Peter Kilkus: nearly two decades of hyper-local news reporting in Napa County
Peter Kilkus: nearly two decades of hyper-local news reporting in Napa County
Intro by Tim Carl and content by Peter Kilkus
NAPA COUNTY, Calif. — Napa Valley Features is pleased to welcome Peter Kilkus as a new contributor. An important figure in local journalism, Kilkus has been delivering community news in Napa County for nearly two decades. His journalism journey began in 2005, when he joined Shirl Katleba, the previous owner of the Lake Berryessa News, a publication he later purchased for $2,000 in 2009. Kilkus' extensive knowledge of Lake Berryessa, a haven for watersports and outdoor enthusiasts in Napa County and the broader region, is comprehensive.
Kilkus' deep ties to the region and academic credentials that include a master of science degree in physics from UC Santa Barbara lend a distinct perspective to his journalistic work. He is also the author of three books, including " Lake Berryessa Technical Manual: The Science." His commitment to the local community and in-depth understanding of regional intricacies make him an important addition to our Napa Valley Features team.
Kilkus' contributions as a journalist have been particularly significant during challenging times, such as when destructive wildfires ravaged Napa County. In 2017, the Los Angeles Times highlighted the important role he and his son, Evan, played in providing timely news and support to the Berryessa Highlands community during the wildfires. Their GoFundMe initiative to support fire victims and local volunteer firefighters exemplified their dedication and community unity at such times.
Another noteworthy aspect of their work is the eagerly awaited “Glory Hole” watch at Lake Berryessa, a rare natural occurrence. This event, which happens when surplus lake water cascades into the Glory Hole, creates a stunning display reminiscent of a watery black hole. Onlookers and photographers travel great distances to see the spectacle, which has become a rite of passage for many local families. This moment is often marked by the Kilkuses' timely updates. Evan's drone footage of the event recently captivated more than 60 million viewers globally, showcasing the awe-inspiring power and beauty of nature.
At Napa Valley Features, we are eager to share Kilkus’ voice. His first article, emphasizing the importance of local election participation, aims to statistically inspire our readers to vote. And with the recent heavy rains, we might soon receive updates on the current status of the Glory Hole, which was a little over 6 feet away from activating at last count.
A statistical look at election results
By Peter Kilkus
NAPA VALLEY, Calif. — Surveys and statistics may or may not predict election results, especially in races with relatively large statistically random populations of voters. That’s also true of Napa County supervisorial elections with smaller populations of voters.
Election seasons begin with public-opinion polls. It only takes 50.1% of any vote to win an election. If we want a 95% confidence level (the level used in most surveys) with a confidence interval of plus or minus 2% that a candidate will win with 52% of the vote (50% - 54%) in a 13,000-voter population (typical Napa County supervisor district), we only need to survey about 2,000 random voters. To get a 99% confidence level with a margin of error of 2%, we would need to interview about 3,000 voters.
As recent history has shown, pre-election public polls can be wrong for various reasons. The survey questions could be poorly worded, the interviewees might be unsure of their own opinions, or the interviewees could lie because they are contrarian and resent being considered a statistic. This voter-response bias has increased because the sheer number of pollsters has exploded over the last 20 years. This has created voter fatigue, tedium and less willingness to respond for privacy and social desirability reasons. Fortune Magazine addressed this issue in "Pollsters got it wrong in 2018, 2020, and 2022. Here’s why political polling is no more than statistical sophistry."
After the polls close there is usually an initial vote count made up of mail-in ballots that have been previously processed. In Napa County that vote count is released at 8:01 p.m., immediately after the polls close. Two additional releases later in the evening consist of live ballots voted at the Election Division beginning 29 days before the election and live ballots brought in from satellite vote centers that evening. Additional releases begin the Friday after the election and continue until 95% of the ballots have been counted. The final 5% are in the certified tally after the manual tally is completed and all damaged ballots have been duplicated and counted.
The difference between the initial vote count and the final vote count in an election is distinct from the difference between a final vote count in a close election and the results of a final vote recount in that election.
According to Fair Vote, which describes itself as a "nonpartisan organization working for better elections for all," "An analysis of statewide election recounts 2000-2023 shows that statewide recounts are rare. Out of the 6,929 statewide general elections between 2000 and 2023, there were 36 statewide recounts. In other words, there was one recount for every 192 statewide elections. Outcome reversals are even rarer. Recounts resulted in only three reversals, or one out of every 2,310 statewide elections. All three reversals occurred when the initial margin was less than 0.06% of all votes cast for the top two candidates. Recounts tend to shift only a small number of votes. Statewide recounts resulted in an average margin shift of 551 votes between the frontrunners, representing 0.03% of the vote. Recounts typically widen the gap between the top two candidates instead of decreasing it."
This is the main reason political organizations work to “get out the base,” especially in closely contested elections. The more advocates of a specific political view who go to the polls, as opposed to a random sample of voters, the more chance that view can have an influence in a close election. In this case, lower voter turnout favors the more active “get out the vote” proponents.
This is also why rural populations such as the Lake Berryessa region do not have as much influence in elections where they are a small part of a total electorate. In the 2010 District 3 election, there were about 2,500 rural voters (Lake Berryessa, Pope Valley, Angwin) matched against more than 10,000 urban voters (St. Helena, Calistoga and parts of Napa). If one candidate got 80% of the rural vote (2,000) the other candidate would only need to get 60% of the urban vote (6,000) plus the 20% of the rural vote (500) to win, 6,500 votes to 6,000 votes. Candidates only need to appeal to urban voters who may not understand the actual needs of their rural neighbors.
In 2010 I wrote a series of stories for Lake Barryessa News about the Napa County District 3 supervisor’s election contest between incumbent Diane Dillon of St. Helena and challenger Jeff Parady of Pope Valley.
The original versus final vote results in that election confirmed the lack of influence of rural voters as shown in the table below. With all of the initial 9,073 ballots counted, Dillon claimed 4,660 votes, besting Parady’s 4,413. Following the final update on election night, Dillon held a 161-vote lead over the challenger. Weeks later Dillon’s lead had grown by 86 votes to 247, as 9,798 ballots made up the final count. As the table of precincts shows, Parady won more than 80% of the rural votes and still lost the election by 247 votes out of 9,798 votes cast.
After the initial vote count some people held out hope that Parady could pull out a last-minute miracle. While the uncounted ballots made up nearly 40% of all votes cast in the Nov. 2 election, or nearly a third in the case of the District 3 supervisors’ race, Registrar of Voters John Tuteur said that he didn’t expect the remaining votes to swing any of the close races. He said that in his experience, large pools of voters tend to adhere to the same set of trends, meaning that the final 40% of ballots will likely favor the same candidates seen in the first 60%. Tuteur’s experience was supported by mathematical statistics.
The basic question is how many votes really need to be cast and counted to determine the likely outcome of a District 3 election (or any election). The answer came to me through a Pope Valley connection. Bill Eaton, vice president of Creative Research Systems at the time, read my article, "Parady Loses to Dillon — Finally!" and sent me a paper he had written about the basics of statistical sampling.
The confidence level in a poll tells you how sure you can be of the result. It is expressed as a percentage and represents how often the true percentage of the population who would pick an answer lies within the confidence interval. The 95% confidence level means you can be 95% certain; the 99% confidence level means you can be 99% certain. Most pollsters use the 95% confidence level.
The larger your sample size, the more sure you can be (confidence interval) that their answers truly reflect the population. This indicates that for a given confidence level, the larger your sample size, the smaller (better) your confidence interval. However, the relationship is not linear (i.e., doubling the sample size does not halve the confidence interval).
As Eaton points out in his article, “To achieve a 4% confidence interval in a sample you need to survey 600 people. To halve that error to 2% you need a sample of 2,400 or four times as many. Once you get over a population size of 10,000 it makes little difference, and once you exceed 50,000 there’s none really. So, we can effectively define a large population as larger than 10,000.”
It doesn’t make a difference whether you have a large population or not. The same law of large numbers applies. For example, Eaton writes, "To get a 95% confidence level with a 2% margin of error of our nearest city (Santa Rosa, California), which has a population of about 150,000, you need a sample of 2,364. To get the same 2% confidence interval on the population of the USA (308 million in 2010) you need a sample of 2,401. That’s right, only 37 more people!”
The population of registered voters in District 3 in 2010 was approximately 13,000, technically a large statistical population. Using the Sample Size Calculator, we can determine a range of possible sample sizes and confidence levels with confidence intervals.
Let’s investigate the predictive value of the initial 2010 vote count of 5,933 (our sample size, 45.6% of registered voters). The vote count showed that Dillon was ahead by 161 votes with 51.4% of the vote. At a 99% confidence level, that percentage lead was accurate with a confidence interval of 1.23%. Her actual percentage lead in the whole population of voters was thus statistically between 52.6% and 50.2% — more than enough to win.
The final sample size was 9,798 votes. Dillon’s lead had actually remained at 51.4% of the vote. The confidence interval, though, had narrowed to 0.65%. Therefore, 5,933 votes (45.6% of registered voters) accurately predicted the outcome.
While later results may not change the final outcome, “It is my duty to make sure every ballot cast is counted accurately,” Tuteur said.
What does this exercise in statistics say about how many votes are needed to predict an outcome? If we want to have a 99% confidence level in a population of 13,000 voters that one candidate will receive 50.1% of the vote or more, the number of actual votes counted would have to be 7,299, or only 56% of the 13,000 registered. This would be considered a close election.
Every vote counts, whether in the initial count, subsequent counts or the final count. Recent history has shown that contests are being decided by smaller and smaller margins.
“I am concerned that statistical discussions can give voters the wrong impression about how important it is to cast their ballot,” Tuteur said.
Data from the Napa County Elections Department confirm that Napa County voters are very involved, with about a 75% turnout rate. But elections are about more than the winners and losers. They are a barometer of the basic strength of a democracy and the importance of an educated and involved citizenry.
If today's story captured your interest, explore these related articles:
Peter Kilkus is a Napa-County based journalist who has owned and operated the Lake Berryessa News since the 2009.
Appreciate the engaging tone of this article that helps the reader understand the ins and outs of polling. The repeated explanation of statistical confidence (95% or 99%) nicely reinforces key points for anyone and particularly for someone not versed in statistical language. (If I were still teaching, I would share the article with my students to help make this very point!) Thank you so much.