Election data criteria

I’m a bit confused about the criteria for the election data. The UK reviewer comment says:

The submission does not meet the definition. There is no detailed data below the constituency level (up to a 100.000 people).

I think that is because of this criterium:

All data should be reported at the level of the polling station

In the UK there are commonly several polling stations in each electoral “ward” (example) A ward is the area for local elections. Several wards make up a “constituency” used in the national elections.

So UK provides the breakdown of votes at constituency level. I don’t understand the benefit of knowing the vote breakdown at the ward level or polling station level. I don’t believe this is collected in the UK, nor could I see this data available for other countries that I checked (Australia & Canada). Can someone explain?

Australian results by polling station for the 2013 federal election are available via the link in the census.

As well as being available for download, data can be viewed online e.g.

I think that the elections data this year is fascinating.
This year, we used the openelections principals to define the granularity of the dataset. You can find it here - Principle 2: Election data is open when it is granular | Open Election Data Initiative

In short - it’s easy to conceal data if it’s not counted in the most prinary level, i.e poll stations. Poll level help to monitor easily and to allow to see mistakes and to avoid elections fraud. @nilleren has a wonderful story of how he found an error in the counting of one of the poll stations in Denmark by using poll station level data.

As for Canada, they do count and report in poll station level, but they publish and display in electoral districts - Election Night Results - Electoral Districts.

I think this shows that we still have work to do when it comes to elections results.

1 Like

Thanks for the examples and explaining, Mor.

I guess it’s a position that OKF is taking as part of a campaign. I imagine that if so many nations are doing the extra effort and money to break down the counts to polling station level, then it’s because of proper research, rather than just a theory that more data is better and the anecdote in Denmark. No doubt there is some proper justification if I was to read through the Open Elections stuff, so that’s fine.

However it just seems crazy that because this one point is not met, OKF is scoring UK worse than Russia and Cambodia. Yet UK elections are considered decent in terms of fraud and fairness, whilst all the results data that is collected is published with easy access, in bulk, with open license etc. It seems an incredibly harsh way of scoring to give 0% purely because OKF’s view is that the data is not detailed enough for the purpose of researching fraud, when it would no doubt agree that it is pretty useful for so many other purposes, such as understanding which MPs are at risk at the next election and analysing that in terms of behaviour over the parliament, campaign spend, etc.

I guess this is a general problem with picking one single dataset to represent each field. No doubt you have many examples of where you can’t accept a similar dataset, to be strict and fair, and yet it yields a crazy score like this.

I agree with David and this is something OKF wants to be very careful about. One thing is taking points off for non-compliance (like granularity below constituency) and another is to null the whole of a country because of one non-compliance (specially one that does not in itself indicates non transparency by default). There will be very little respect for the index if countries that are known and recognised as following proper democratic processes to come off worse than some with less than transparent political processes, it will undermine how seriously the index is taken in the future.


I have just heard about your survey as the Taiwanese government has (understandably), been promoting it, today.

FYI, Taiwan tends not to do too well in International surveys and with a national election coming up on the 16th of January, the government is looking for any good news.

I must admit the huge jump in places does appear odd. However, when I look at the tables and see the UK’s result in RED on the election data criteria, and the decision rationale of the reviewer, it makes me doubt the whole thing.

I agree with Dread and Antonioacuna, that it hurts the Index’s credibility.

I would suggest you review that criterion and revise the results, as quickly as possible.

As for the comparison with Australia, I think that only highlights why you need to change the criterion. There are many differences between the UK and Australian systems - compulsory voting in Australia, for example - but the main differences lie in proportional representation (Australia), versus First Past the Post (UK). Proportional representation requires the use of multiple-member voting districts, whereas First Past the Post results in single-member legislative districts i.e., to elect a person to represent the constituency (local area). Perhaps that is why Australian results are reported at the polling booth level and why the UK doesn’t need to? I note that Canada is also First Past the Post and whilst it can produce at polling level, it publishes at the electoral district level. Whatever the differences, I would still have the UK ahead of Russia and Cambodia!

@Tell10 I have to say that the work in Taiwan appears to be remarkable. Clearly the government has worked hard to go digital and open in lots of these areas and it’s paid off in getting the top ranking. No doubt this will bring plenty of more tangible benefits too in going forward.

That doesn’t change the fact that the index could do with some tweaking, as you suggest. UK probably feels unfairly knocked down by 10% because the electoral category is questionable. But in fairness it is made up for on the spend data, where the whole criterion is potentially called into question by the fact that hardly any other country doing what UK has done in publishing this sort of data.

So I’m very keen to hear more about Taiwan’s open data effort, how it’s achieved such improvement over the past year or two and what the rest of the world’s governments can learn from it.

On election data criteria, Oliver Buckley from UK government has more comment, including this point:

The 1983 Representation of the People Act explicitly requires votes from polling stations to be pooled in the interests of protecting the identity of voters in the event that a candidate receives a very small number of votes.

Looking more at Open Election Data’s call for granularity down to polling station, I think it relates to this guidance in Patrick Merloe’s 2014 work:

Tally sheets at intermediate and final counting centers record dis- aggregated votes, as well as aggregated results tabulations, are easily verified, publicly posted and provided to representatives of political parties, candidates, groups supporting or opposing refer- enda or other ballot initiatives, domestic nonpartisan election monitors, news media and international election observers; and Tally sheets from intermediate and final counting centers are posted on the Internet and other public places at an early date.

Leading to stating the it should be down to the polling station level:

Do the law and regulations require that a copy of all tally sheets, showing the vote numbers entered from all levels, starting with polling site and including all intermediate levels up to the aggregated totals for each vote tabulation center (i.e., disaggregated results as well as aggregated results) be displayed for public inspection?

However I can’t see in the text any more discussion of this, aside from saying that in general a balance must be reached for competing interests and ‘privacy’ must not eclipse ‘transparency’.

His work, like the Open Election Data from OKF just feel like a shopping list of ideals, with little reasoning and pointing to no evidence to back it up.

I’d love to see some actual research in this area, if someone can point us at it, to provide a lot more examples than alluding to a single anecdote from Denmark.

And I think it makes sense for OKF to rejig their scoring next year, so that publishing lots of data, with an open licence, with bulk download, etc, but just not quite as detailed as desired, leads to some points, rather than none.


I believe that the election system is not the caused to the counting systems that each country is using (or showing). The British system, de facto, does not count votes in the most granular level - i.e - poll stations. Why is this the case? Due to legislation. Why it was done this way? I actually do not know. I do find it very positive that British people trust this counting, however, I don’t think it is the most reliable counting of votes (you just need to watch the first season of the Good Wife to understand why…).

I will leave more to NDI to answer here soon, but I want to flag that if we are doing a global measurement, we need to understand that it can not benefit all countries. We are setting these strict criteria so we can get the data we need globally. In this case, to help others around the world to monitor elections.

And for the NDI summary - Unlocking Election Results Data: Signs of Progress but Challenges Still Remain – Open Knowledge Foundation blog

1 Like

The legislation is there to protect individuals identity and voting choice if a small number of people vote at a given polling station. We have some polling stations in rural areas where only 9 or 10 people vote (and that being the total number of people in that community, say in Scotland or Wales) publishing data at that level would in principle reveal the vote of each individual (or make it very easy to ascertain) and given that voting in the UK is cast, by legislation, as a secret ballot, it would be illegal and a breach of the law (and probably some tangential extrapolations of the Data Protection Act) to publish something that by nature can be easily back engineered to the voting choice of individuals.

Just out of curiosity, how can we know that there is actually a polling station for 10 people? I tried to look for the list of all polling stations in the UK and failed to find one. Can you point me for such a list? In a country when one can vote also by post or by proxy, it seems odd to me that there will be a polling station for a congregation of 9 people in total (it also costs a lot of money to set up so many of them).

No doubt though, that privacy and anonymous elections are important, but poll stations usually consist more that 10 people and a station… One solution can be to have poll station for the amount of X people, as it is in many other countries in the world.

Again, out of curiosity - why when it comes to granular elections data I hear more UK people opposing to it than lets say, granular spending data? Does British people have more trust in the elections system than the rest of the world? (If they do, good for them!)