Our principle data source for voting data post-1945 was IDEA. However, they report having stopped recording data on elections held after 30 Septemebr 2001 (with some undocumented exceptions). In practice, they appear to have stopped somewhat earlier for some countries. Beginning with elections in the late 1990s we sought supplemental data sources, adhering to the decision rule that IDEA was used whenever it was available, but we also recorded data from IFES, IPU, Psephos (Adam Carr’s electoral archive), CNN election watch, and Wikipedia (which has taken over Electionworld.org), usually in that order.
We attempted to adhere to IDEA decision rules in updating data for recent years. Elections that were not meaningful exercises in democratic participation were not included. That does not require them to be entirely fair and free, of course. IDEA stated two conditions for inclusion: that there was a degree of competitiveness and that the franchise was universal.
In practice, it appears that the universal franchise rule was seldom the basis for exclusion by IDEA, not least because they made frequent exceptions to it (they acknowledge at least seven that excluded women). Most countries with a narrow enough franchise to warrant exclusion (the cut-off point was not stipulated) would also be excluded by the competitiveness rule, especially from the 1990s onward. They specify as examples only South Africa and Western Samoa - that could be the entire list. When we obtained data from sources that did not apply such a rule, we did not screen for the breadth of franchise, on the grounds that the actual vote total would encompass this kind of information more fully.
IDEA defines loose criteria for the competitiveness rule: "more than one party contested the elections, or one party and independents contested the elections, or the election was only contested by independent candidates. This criterion excludes the one-party states of North Korea, China and the Soviet Union, but led to the inclusion of elections such as those in Uganda 1995 (where parties were banned) and Egypt in 1976 (where a number of independent candidates ran against the ruling party).Within this ‘gray area’ of competitiveness we have erred on the side of inclusion."
We have tried to follow the same principle when judging data from other sources. To maximize continuity with IDEA, we have assumed that qualified/disqualified countries continue that status unless there is evidence to the contrary. We have relied on the judgments of international observers as to the legitimacy of the election, where available, and on various other criteria – including whether political parties were allowed – where no international judgments were available. For example, IDEA included the 1995 election in Belarus, but we have excluded the elections of 2001 and 2005. Polity and Freedom House both record a steep increase in autocracy since 1995 which warrant this discontinuity. OSCE observers ruled that the 2001 election did not meet their criteria and the very high pluralities for the incumbents (around 90%) in both elections also justified exclusion.
Various special codes were initially recorded and then later recoded in Stata.