Manifesto Project Dataset
Manifesto Project (WZB Berlin Social Science Center)
3.3 million sentences from party platforms in 67 countries, each hand-coded by policy topic.
Party election manifestos coded sentence by sentence into policy categories, aggregated into each party's emphasis on topics such as welfare, the military, and the environment, plus a left-right score. It covers 1,429 parties across 884 elections in 67 countries. The corpus also offers the original texts.
Access
- Cost
- FreeFree for scientific research after creating an account.
- License
- Open, non-commercial onlyTerms of use for scientific research; redistribution forbidden without written permission
- Sign-up
- Free account
- How you get it
- API and Bulk downloadCSV, Stata, SPSS, R API
What's in it
- One row is
- One row per party per election (main dataset); one row per quasi-sentence (corpus)
- Covers
- 67 countries, mostly OECD and post-communist democracies
- Years
- 1945-present
- Updated
- Annual
- Size
- 5,330 manifestos and 3,356,229 coded quasi-sentences
Quality
Decades of funded, documented coding; the standard source for party positions in comparative politics.
- The rile scale is criticized as a crude left-right measure
- Coder reliability varies across countries and decades
Connects to
- Constituency-Level Elections Archive (CLEA): Join on party and election to relate positions to district votes
- V-Dem (Varieties of Democracy) Dataset: Compare party platforms with regime-level indicators
Questions it could answer
- Have center-left parties in Europe moved toward or away from welfare spending since 1990?
- How much did environmental emphasis rise in manifestos after 2015?