Sandık Atlası

Sandık AtlasıAnalysis of the Week › Turnout & invalid votes
Analysis of the Week

Where is turnout lowest and invalid votes highest in İstanbul and Ankara?

Election results are usually read through a single question: which party got how many votes. This analysis looks at two different indicators instead — turnout (the share of registered voters who cast a ballot) and the invalid-vote rate. In İstanbul and Ankara, these two indicators reveal clear geographic differences at the neighborhood level.

29 June 2026 · Official YSK data · neighborhood level · 2023 presidential (1st round) · İstanbul & Ankara

Method and stance. This analysis carries no commentary; it shows not parties but two technical numbers: turnout (ballots cast over registered voters) and invalid votes (ballots ruled invalid over ballots cast). Scope: İstanbul and Ankara, the 2023 presidential first round. Only neighborhoods with at least 500 voters were included; institutional units such as prisons were excluded — 1,374 neighborhoods in total. As you will see below, some patterns were verified across multiple elections and back in time. No candidate or person is compared.

Turnout: about 90% almost everywhere

The first thing that stands out about turnout is how small the differences between neighborhoods are. In both cities, neighborhood turnout is remarkably uniform: both the mean and the median sit around 90%. Turnout in Türkiye is generally high; in İstanbul and Ankara, regardless of income or district, most neighborhoods voted in the 88–92% band. So there is no general picture that would let us call certain districts "politically indifferent".

That said, this homogeneous distribution does have its outliers. In a very small number of neighborhoods turnout drops below 70%, and in one below 50%:

District · NeighborhoodVotersTurnout
Ankara · Altındağ — Hacı Bayram2,39149.4%
İstanbul · Beyoğlu — Şehit Muhtar92569.1%
İstanbul · Fatih — Katip Kasım92572.8%
İstanbul · Beyoğlu — Çukur1,48672.9%
İstanbul · Beyoğlu — Kamer Hatun58773.4%
İstanbul · Beyoğlu — İstiklal3,42875.2%
İstanbul · Zeytinburnu — Kazlıçeşme2,75976.6%

Bu liste rastgele değil. İstanbul ayağı neredeyse tümüyle İstiklal–Beyoğlu çekirdeği ile tarihi Fatih'ten oluşuyor: şehrin ticari ve turistik merkezi, konut yoğunluğu düşük semtler. Ankara ayağının başındaki Hacı Bayram ise başkentin en eski mahallelerinden biri; tarihî dokusu, yıllardır tamamlanmamış kentsel dönüşümü ve önemli ölçüde değişmiş nüfusuyla öne çıkıyor.

A one-off or a lasting pattern?

Looking at a single election and declaring "these neighborhoods are indifferent" would be misleading; low turnout in one election may stem from circumstances specific to that day. To check, we tracked the same neighborhoods across four elections falling in four different seasons: autumn, spring, summer and late spring.

The same central neighborhoods, four elections (turnout %)

İstanbul's average is 88–91% in every election. These neighborhoods are 15–20 points below each time.

Neighborhood (Beyoğlu/Fatih)Nov '15Apr '17Jun '18May '23
Şehit Muhtar72717369
Kamer Hatun73767773
Çukur77777373
İstiklal73757775
Katip Kasım (Fatih)77707073
The conclusion is clear: this shortfall is not a one-off but persistent. Not a temporary effect of the season or the day — it is a structural feature of these neighborhoods.

So why? Here the data takes us only so far, no further. While İstanbul's registered electorate grew 10% from 2015 to 2024, the electorate of these neighborhoods did the opposite and shrank: down 28% in Çukur, 26% in Katip Kasım, 21% in Şehit Muhtar. These are central areas that thin out and shrink even as the city grows.

The most plausible explanation is a mismatch between voter registration and actual residence: a person may still appear on the population register at a central address while living elsewhere, and so never shows up at that ballot box. But it would be dishonest to claim ballot data proves this. Low turnout could reflect this registration–residence mismatch, or an altogether different resident population. To separate the two cleanly, one would have to set address-based actual-population data alongside the voter count; we do not have that data. So we state the fact plainly — turnout in these centers is persistently low — but make no firm claim about the cause.

Invalid votes: low on average, but with a pattern

Invalid votes run low in both cities: around 1.5% on average (1.45% in İstanbul, 1.61% in Ankara). Yet here too there is an extreme — and this time a geography that is the exact opposite of turnout. The highest invalid-vote rates cluster not in the urban core but in the countryside:

District · NeighborhoodInvalid
Ankara · Şereflikoçhisar — Sarıkaya5.60%
İstanbul · Şile — Göce5.48%
Ankara · Haymana — Sındıran5.44%
Ankara · Şereflikoçhisar — Akseki5.25%
Ankara · Beypazarı — Zafer4.97%
İstanbul · Şile — Geredeli4.50%
Ankara · Şereflikoçhisar — Sanayi4.28%

Three to four times the average. The list concentrates in the rural south and west of Ankara — especially Şereflikoçhisar, which places four neighborhoods in the top ten — and in İstanbul's forest villages, Şile and Çatalca. In dense urban neighborhoods invalid votes are rare.

A consistent gap by party base

We did not leave this pattern to a single election either. We grouped neighborhoods by the party that came first in the 2023 parliamentary election and lined up the invalid-vote rate across four separate elections:

Invalid-vote rate by the neighborhood's party base (four elections)

Neighborhood groups by the 2023 parliamentary winner. Neighborhoods where AK Parti led show a higher invalid-vote rate in every election.

Neighborhood base2018 pres.2023 pres.2023 parl.2024 local
AK Parti ahead2.09%1.71%2.42%4.11%
CHP ahead1.41%1.17%1.59%2.58%
The gap is consistent in every election: AK Parti–base neighborhoods have about 1.5× the invalid votes of CHP-base ones. (Everyone is higher in the local election — four separate ballots mean more accidental errors — but the ratio holds.)

The finding itself is clear and robust: in neighborhoods where AK Parti led, the invalid-vote rate is about 1.5× that of neighborhoods where CHP led, in all four of the four elections. Not a one-election anomaly but a recurring, consistent pattern — and the gap is far from small. The real question is how we read it, because it is very easy to draw the wrong conclusion. This table does not mean "AK Parti voters cast more invalid ballots." What drives invalid votes is the neighborhood's profile far more than its party: the places with the highest rates were largely rural settlements — and these are where AK Parti and MHP are strong across Türkiye. The academic literature points the same way: as literacy and education levels fall, the invalid-vote rate rises. In short, the gap we see most likely reflects not voters' intentions but the socioeconomic fabric of the neighborhood (rurality, education level).

To test this, we checked whether the gap comes only from rurality. The result: even if we drop rural neighborhoods entirely and look only at urban İstanbul districts, neighborhoods where AK Parti led still show more invalid votes than CHP ones — 1.51% versus 1.05%, roughly 1.4× even here. So the gap cannot be explained by the rural–urban divide alone; socioeconomic factors such as education, age and income may also play a part. But pinning that down would require demographic data beyond the ballot box — what we have is a correlation, not a cause.

Note — an invalid vote is not "fraud". A ballot is ruled invalid when the stamp is misplaced or missing, the ballot is left blank, or it is cast as a deliberate protest; it is not a sign of irregularity. The neighborhood-level party-base gap, likewise, reflects not voters' choices but the neighborhood's socioeconomic profile (education, age, rurality). Ballot data does not show who voted how; individual behavior cannot be inferred with certainty from aggregate numbers (the limit of ecological inference).

Two numbers, two geographies

In short, low turnout is concentrated mainly in the city's central, thinning neighborhoods; behind it most likely lies a population and residence factor, though this cannot be shown conclusively with ballot data. High invalid-vote rates, meanwhile, gather in rural neighborhoods with a different socioeconomic profile. Both indicators relate not to parties but to neighborhoods' demographic and settlement features — two layers that sit beneath the usual "who won" reading of an election.


Data and method. Source: Supreme Election Council (YSK), neighborhood level. Main cross-section: the 14 May 2023 presidential first round; İstanbul and Ankara; neighborhoods with ≥500 voters; institutional units such as prisons excluded (1,374 neighborhoods). Turnout = ballots cast / registered voters. Invalid rate = invalid votes / ballots cast. Persistence tests use the 1 November 2015, 16 April 2017, 24 June 2018 and 31 March 2024 elections; party base is set by the 14 May 2023 parliamentary result. The mechanism (registration–residence mismatch, education effect) is an interpretation supported by correlation and external sources; it cannot be proven from ballot data alone. Individual voting behavior cannot be inferred from aggregate data.

← Analysis of the Week · Sandık Atlası