| Figure 18a |
DIFFERENCE BETWEEN FOUNDING
AND SUBSEQUENT ELECTIONS |
 |
| Key: VAP = voting age population |
| Figure 18b |
DIFFERENCE BETWEEN FOUNDING
AND SUBSEQUENT ELECTIONS |
 |
| Key: VAP = voting age population |
One of the prominent articles of conventional wisdom in the study of elections is that turnout is higher in the first founding election after a period of authoritarian rule, but then in subsequent elections, participation rates tend to drop dramatically. However, the International IDEA data base disproves this thesis. After taking out all states which had held democratic elections before 1945, and those nations which have held only one election, we analysed the average turnout for first elections, and then the average for subsequent elections in the remaining 81 countries. Surprisingly, we found that the average for first elections (61%) was actually lower than the average for subsequent elections (62%) (see Figure 18a). The mixed pattern was backed up by the fact that turnout in 41 countries dropped between the first and second elections but turnout actually rose in the other 40 countries.
If we go deeper into this question by looking at the data in Figure 19 we find that in three regions (Asia, Western Europe, and the Middle East) turnout actually increased after the founding election - most dramatically in the Middle East where it rose by nearly 25%. In two regions (the Caribbean, and Eastern and Central Europe and the Commonwealth of Independent States) turnout did drop off between founding and subsequent elections, but only in the former communist states was the fall significant. In this region, every post-communist state which held their first democratic elections in the 1990s featured a marked drop in participation at the next election: a situation which was most pronounced in the three Baltic states, which featured a dramatic decline of over 20% in each case (a figure undoubtedly influenced by the change in citizenship laws in these countries). By contrast, in other regions such as Africa, Oceania, and South and Central America, the difference between first and second generation elections is negligible.
While the leaders in turnout during the 1990s are predominantly new democracies, when one looks at broader figures there does appear to be a difference in turnout between "established democracies" and the rest of the world. When classifying "established democracies" we follow the categorization of political scientist Arend Lijphart, who includes all countries that are democratic now, and have been democratic for the last 20 years, which have a population of at least a quarter of a million people1. The overall average turnout in the post-war period for Lijphart's 36 established democracies is 73%, which contrasts with the an average of 59% for the remaining 136 countries. Nevertheless, as Figure 15 illustrates, turnout rates in both established democracies, and the rest of the world have been converging over time. If we assess elections by the environment of political rights and civil liberties in which they are held (Figure 16) then we find that the 353 elections held in countries which were ranked as "free" had average turnout rates of 72%. The 41 elections in "not free" countries gave rise to a rate of 63%, but interestingly the lowest turnout rates, averaging 59%, are found in "partly free" countries. As Figure 17 graphically illustrates, over the last 30 years voter participation in elections held in "free" countries has slowly declined - from a high of 74% in the 1970s to 71% in the 1990s. Conversely, elections held in "not free" countries have demonstrated a marked rise in voter turnout over the last 30 years - advancing from a low of 51% in the 1970s to 65% in the 1990s. Elections held in "partly free" countries did have higher turnout between 1970 and 1990, but since then they have fallen to a low of 57%.
Figure 19
FOUNDING AND SUBSEQUENT ELECTIONS BY REGION 
