Friday, August 28, 2015

Voter Registration and Turnout Figures



I found Gardner Selby’s Politifact Texas article in Sunday’s Austin American-Statesman interesting and decided to do some of my own checking. The article assessed whether Travis County Tax Assessor-Collector Bruce Elfant’s statement that Texas ranked near the bottom in registered voters was accurate.

There are various ways to compute the percentage of registered voters. Most commonly, the percentage of registered voters is calculated by dividing the number of registered voters by the voting age population (VAP) –residents who are 18 years-old and older. For example, if the number of registered voters is 14,025,441, and the voting age population is 18,915,297, then the percentage of registered voters is 74.15 percent.  These are the figures used by the Texas Secretary of State to calculate percentage of registered voters.






This is problematic as some of the voting age population may not be qualified to vote because they are not citizens of the United States or are incarcerated felons and unable to vote in most states. A better indication of the percentage of registered voters would be calculated using the number of registered voters divided by the voting eligible population, which would exclude non-citizens, felons unable to vote, and paroled felons who are not able to vote. Associate Professor Michael McDonald, University of Florida, provides these figures for each state at his United States Elections Project Web site. Here are his calculations for Texas’ voter eligible population:







The Census Bureau’s data come from the Current Population Survey (CPS) and are described below:
The data in this report are based on responses to the November CPS Voting and Registration Supplements, which survey the civilian noninstitutionalized population in the United States. Voting estimates from the CPS and other sample surveys have historically differed from those based on administrative data, such as the official results reported by each state and disseminated collectively by the Clerk of the U.S. House of Representatives and the Federal Election Commission (FEC). In general, voting rates from the sample surveys such as the CPS are higher than official results (Bauman and Julian, 2010; De Bell, et al., 2015). Potential explanations for these differences include misreporting, problems with memory or knowledge of others’ behavior, and methodological issues related to question wording, method of survey administration, and nonresponse. Despite these issues, the Census Bureau’s November supplement to the CPS remains the most comprehensive data source available for examining the social and demographic composition of the electorate in federal elections, particularly when examining broad historical trends for subpopulations.
Here are the Census Bureau’s data for Texas:


Now let’s compare the data. The VAP is 18,915,297 according to the Texas Secretary of State, 19,847,831 according to the Elections Project, and 19,354,000 according to the Census Bureau’s CPS. The total number of registered voters is 14,025,441 according to the Texas Secretary of State, and 9,946,000 according to the Census Bureau’s CPS. The voting eligible population in Texas is 16,679,393 according to the Elections Project, and 16,844,000 according to the Census Bureau’s CPS. The greatest discrepancy is in the number of voters. The Texas Secretary of State calculated that 4,727,408 Texans voted whereas the Census Bureau’s CPS found that 5,836,000 Texans claimed to have voted in the 2014 election.

What should one think about this information? I would argue that the data on number of registered voters for an election would be more accurate from the state official responsible for voter registration. The data from the Census Bureau could be inflated. Isn’t one expected to be registered? If one isn’t registered, is that a sign that the person is not accepting his or her responsibility in a democracy? I would make the same argument for number of voters in an election and for the same reason. Concerning VAP and VEP, I’m inclined to take the Election Project’s data as more accurate than either the Texas Secretary of State’s data or the Census Bureau’s data.

So, how many Texans are registered to vote? I would say that 14,025,441 is the correct number of registered voters. What percentage of eligible Texans are registered to vote? Divide the number of registered voters by the VEP, which results in 84.01 percent. What percentage of Texans voted in 2014? The number of voters was 4,727,208. The percentage of VEP that voted was 28.34 percent.
That's nothing to be proud of!

The full table by the Census Bureau is below:


Thursday, August 27, 2015

Campaign Spending: At-large vs. Single-member District Electoral Systems



Serendipitously, I recently came across a capstone report by Jason Malinowski entitled “Campaign Spending in City Council Elections: A Comparison of At-Large and District Contests.”  In it, Malinowski examines the relationship between campaign spending in City Council elections and the type of electoral system—at-large, mixed, or single-member district—employed by the city. One of the arguments for single-member district electoral systems is that candidates do not have to raise and spend as much money as they do in a city with an at-large electoral system. This increases the opportunities for candidates to run for city council seats. But is it true that winning a city council seat in a single-member district electoral system is less expensive than competing for a city council seat in an at-large electoral system?

Malinowski notes that there are few studies comparing campaign spending and type of electoral system. He provides the table below comparing three empirical studies:

Malinowski compares the mean spending by winning candidates in 19 large cities that employ different methods of electing council candidates. The graphic displays his results:

In Austin, according to Malinowski’s data, a winning candidate spent a mean of $161,000 in the at-large system. In 2014, the mean spending by a winning candidate was $119,153. The difference is $41,847, a 26 percent decrease in mean campaign spending. The difference is less than what Malinowski found, but the effect is noteworthy. In his conclusion, Malinowski states:

. . . [T]his study finds a strong relationship between city council electoral system and the amount of campaign funds spent by winning candidates. Specifically, it finds that candidates for at-large contests expend a significantly greater amount of funds than district candidates. This observation appears consistently across a set of description statistics comparisons and regression analyses. The difference in spending is estimated to be approximately $76,000. These findings suggest that cities with at-large systems, who want to increase political participation and decrease the influence of moneyed special interests, may wish to evaluate a change to a district or mixed electoral system.

Sunday, August 23, 2015

Voter Turnout and Percentage of the Vote for Minority Candidates in the 2014 Austin Council Election



Zoltan Hajnal (America’s Uneven Democracy, pages 76-78) argues that higher voter turnout is conducive to ethnic minorities’ representation on city councils. He states that it is more advantageous for Hispanics and Asian Americans than for African Americans. The Austin City Council election in November 2014 provides an opportunity to test whether Hispanics and African Americans benefitted in districts where those minorities constituted either a substantial percentage of the population of the district or constituted a majority of the district. 

District 1, located in northeast Austin, is considered an opportunity district for African Americans. Although African Americans do not constitute a majority of the voting age population, their percentage of the population is considered large enough to allow an African American candidate an equal opportunity to win the election in the district. According to the 2010 census, African Americans constituted 29.68 percent of the district’s voting age population (VAP)—residents who are 18 years of age or older; Hispanics were 37.06 percent of the VAP, and Asian Americans were 3.59 percent of the VAP. Because Hispanics are more likely than African Americans to be non-citizens, the district is an African American opportunity district.

Among the nine candidates in District 1, five were Anglos, and four were African Americans. Table 1 depicts the combined votes for the five Anglo candidates and for the four African American candidates, by precinct.

Table 1:  Votes for Anglo Candidates, Votes for African American Candidates, Voter Turnout in District 1, by Precinct
Precinct
Votes for Black Candidates
Votes for Anglo Candidates
% Vote for Black Candidates
% Vote for Anglo Candidates
Voter
Turnout
101
141
56
71.6%
28.4%
37.81%
102
110
121
47.6%
52.4%
22.67%
103
320
244
56.7%
43.3%
34.69%
104
232
189
55.1%
44.9%
34.20%
108
264
154
63.2%
36.8%
28.03%
117
426
148
74.2%
25.8%
21.38%
118
107
48
69.0%
31.0%
18.26%
121
200
31
86.6%
13.4%
23.69%
122
837
122
87.3%
12.7%
31.73%
124
1,191
293
80.3%
19.7%
27.68%
126
1,313
340
79.4%
20.6%
31.33%
129
652
142
82.1%
17.9%
26.78%
130
472
198
70.4%
29.6%
32.99%
132
760
177
81.1%
18.9%
33.43%
133
379
142
72.7%
27.3%
32.44%
139
51
30
63.0%
37.0%
19.42%
141
54
37
59.3%
40.7%
19.61%
151
632
284
69.0%
31.0%
36.60%
153
594
388
60.5%
39.5%
27.77%
154
481
238
66.9%
33.1%
30.75%
156
267
97
73.4%
26.6%
34.50%
203
29
24
54.7%
45.3%
24.54%
227
16
14
53.3%
46.7%
13.04%
325
4
8
33.3%
66.7%
23.53%
444
24
9
72.7%
27.3%
12.27%

I calculated the relationship between percentage of the vote for African American candidates and voter turnout for each precinct, which resulted in a Pearson product moment correlation of 0.25, which is a weak relationship. The scatterplot of the African American vote and turnout is in figure 1. The linear trend line reflects the weak, but positive relationship.

Figure 1: Graph of Relationship between Percentage of the Vote for African American candidates and Voter Turnout

For candidates, perhaps the campaign’s concentration is on those precincts that provide a larger percentage of the district’s total vote. This is reflected in the precinct’s contribution to the total vote in the district. Table 2 displays the total votes cast in each precinct and the percentage of the total vote contributed by each precinct.

Table 2: Total Votes and Percentage Contribution, by Precinct
Precinct
Total Votes
% Vote
Black Candidates
% Contribution
101
197
71.6%
1.5%
102
231
47.6%
1.8%
103
564
56.7%
4.3%
104
421
55.1%
3.2%
108
418
63.2%
3.2%
117
574
74.2%
4.4%
118
155
69.0%
1.2%
121
231
86.6%
1.8%
122
959
87.3%
7.3%
124
1484
80.3%
11.3%
126
1653
79.4%
12.6%
129
794
82.1%
6.1%
130
670
70.4%
5.1%
132
937
81.1%
7.2%
133
521
72.7%
4.0%
139
81
63.0%
0.6%
141
91
59.3%
0.7%
151
916
69.0%
7.0%
153
982
60.5%
7.5%
154
719
66.9%
5.5%
156
364
73.4%
2.8%
203
53
54.7%
0.4%
227
30
53.3%
0.2%
325
12
33.3%
0.1%
444
33
72.7%
0.3%

13090
73.0%
100.0%

The Pearson product moment correlation of percentage of the vote for African American candidates in District 1 and the percentage contribution of the precinct to the total vote is .53, a stronger relationship. The scatterplot also reflects the stronger relationship.

Figure 2: Graph of Relationship between a Precinct’s Contribution to the Total District Vote and Percentage of the Vote for African American Candidates


An analysis of the majority Hispanic districts will be a future post.