Why 2027 presidential election maths is not a spreadsheet game
The Independent Electoral and Boundaries Commission has set a spending limit of Sh6.1 billion for the 2027 presidential campaign.
What you need to know:
- Ordinary citizens deserve a clearer picture of what mathematics can – and cannot – tell us about the presidential election next year.
- Treating counties and regions as fixed containers of votes that move together by uniform percentages is not dynamical modeling.
During every election cycle in Kenya, mathematics is drafted into politics like a hired witness. Columns of county voter totals are added up, turnouts are multiplied and tidy victory margins are announced months before the Independent Electoral and Boundaries Commission prints a single ballot paper.
In his thoughtful commentary in the Sunday Nation of November 15, Murang’a Governor Irungu Kang’ata offered a candid version of this exercise.
By scaling up the 2022 voter register, testing four historical turnout rates and applying uniform five-percentage-point shifts to binary regional blocs, the governor calculated an upper-bound margin for the 2027 presidential contest.
To his credit, Dr Kang'ata explicitly noted that politics is a social science of tendencies rather than universal physical laws, describing his numbers as an upper bound rather than a prophecy.
Yet in public debate, static “bloc arithmetic” is routinely mistaken for mathematical forecasting.
As a mathematician who models complex, non-linear systems, I believe ordinary citizens deserve a clearer picture of what mathematics can – and cannot – tell us about the presidential election next year.
Treating counties and regions as fixed containers of votes that move together by uniform percentages is not dynamical modeling. It is static spreadsheet accounting.
In ordinary real life, an electorate does not behave like water poured evenly across 47 identical buckets. It behaves like a living network where opinions, economic shocks and local loyalties interact unevenly across time and space.
Three mathematical realities explain why static bloc addition routinely misfires.
Permanent ethnic vaults
First, averages hide decisive local thresholds. Under Article 138(4) of Kenya’s Constitution, winning the presidency in the first round requires more than half of valid votes cast plus at least 25 per cent of the vote in at least 24 of the 47 counties.
Multiplying a single national turnout rate – whether 63.8 per cent from 2022 or 84.6 per cent from 2013 – misses the fact that turnout is deeply differential.
A county with a favourite son or daughter on the ballot may turn out at 85 per cent, while a disillusioned region stays below 55 per cent. Averaging those two figures erases the very geography that decides the race.
Second, voters are not locked into permanent ethnic vaults. Static models start by assuming 100 per cent regional loyalty before subtracting flat percentages for youth drift or incumbency.
Serious electoral mathematics replaces rigid bloc assumptions with stochastic transition matrices – tools that estimate the probability of a voter in one camp shifting to another, staying home or being outnumbered by newly registered young voters.
In an era of severe cost-of-living pressure and fluid youth politics, assuming uniform loyalty across millions of individuals introduces massive compounding error.
Third, political sentiment spreads like a wave, not a light switch. In epidemiological and opinion-diffusion mathematics, ideas and political enthusiasm spread through social contact – across markets, funerals, religious gatherings, university and college campuses and digital networks.
Unpredictable choices
Sometimes a campaign message stalls at county borders. At other times, a sudden economic shock or coalition realignment triggers a rapid, non-linear cascade across previously safe strongholds.
A flat five-per-cent slider cannot capture a tipping point.
So how does genuine quantitative political science forecast an election?
Modern probabilistic models – such as dynamic Bayesian state-space frameworks paired with multilevel regression – do not hand out a single magic number.
Instead, they treat county preferences as moving trajectories that update continuously as new voter registration information, economic indicators and credible polls arrive.
Rather than claiming certainty, they run thousands of simulations to produce a probability distribution – telling us not just who might lead, but how wide the margin of error really is.
Even the finest equations, however, hit a hard wall against human agency. No differential equation can foresee a late coalition merger, a running mate surprise or the entry of a strong third candidate six months before the election day.
For the ordinary citizen, the takeaway is simple. Do not be intimidated – or lulled into apathy – by tidy regional spreadsheets circulating on WhatsApp and other social media platforms.
It should be noted that numbers in politics are useful when they clarify possibilities, not when they pretend the future is already locked in a calculator.
In 2027, the outcome will not be decided by static arithmetic drawn up today, but by the living, unpredictable choices of millions of Kenyans in the ballot booth.
Dr Ogutu Keroboto B. Za'Ngoti is a lecturer in the Department of Mathematics and Physical Sciences at Dedan Kimathi University of Technology. ogutu.keroboto@dkut.ac.ke / okeroboto@gmail.com
KioskNews shows a cleaned-up reading view extracted from the publisher’s page — the original always lives on their site, not ours.