Nov 09: The Cost and Value of Double Checking and Organization

D. The Cost and Value Of Double Checking And Organization

There are no standard methods for counting specified in Connecticut’s audit procedures. Double checking is required by Secretary of the State’s procedures, but frequently does not occur. Towns count in teams of 1, 2, 3 or 4.  Some are very organized in counting; others are organized in counting but not in totaling. We explored the costs of counting and the costs vs. value of various methods.

In each of thirty-two (32) municipalities, we timed the counting, and noted the number of supervisors, the number of ballots counted, the votes counted, and the counting differences[1].  To develop a cost per count we assumed a supervisor is paid $30.00 per hour and a counting official is paid $10.00 per hour[2].

We used cost per count (vote counts and ballot counts totaled) rather than cost per ballot because the number of races counted varied, and the number of candidates in each race varied significantly from district to district.

Based on these calculations, the cost per count, and average counting difference rate for the thirty-two (32) municipalities was:

For All Thirty-Two Municipalities Fully Observed
Average Cost Per Count (Estimated cost divided by the Number of ballots plus votes for all candidates counted for the 32 municipalities)	Average Difference Rate (Average of all differences in races and ballots per 1000 counts for  the 32 municipalities per 1000 counts)
$0.127	2.9 differences per 1000

Table 7: Average costs and differences for thirty-two districts.

We also looked at three groups of municipalities by counting methods and observation details:

1. Those that used teams of four counters, double checked, and where observers had no concerns with the organization of the counting.
2. Those that double checked, and where observers had no concerns with the organization of the counting with counting teams of less than four.
3. All the others.

	Number of Municipalities	Average cost per count	Average difference rate per 1000
1. Teams of four, double checking, organized	6	$0.159	0.5[3]
2. Municipalities double checking, organized, teams less than four	5	$0.124	2.9
3.All other municipalities	21	$0.118	3.6

Table 8: The relationship between counting methods, cost, and accuracy.

We conclude, perhaps not surprisingly, that organization and double checking pay off in higher accuracy for a small incremental cost.

We also note the possibility of a significant increase in accuracy using teams of four counters at a small increase in cost.

These calculations also point to the possibility of savings possible from counting accurately in the first place, if all towns had followed the required procedure to recount in each case that a difference is found.  It would also reduce the cost of subsequent, unnecessary investigation.

---

[1] Differences in each municipality were the difference in ballot counts and race counts considering questionable ballots.  The “difference rate” is the differences in manual and machine counts per 1000 counts in the municipality divided by the total number of votes and ballots counted in the municipality.

[2] We do not know the average hourly rate for registrars and counting officials in Connecticut.  These figures may be close to the average (within a dollar or two per hour) based on anecdotal information.  The purpose here is to provide a simple to understand comparison, rather than complete precision.

[3] Of the six towns which were well organized, double checked, and with teams of four — five of the six reported zero (0) differences, the sixth town reported a difference count of twenty (20).  In general, if methods were improved over time it is likely that greater efficiencies and greater levels of accuracy could be obtained with teams of four.