
(0.0045)
How many years
at curr. Res. (yrs.)
-0.0043
(0.0002)
-0.032
(0.002)
-0.0046
(0.0003)
-0.033
(0.002)
Constant
0.024
(0.005)
-0.062
(0.055)
0.032
(0.006)
-0.12
(0.06)
N
69,837
6,599
69,837
6,599
R2
.006
.05
.005
.04
Mean of dep. Var.
.0038
.068
.0053
.095
Sample: