Voters' self-assessed comprehension of voting methods

By Warren D. Smith, June 2013.   (Skip to conclusion.)

Details

Score and approval voting data are from this French exit poll study (see pages 222 and 257-259) question wording "Le principe du vote par note vous semble-t-il clair?" and "Le principe du vote par approbation vous semble-t-il clair?"; 1256 respondents. The overall percentages of score and approval voters for whom the principle was "unclear" or "slightly unclear" were (9.4±0.8)% and (14.4±1.0)% respectively (including ±one standard error).

Instant runoff based on table 9 in this exit poll; 1633 responding voters in San Francisco (November 2005). IRV enjoyed these comprehensibility advantages versus score and approval voting:

1. San Francisco had decided to adopt IRV in March 2002, i.e. over 3 years earlier (then spent millions of dollars on voter-education campaigns about it), and most SF voters had already experienced IRV in the preceding (November 2004) election.
2. In contrast, in the French approval and score voting studies, these were only an exit poll study using pseudo-ballots, not official voting using real ballots, the warning time was under 1 year, and essentially all the voters presumably had never experienced either voting method before.

The San Francisco study also broke up the pollees by education level ("less than HS" up to "post-grad study"), age, ethnicity, district, first language (Chinese, English, Spanish, Other), income, gender, ideology ("very liberal" to "very conservative"), whether they had voted in the preceding 2004 IRV election, etc. The French study did not do those things although in principle one could deduce their political ideology from their votes.

The overall percentage of IRV voters who "did not entirely understand" or "did not understand at all" was (12.9±0.8)%. The worst self-assessed comprehension levels were exhibited by "less than HS" education level (23.8%) and 21.9% for Chinese as 1st language. The best were post-grad education (10.0%) and Spanish as 1st language (9.0%).

Lesson: Simplicity matters. This is quite large incomprehension rate, especially for the Chinese and low-education people – most people would regard their incomprehension rates as unacceptably large. Advocates of improved voting methods need to focus on simplicity. A good case can be made that even instant runoff voting is too complicated. And now also consider the challenges faced by such countries as Afghanistan (estimated 28% literacy rate in year 2000), Egypt (estimated 72% literacy rate in year 2010 but only 63% among women) and India (estimated 74% literacy rate in year 2011 but only 65% among women) instead of the USA and France which have claimed 99% literacy for decades. Score voting is simple enough to have been used highly successfully by children in kindergarten. How many would misunderstand, say, score voting employing negative numbers as allowed scores? What about Condorcet methods or the "majority judgment"?

*Plurality voting: It also is of interest to compare with plain plurality voting. How comprehensible/incomprehensible is that? Here's a January 2009 telephone poll of 1485 USA citizens aged≥18 asking about their experiences voting.

  Question 16c "I found my ballot to be difficult to understand"
    Disagree strongly................73
    Disagree somewhat................14
    Neither agree nor disagree........2
    Agree somewhat....................7
    Agree strongly....................3%

We warn the reader that all three of these polls were in different contexts, different places, different survey methodologies, and with slightly different wordings, so attempts to compare them directly at high accuracy may be dangerous. (We're doing that anyway since we don't have any choice.)

Conclusion

In order of decreasing self-assessed comprehensibility :

We admit that the fact that 12.9<14.4 contradicts the final ">" but keep in mind:

• The difference between 14.4 and 12.9 was within the margin of statistical error,
• The difference between 66.9 and 51.6 was not,
• The circumstances were biased (described above) to artificially favor the comprehensibility of IRV (and plurality), and the same pollster report shows that the first-time IRV voters did not exhibit incomprehension level 12.9%, but rather 14.5%.

What voters want

Time delays to get election results as one objective (and consequential) way to measure voting method "simplicity"

Which scales score voters prefer

Return to main page