statsjunk.elections¶

laakso_taagepera ¶

FragmentationResult ¶

Bases: BaseModel

Laakso-Taagepera effective number of options (candidates or parties).

compute_laakso_taagepera ¶

compute_laakso_taagepera(
    votes: Sequence[float],
) -> FragmentationResult

Laakso-Taagepera effective number of options for a vote distribution.

What this solves

"How many parties/candidates does this election really have?" is a surprisingly tricky question — counting every name on the ballot overstates it if most votes go to just one or two of them. This gives a single number that answers the practical version of that question: it's close to 1 when one option dominates, and approaches the raw count of options only when the vote is split evenly between them. It's a standard way to compare how fragmented or concentrated the vote is across different elections, districts, or time periods.

N = 1 / Σ pᵢ² where pᵢ is option i's share of the total. It is 1 when a single option takes every vote and approaches the number of options as the vote splits evenly — a fragmentation / dispersion measure.

Parameters:

Name	Type	Description	Default
`votes`	`Sequence[float]`	Vote count per option (candidate or party). Zeros are allowed; options with zero votes don't affect `N`.	required

Returns:

Type	Description
`FragmentationResult`	`effective_number` (Laakso-Taagepera `N`), `largest_share` (the leading option's vote share), and `n_options` (options with at least one vote).

Raises:

Type	Description
`ValueError`	If: - no options are provided - any vote count is negative - the votes sum to zero

References

Laakso, M. & Taagepera, R. (1979). "'Effective' Number of Parties: A Measure with Application to West Europe." Comparative Political Studies, 12(1), 3-27.