Political Ideology Scaling
Also known as: NOMINATE, Ideal Point Estimation, IRT Ideology Scaling, Spatial Voting Scaling
Political ideology scaling estimates actors' positions on one or more latent ideological dimensions from their observed choices, most often legislators' roll-call votes, but also survey responses and donations. The dominant methods are Poole and Rosenthal's NOMINATE (1985) and the Bayesian item-response-theory (IRT) approach of Clinton, Jackman and Rivers (2004), which place legislators and the proposals they vote on in a common spatial map.
Key highlights
- Recovers continuous, comparable latent positions from binary choices with strong measurement properties.
- NOMINATE provides a consistent historical scaling of the U.S. Congress across two centuries.
- The Bayesian IRT formulation yields posterior uncertainty (credible intervals) for each estimate.
- Generalizes beyond roll calls to surveys, courts, donations, and social media via the same latent-trait logic.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use ideology scaling when you need continuous, comparable estimates of actors' latent positions from their binary or ordinal choices, classically legislators from roll calls, but also justices, survey respondents, or social-media users. It is appropriate for measuring polarization, mapping coalitions, and validating ideology measures. Choose NOMINATE for the canonical legislative time series, Bayesian IRT for uncertainty quantification and flexible extensions, and always impose identifying restrictions and check dimensionality.
Strengths & limitations
- Recovers continuous, comparable latent positions from binary choices with strong measurement properties.
- NOMINATE provides a consistent historical scaling of the U.S. Congress across two centuries.
- The Bayesian IRT formulation yields posterior uncertainty (credible intervals) for each estimate.
- Generalizes beyond roll calls to surveys, courts, donations, and social media via the same latent-trait logic.
- Estimates are identified only up to rotation/scale, so anchoring and orientation choices are required and consequential.
- Low-dimensional summaries may miss substantively important secondary dimensions.
- Scaling assumes the choices reflect a stable spatial utility; strategic voting and agenda control can distort estimates.
- Cross-time or cross-chamber comparability requires bridging assumptions (common actors or items) that may not hold.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
How is roll-call scaling related to item-response theory?
They are formally equivalent. In educational testing, IRT recovers a student's latent ability from correct/incorrect answers to items of varying difficulty and discrimination. In ideology scaling, the 'items' are votes, the 'ability' is the legislator's latent ideology, and item discrimination indicates how strongly a vote separates left from right. Clinton, Jackman and Rivers made this connection explicit, enabling Bayesian estimation.
Why must the latent scale be anchored?
The likelihood is invariant to rotating, reflecting, and rescaling the latent space, so the model cannot by itself tell left from right or set the units. Researchers impose identifying restrictions, fixing the positions or signs of known anchor actors (e.g., a clearly liberal and a clearly conservative legislator), to orient and scale the dimension so that estimates are interpretable and comparable.
When should I use more than one dimension?
Use additional dimensions when fit diagnostics (correct-classification rates, eigenvalues, predicted-vote improvement) show that one dimension leaves systematic structure unexplained. Historically a second NOMINATE dimension captured the cross-cutting cleavage over civil rights in the U.S. Most contemporary congresses scale well in one dimension, but courts and multiparty legislatures often require two or more.
Sources
- 1.Poole, K. T., & Rosenthal, H. (1985). A spatial model for legislative roll call analysis. American Journal of Political Science, 29(2), 357-384.
- 2.Clinton, J., Jackman, S., & Rivers, D. (2004). The statistical analysis of roll call data. American Political Science Review, 98(2), 355-370.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 22). Political Ideology Scaling. ScholarGate. https://scholargate.app/political-psychology/political-ideology-scaling