The paper describes and discuss the use of a weighted principal component analysis model (W-PCA) for a spatial analysis of roll-call votes.
I propose to use a set of weights to treat several practical problems that occur in spatial modeling in different legislative bodies as very unequal importance of the divisions or different attendance of the representatives.
This model allows easy and consistent modelling of dynamics within a single legislative term, which is an interesting question that is rather overlooked in the literature.
Most of the studies in the field use (DW- or new alpha-)Nominate or Ideal methods that are based on the theory of ideal points. The W-PCA model does not need such theoretical assumptions and it may be also looked at as simply a way to provide “the best view of the data”. PCA models in general are standard statistical methods with well-known behaviour (being first developed by Pearson in 1901), broadly used by the scientific as well as non-scientific community. There also exist a few cases of using PCA for an analysis of roll-call votes.
Examples of Czech Lower Chamber of Parliament (1998-2002), Prague City Assembly (2010-2014) and from the well-studied 109th US Congress illustrate the model. The example of the Czech parliament and mainly the Prague example demonstrate the indispensability of modelling dynamics within a single legislative term in some cases.
