CHAIRPERSON QUESENBERY: This first panel is looking at issues around the issues of design of ballots and usability as part of the voting experience. I would like to welcome our first speaker, Kimball Brace. Introduce yourself and background for the people who may only be reading what you said later.
MR. BRACE: I many times I tried to figure out who I am or where. I appreciate the opportunity to come and testify. I must say in full discloser my name is Kimball Brace. I am president of a company called Election Data Services out of Washington D.C. or as I always say the nearest airport.
I was up in Rhode Island yesterday, and I have to fly back to Rhode Island this afternoon. But it is a quick time to come in and see you.
I was about to say in full disclosure I must say that two of the four members on the front panel are current clients of mine, and one of the other ones was -- worked against me, I guess, in his earlier days on that side.
But as many of you know, we are involved with data collection. We collect an awful lots of data. I have for a number of years. Election Data Services I started back in 1977 and began work actually in election administration before then when I was associate editor of Dick Smokin Election Reports. He got me involved in Election administration more than 30 years ago.
It's because of his interest imparted upon me I have been become a collector of a lot of information. Back in 1980, we ended up starting an effort to collect what kind of voting equipment was used in every county in the country.
We continued to collect that and maintain for every two years since 1980. It wasn't until the year 2000 that anybody cared about that. It was a labor of love and little pay, if anything, at that point in time. Then in that 2000 everybody was interested in what was going on.
We have -- Because of this data collection we have been intensely involved in looking at voting information and voting data, and we have become very much aware of the consequences of different types of voting equipment because we do maintain the only database that there is around there that allows us to bring together voting equipment information with electrician results.
We have maintained that we have a data set going back to 1948 of the county level election returns for the entire country.
We don't have this at the precinct level. But we have a county level data set that lets us take a look and work together with our election information data to see what kinds of patterns one might find.
We have been the source of much data that many of the academics have been utilizing over the past four years to analyze different voting machines and dropoff rates. Residual rates. That sort of thing.
As I said, this is just at a county level, and that's where most of the analysis has taken place. So we are involved with an effort on behalf of the EAC to see for the year 2004 what might be available at the precinct level from the various states.
So, we have been in the last two and a half weeks talking with -- trying to get hold of every single state in the nation to see what kind of data collection efforts they have underway and what might be available at a lower geographic level.
We have over the past 25 years done a lot of analysis on court cases particularly in the voting rights area. And at that level, we are equating election data with census data. That is the only level you can get a firm feel of the demographic characteristics of voting behavior and how that might have an inference on what's going on.
And so it's my belief that you need to have a precinct level data where one can analyze more closely the patterns that one would see at the county level.
As I said, we have done a bunch of analysis at that county data, and we were called upon in -- in 2000 just immediately after Florida to take a look at the data coming out of the State of Florida to see what was happening in terms of ballot design. It became clear to me very, very early that ballot design was a critical factor in a lot of how people perceive and understand the voting process.
We could see in the data coming out of Florida that, in fact, the impact of the butterfly ballot we now all know and not love, I guess, coming out of Palm Beach County. But we could see it in the data because when you looked at the people residual rates, I call them dropoff rates, you could see an abnormally high degree of dropoff in that particular jurisdiction.
But what was more important to take a look at is the difference of what makes up dropoff. For purposes of definitions, I'll provide. I call dropoff the difference of how many people went to the polls and voted compared to the votes cast for which ever office you're interested in analyzing.
We know from history that, in fact, not everyone that goes to the polls votes for the highest office. In my May 5th testimony to the EAC, we provided a chart to the EAC that looked back to 1948 to the dropoff rates for President and for highest office in non-Presidential elections. One sees back in time generally you see one and a half to maybe two percent of the people that went to the polls did not cast votes for the highest office.
That percentage increases when you go to a non-Presidential election. It goes to about three and a half to four percent of the people that went to the polls didn't cast a vote for the highest office.
Why didn't they cast at vote for the highest office? That's part of the investigation. Generally, there's two components of a dropoff. One is over-votes and the other is called under-votes.
Generally over-votes occur when people vote for more than -- more candidates than they are allowed for that particular office. If you have a vote for three, you cast four votes or something like that.
Under-votes is where you don't have any votes recorded for that office. What we have seen over the years is when you can get hold of that over-vote and under-vote, you generally have about 90 percent of that dropoff is accounted for by under-vote. Only about ten percent is accounted for by over-vote.
So, that's the normal pattern that one sees when you look at election results in analyzing that residual vote rate. What we were finding in Florida was the exact opposite. We found 90 percent of the votes cast were over-votes, not under-votes. And that led us to take a look at the ballot design.
And that's certainly where one starts by analyzing and being able to look at what are the numbers that come out of the system on election night or shortly thereafter.
So, that's where we began, and we certainly encourage anyone looking at the process to begin at that process. But that let us see where there might be problems. And we always encourage jurisdictions particularly counties and local governments to look at those kinds of data at the local level. So, that's why we encourage them to look at the precinct level and be able to figure out where are the abnormalities in your jurisdiction. And that is part of the effort that we will be undertaking for the EAC behalf.
We'll then be able to couple that with the voting equipment information. That is the other important component to see what kind of patterns that might be determined. With that, you have an overview of what we're looking at.
CHAIRPERSON QUESENBERY: Thank you. Does anyone on the panel have a question? Are you actually looking at not just the equipment and numbers but the actual ballot design?
MR. BRACE: We are. Part of our data collection is attempts to get sample ballots. Generally from each jurisdiction for each ballot format. So that one could down at the precinct level see where ballot format might be an issue.
So, that's one of the desires we have in terms of collecting that kind of information. That is a monumental collection effort. It greatly depends on the skills of our vice chairman here in persuading Congress in terms of money for this kind of research effort.
If it's there, we think there is a way we can collect all of that.
CHAIRPERSON QUESENBERY: On the phone, Jim Elekes.
MR. ELEKES: With the data you are collecting, is there any data that has been collected relative to ease of ballots for persons who may have physical disabilities, or may have problems marking the ballots, or sensory disabilities, and may require an alternative means of marking the ballot?
MR. BRACE: Good question. I don't know of any independent research, and we're going to be collecting aggregate data without looking at individuals and talking to individuals. That's more of an individual investigation, I believe, to try to determine that kind of thing.
So, unfortunately we wouldn't have that kind of data. I think it's a good piece of information to try to gather if you can. The difficulty would be in trying to gather that.
You are talking about interviewing poll workers basically to determine what they saw on election day or collecting information from individual voters that might come forward to say that they have difficulty. That's your only two data sources.
MR. ELEKES: Thank you.
CHAIRPERSON QUESENBERY: Thank you very much, Mr. Brace.