This question is posted on behalf of Linda Maguire, member of the Practitioners' Network.

Is there a general standard or benchmark for the number/percentage of invalid votes cast that is considered 'normal' or 'acceptable'? Obviously the kind of voting system in use plays a big role, but are there any comparative figures or analyses out there to consult on the issue?

Three to four per cent is considered by one expert as the “generally accepted high level for rejected ballots,” a proportion that is supported by data from International IDEA’s voter turnout database, where the global average proportion of invalid votes is slightly less than three per cent. The same data also indicates that there are, on average, fewer invalid votes in presidential elections than parliamentary ones, as well as slight geographical variation. It is also pointed out that even a level around the three to four per cent mark can be contentious in a very close election.

India is cited as a case where there used to be a great deal of invalid votes (before electronic voting was introduced). Although there was no benchmark as such, a higher than normal rate of invalid votes often lead to a recount.

As well as the type of voting system, the instructions given to polling staff can also have an impact on the number of invalid votes, the case of Sierra Leone in 2007 being given as an example.

Examples of related ACE Articles and Resources

Encyclopaedia:

•    Criteria to determine when votes should count as valid

Materials:

•    OSCE: Sample of invalid and valid ballots 

•    International IDEA’s Voter Turnout Database

Names of contributors

1.    Ola Pettersson

2.    Needamangalam Gopalaswami

3.    Ron Gould

Although it should not be perceived as a standard or benchmark, the average percentage of invalid votes could be an interesting reference points. Data from some sources, such as International IDEA's voter turnout database, indicates that the global average proportion of invalid votes is a bit less than 4%.

There seem to be a slightly less proportion invalid votes in presidential elections than in parliamentary elections, and the proportion of invalid votes seem to be slightly lower in Oceania, Europe and Asia (closer to 3-3,5%) than in the Americas and Africa (closer to 4,5%).

Yes, the kind of voting system plays a big part in regard to invalid votes. In India there used to be many invalid votes at the time where ballot papers were used (rather than today's voting machines) and there were even cases where the number of invalid votes was in excess of the margin of victory!

As far as I could learn from the old-timers in the Election Commission of India, there was no such benchmark as acceptable level of invalid votes, but a high number of invalid votes always meant that the trailing candidate asked for recount and generally the request for recount used to be granted by the Returning Officers.

But the recount cannot be vague or of the entire lot of votes counted but request had to be specific - to a particular polling station and in a particular round of counting. Each round of counting typically involved about 1 415 000 votes and in each table or polling station the number of votes would be in the region of 1000 to 1200 votes.

I must add that since the introduction of Electronic Voting Machines in Indian elections, there are no invalid votes at all. A big problem related to the ballot papers was the confusion regarding where to stick the rubber stamp and often the stamp used to be put at the space in between two candidates or at back of the ballot paper and so on leading to rejection of the vote as invalid.

Needamangalam Gopalaswami

On behalf of Ron Gould:

The problem in Scotland was that in some areas of Glasgow and Edinborough the rejected ballots were as high as 12 per cent, compared to 3 or 4 percent in previous elections.  In addition, in sixteen ridings, the number of rejected ballots exceeded the number of votes attained by the winning candidate!

All of this to say that in my experience,  3 or 4 percent is the generally accepted high level for rejected ballots,  but in a very close election even this level could result in serious challenges to the results.