Over the last few days I have been analysing the publicly available results announced by the Electoral Commission of Zambia alongside the principal independent assessments of the 2026 election: most notably the Christian Churches Monitoring Group's parallel vote tabulation and assessment and the EU Election Observation Mission's preliminary findings. I have also considered the competing interpretations advanced by Nic Cheeseman and Nicole Beardsworth, McQueen Zaza, and Blessings Vava and Tamuka Chirimambowa. There is a great deal of material here, and I would encourage anyone interested in the argument to read those assessments for themselves.
There is broad agreement on one important point: aspects of the result require explanation. CCMG identified 30 constituencies in which official results fell outside its PRVT confidence ranges; the EU EOM separately raised concerns about elements of the tabulation process. The dispute is over what those observations mean. Are they evidence of electoral malfeasance, administrative error, legitimate political heterogeneity, or some combination of the three?
My study attempts to answer as much of that question as the currently available evidence permits. Rather than beginning in 2026, I reconstruct Zambia's electoral development from 1996 to 2026 at constituency level. I then combine longitudinal party-lineage and transition analysis, econometric testing, Monte Carlo simulation, counterfactual analysis and game-theoretic models of voter and political-party behaviour. Competing explanations are subjected to placebo tests, alternative specifications, leave-one-province-out validation, stress tests and attempts at falsification.
The study also uses my published pre-election forecast as a frozen prior. That is useful not simply because its directional prediction proved correct, but because it records before the votes were counted the political structures and mechanisms I expected to matter. Its national expectation was also reasonably close to CCMG's subsequent PRVT estimate. The official result can therefore be evaluated against two independently generated pre-result expectations, while the constituency data allow us to investigate what each national estimate may have missed.
The Actual Report 🇿🇲
TL;DR: Zambia's 2026 election was not an inexplicable electoral discontinuity. Three decades of constituency-level evidence establish that the country entered 2026 with a fundamentally different political structure from the one that preceded 2015: MMD's collapse divided its electoral estate between the two emerging poles, but UPND subsequently integrated its inheritance much more completely than PF and later NRPUP. The result is a demonstrably asymmetric system—exceptionally durable UPND heartland seats, presidential paramountcy alongside continuing parliamentary competition, extreme vote concentration around the two viable presidential coalitions, and a demonstrably shrinking presidential market for third parties even where parliamentary pluralism survives.
The study further establishes that the extraordinary rural–urban divergence in 2026 is not random: it follows historical political lineage, organisational depth and socioeconomic exposure, while social-policy continuity, loss aversion and perceived political risk provide empirically supported mechanisms consistent with particularly strong rural-heartland consolidation. These findings survive longitudinal, specification, placebo, counterfactual, leave-one-province-out and stress testing. The proposition that Zambia's national PRVT deviation can be interpreted as though it arose from a politically homogeneous electorate is therefore empirically false, and the political geography capable of generating unusually large Hichilema margins is real rather than a statistical artefact.
What the study does not—and cannot—settle from aggregate data is whether every ballot contained in the official result is legitimate. CCMG documented direct polling-station discrepancies, the available official totals contain unresolved reconciliation questions, and neither the structural model nor the PRVT can determine how much, if any, of the remaining PRVT–official difference reflects legitimate geographic heterogeneity, administrative error or malfeasance.
That question is no longer principally statistical: it is an accounting problem. Final ECZ polling-station and constituency records must reconcile the votes from station to constituency to national declaration. The study therefore settles the central political question while leaving the forensic one open: Zambia's 2026 result arose from a real and measurable electoral structure capable of producing extreme UPND presidential margins; whether the declared numbers are an exact accounting of that underlying political reality can only be established by reconciling the count. The political map is explained. The count must now be proved.
