Electorate segmentation demo

Anonymised demonstration of a real Latenta engagement - a voter-segmentation & campaign-targeting study in a MENA democracy (CAWI + CATI, n = 6,892, 2026). Country, parties and politicians are generalised; every figure is a real study result. This demo profiles only the secured base and the persuadable / mobilisation target segments; the large non-target group is excluded from profiling and analysis.
Chapters: Link strength: all (≥1.12×) Aversion strength: all (≤0.70×) Hub size:
Closer to the centre = stronger over-index vs the electorate average · dot size = prevalence in the segment · up to 18 sharpest traits
Values are percentages within each segment; colour shows the relation to the electorate average (NET): blue = above average, orange = below average. An arrow marks an index ≥ 1.3 or ≤ 0.7 that is statistically significant (FDR p = 0.1, technical report).
Compare:vs

Co-occurrence of traits

Which traits travel together across the seven segments - pairs that over- and under-index in the same voter groups (weighted correlation of segment indices ≥ 0.70; darker violet = stronger). This is derived from the segment-level patterns, so it shows the trait families the segmentation produces; the full respondent-level co-occurrence is computed in the client engagement. Hover a cell for the pair, a trait name for its source question.

Strongest pairs

Priority matrix - where to spend

The six persuadable / mobilisation segments by readiness (persuadability, horizontal) and strategic reward (the campaign's opportunity index = share × reach × turnout × persuadability, vertical); bubble size is the number of voters, dashed lines are medians. Top-right is where persuasion is both most likely to land and most valuable - the first move. The secured Party Core base is held, not persuaded, so it is not plotted here.

Campaign what-if simulator

A deterministic planning model on measured quantities. Projected governing-party votes = secured base (core × turnout × retention) + won persuadables (segment × reach × persuadability × conversion × turnout, with a mobilisation uplift for low-turnout segments). Move the sliders - everything recomputes instantly. Bars use a fixed scale (full bar = the ambitious scenario of the largest contributor).

Projected parliament - 300 seats

Quiz - which voter segment are you?

Answer a few questions from the instrument and the tool places you in the closest segment, based on the study's response patterns. An indicative classifier on response patterns, not a formal scoring model.