Frequently Asked Questions
All parties that gained at least one seat in parliament are included in the dataset. For Central and Eastern Europe as well as South and Middle America, this threshold is two seats. For presidential elections in South and Middle America we only collect manifestos of candidates who won at least 5% of votes in the first round. However, exemptions do exist. In some cases not all parties with one or two seats respectively are included, this is for example the case with the UK, where only the major parties have been coded in the past. In other cases we have included parties with no seats at all. This usually happens for example when a party used to be very important (e.g. it was part of the government in the previous legislative term) and suddenly lost all its seats. In general, we are interested in collecting all sorts of electoral manifestos. If there is a manifesto you are interested in, feel free to share it with us.
We currently only cover the countries already included in our datasets. Occasionally we add new countries to the dataset. In general, we only add democratic countries.
If you are interested in a recent election not yet covered, we are most likely currently collecting the data. We update the dataset once a year (mostly in late spring). You can either check the website or create an account on our website to receive an email when the new update is out.
We only collect the data for manifestos of parties that won at least one seat (in Central and Eastern Europe and South and Middle America two seats) in a respective national election to the lower house. For presidential elections in South and Middle America we only collect manifestos of candidates who won at least 5% of votes in the first round. We do not collect data on extra-parliamentary, regional or local parties. However, if you need this data and would like to code the manifestos yourself, we are happy to support you. In this case, please contact us. Please also check the Euromanifesto Project and the Regional Manifesto Project.
Since manifesto coding is a lengthy procedure (collecting the manifestos, finding and training a coder, as well as the coding itself) and the availability of our coders varies, we cannot guarantee instant access to data after an election.
Not all parties issue an ideal type of election manifesto. In the cases where no manifesto was available we used substitutes or estimates. You can find the relevant information on document type under the variable ‘progtype’ in the dataset.
Our webpage provides access to all manifestos the project has available. You can download them via our dataset browse dashboard or manifestoR (for R).
We recommend the dataset browse dashboard for a simple and convenient experience. Logging in is required to download the manifestos. Note: If there are no manifestos displayed in the dashboard, we usually do not possess them.
Unfortunately, we do not hold all manifestos we have data for. Some stayed with the respective coder, others were lost over time. We apologise for any inconvenience related to this. There are, however, other manifesto-collecting websites that could potentially hold the manifestos you are looking for, see e.g. polidoc.net.
You can access our Corpus here. The corpus contains all programmes that are currently available in a machine-readable format. Furthermore, the project team is working on digitalising programmes from older project phases. As soon as machine-readable versions can be provided they will be included in the corpus.
The Manifesto Corpus contains English translations of electoral programmes, which we have as machine readable and annotated files. For more information see here.
You can find the definition of each category in the Handbook.
The numbers you see in the dataset for all “per” variables constitute the relative share of statements for each category in relation to all statements in the manifesto. A “0.35” means that 0.35 percent of the manifesto was devoted to that category. Since this is a relative share, the scale can run between zero (no statement at all) and 100 (the whole manifesto is about this category). Please, also consult the Codebook for this and similar explanations.
Excel inserts these signs if the numbers in the cells are much wider than the column. Just expand the column to the right and you will be able to see the data in the cells.
Please have a look at our page for the Election Level do-file, where we provide a Stata script to calculate information about the median voter. Additionally, manifestoR (for R) can also calculate the median voter position.
Currently, we do not provide the government score. For the time being, please consult our links to external projects that extend our data. There you can find the Party Government Data Set updated by Seki & Williams which contains such data.
Information on our core indicators ('rile', 'planeco', 'markeco') can be found in the Codebook of the dataset. Arguably the most prominent is the 'rile' (defined by Michael Laver/Ian Budge (eds.): Party Policy and Government Coalitions, Houndmills, Basingstoke, Hampshire: The MacMillan Press 1992). By combining different categories it gives an estimate of the ideological position of manifestos along a general left-right scale.
In our online visualisation dashboards we have added even more indices. These are explained on the visualization page.
'H' marks sentences that have served as some form of heading in the manifesto, those sentences thus help to document the underlying structure of the election program. For most of the classical text analysis work with the annotated text documents and the manifesto labels they can simply be understood as 'NA'.
This depends on the manual used (see variable manual in the dataset).
For manifestos coded with a manual version < = 4 two scenarios apply.
First, for manifestos from countries that include four digit variables (per 1011-per7062), mostly CEE and Cyprus, one has to add the main categories (per101-per706) plus the four digit variables (per 1011-per7062) plus peruncod:
sum(matches(per[0-9]{3}$)) + sum(matches([0-9]{4}$)) + peruncod
Second, for all other countries it is the sum of the main categories (per101-per706) plus peruncod:
sum(matches/per[0-9]{3}$ + peruncod
Manifestos coded with the newest manual version 5 have four digit variables with underscore (per103_1-per703_2), which are sub-categories of the main categories. Scholars interested in long-time series can continue to add up the main categories and peruncod as for older manual versions. Scholars who work exclusively with manifestos coded with manual version 5 can otherwise also calculate the sum of the main categories that do not have subcategories plus the new sub-categories and peruncod. From the latter, though, they need to substract three subcategories, which are not summed up into their main category but into peruncod:
sum(matches(per[0-9]{3}$)) - (per201 + per202 + per305 + per 416 + per601 + per602 + per605 + per606 + per607 + per608 + per703) + sum(matches([0-9]{3}_[0-9])) + (peruncod - (per202_2 + per605_2 + per703_2))