Statement re allegations of misuse of http://www.boards.ie/vote

Darragh

Hi there

Comments have been made regarding the voting preferences of members on http://www.boards.ie/vote - in particular questioning whether certain candidates had such high preferences because they were somehow manipulating our system or "vote stuffing" as we call it here.

There are over 550 candidates running in this election and the first preference votes for them, as they stand on Boards.ie/Vote, are legitimate votes from people who have registered with us.

Over 11 years operating, we have developed sophisticated systems and processes to identify duplicate accounts.

We have checked any suspicious patterns of voting and reviewed each individually. To date 117 extra votes have been identified for 17 different candidates and discarded and the account holders contacted. These seem to be primarily enthusiastic supporters eager to support their candidates.

We will not be commenting on who the candidates are.

At the time of writing, there have been 13,950 votes on the system. Of those accounts, over 11,600 had registered before the /Vote went live on Feb 4. We are aware that people have signed up to vote, but also confident there are little to no duplicates registering to manipulate the system.

We are happy that, where possible, every vote on Boards.ie/vote is due to member preferences and/or candidate campaigning. Many candidates have linked to the poll on Facebook or twitter for their supporters and have said that they are "buoyed" by what the results could indicate.

We firmly believe, as we have said previously, that any manipulation of the system by candidate supporters is more-so fooling themselves than anyone else. Our experiment is to see how accurately our members can predict the General Election, nothing more.

Thanks

Darragh

Laminations

Hi Darragh, can you comment as to who made the allegations?

Also, should it not read 'Statement re allegations of misuse of....'

Darragh

Amended, thanks.

I've seen suggestions in a number of places and about a number of candidates and was made aware of at least one tabloid newspaper looking at running a story with the allegations.

jfk75

Darragh,from what I've seen of Boards.ie I find it very hard to believe that you would involve yourself in something so serious and stupid as attempted vote rigging. Your moderators seem very quick to stamp out illicit stuff & operate a good fairness policy on the forum and I just can't see what benefit there would be to either yourselves or to a particular candidate in doing such a thing. In fact it would not only pretty much destroy the image of Boards.ie but would utterly ruin the candidate involved too. Make sure you stand up to whatever excuse for toilet paper tabloid is making these allegations.

Wolfe Tone

No doubt there is slight misuse, but an internet poll can never be accurate, will always be abused/misused and inaccurate imo...


Or maybe The Wolfe Tones "A Nation Once Again" really IS the worlds most popular song as voted for by hundreds of thousands of voters from over 155 countries?


http://www.bbc.co.uk/worldservice/us/features/topten/

Clearly Darragh-some redtop journo has ran out of stuff to write about.I agree that candidates vote stuffing are only fooling themselves.
I think the pm asking people to revote if for example they have swung a different way helps accuracy.That was good to see as one of my concerns was one off voting would miss new trends.

That said,everything always has some flaws,it doesnt take from its good points-my main issues:

Is there any way of stopping voting coming from facebook links? Because obviously if 1oo's from various candidates fan pages come on as a result,party percentages will skew.
Have you any block on uk ip's including NI? if not, Do you know what percentage of the vote is from NI,the UK,abroad or under age?
I hold the view that boards does contain a higher percentage of party followers than real life-more especially from Sf,fg and labour.that to my eyes isn't affecting the national percentages you are showing.But it could mean seats going where they shouldnt in some constituencies imho especially where total vote is smaller.
Just my 2cent anyhow.

jimaneejeebus

Deleted User wrote: »

But it could mean seats going where they shouldnt in some constituencies imho especially where total vote is smaller.
Just my 2cent anyhow.

I don't think it will make a blind bit of difference.

jimaneejeebus

jimaneejeebus wrote: »

I don't think it will make a blind bit of difference.

I suppose not if you support the candidate

Sponge Bob

For a candidate anywhere with minimum 10 votes or more are there any candidates anywhere whose 10+ votes include 50% or more from noobs who registered in Feb 2011 Darragh ??

SugarHigh

This would be like cheating in the leaving cert mocks, which I'm sure someone probably did.

gandalf

It just shows how desperate the rag tops are when they run stories like this.

I see this Boards vote as a indicator of how the members of the site intent to vote. We all know that the demographics are not the same as the real world with a short coming especially of the more aged citizens.

Where I think the parties should take notice of the boards results is the future elections in this country. If they fail to take proper notice of the youth of this country then they will be in for a very rude awakening soon. That is the real meaty story with this poll not the fantasy that boards is rigging the election. If that allegation has come from a political party you should publicise which one made the accusation.

With regard to overseas IPs remember that a lot of people who work for MNC will have their traffic routed through foreign routers. So if they are naughty and checking the site from work it will appear that they are accessing from another country.

UpTheSlashers

SugarHigh wrote: »

This would be like cheating in the leaving cert mocks, which I'm sure someone probably did.

Don't be ridiculous, that has never happened. And even if it did happen, its not as if theres some sort of "Leaving Cert" forum on Boards.ie where students look for the papers off students from other schools. :pac:

Darragh wrote: »

We firmly believe, as we have said previously, that any manipulation of the system by candidate supporters is more-so fooling themselves than anyone else. Our experiment is to see how accurately our members can predict the General Election, nothing more.


Darragh

This is what I'm thinking. What advantage do candidates or supporters think they can get from manipulating this project

Old_-_School

I received polling cards in 2 separate addresses in separate constituencies and I will be voting in both constituencies so should I vote in both in boards.ie vote as well?

jimaneejeebus

Deleted User wrote: »

I suppose not if you support the candidate

No I mean..seeing as how the poll [stupidly] allowed double votes from over zealous members...BANG there goes any reliable statistical mean.

Next the fact that idiots with too much time on their hands can use a proxy..and vote multiple times...BANG statistical mean ruined.

Next the fact that there will be less of the grey vote.....and therefore not representative of the entire electorate anyways.

Under 18's who cannot vote

People from outside the 26 counties who cannot vote


I mean, when I saw it, I just thought it was a bit of fun. I can't believe anyone is taking this with more than a pinch of salt?!?

Bogger77

Old_-_School wrote: »

I received polling cards in 2 separate addresses in separate constituencies and I will be voting in both constituencies so should I vote in both in boards.ie vote as well?

and that would be illegal, in the real world to vote twice

DeVore

jimaneejeebus wrote: »

No I mean..seeing as how the poll [stupidly] allowed double votes from over zealous members...BANG there goes any reliable statistical mean.

Eh? It doesnt do that. Only your last vote counts, it effectively overwrites your previous vote.

Next the fact that idiots with too much time on their hands can use a proxy..and vote multiple times...BANG statistical mean ruined.

Its one vote per account, not one vote per IP. Proxying wont make much difference. Also, one drop in an ocean isnt much of a problem. We have 13,000 votes now. 11000+ from accounts prior to this being announced.

Next the fact that there will be less of the grey vote.....and therefore not representative of the entire electorate anyways.

We are currently working on unskewing our data and then RE-skewing it as per exit polls of age/gender from 2007. Then we will rerun the PR/STV and *that* will negate this objection.

Under 18's who cannot vote

From our demographic data, this doesnt seem to be an issue at all. Almost none of them have.

People from outside the 26 counties who cannot vote

Data is being monitored for this, its not perfect but it doesnt seem to be an issue either.

I mean, when I saw it, I just thought it was a bit of fun. I can't believe anyone is taking this with more than a pinch of salt?!?

It is a bit of fun but with 13,000 votes we can seriously begin to datamine. A lot of people say they cant believe it because it doesnt match with their expectations (!?) or that it doesnt match with the RedC/MRBI polls... polls which are done on much smaller sample sets, dont have PR/STV and have varied quite seriously from their own prior prediction time they are run... while ours has remained more or less rock steady as it grew and grew.

This IS just a bit of fun, but we take our fun seriously.
Come 26th of Feb we will get a better idea of how accurate we are compared to the only poll which actually matters.

DeV.

@ gandalf-it should be easy enough to identify a mnc ip and allow for irish based users.
Time would be the issue with filtering that

A bigger issue would be NI isp's and also voters aged 13 to 17.
Views in that age group would be more radical(I know mine were) and not reflected in either the real vote or a red C
I take DeV's point in another thread that red c's etc are only 1000 voters but their methods of choosing interviewee's are well accepted and tested norms less susceptable I think to anonymous people on the internet who have time to think about skewing or cheating if they are that way inclined.
Those people are only fooling themselves and can't avoid the results of the real vote so are wasters in that regard.

Btw getting people to revote so as to catch trends I repeat is excelent.It remains to be seen if the issues I mentioned are big enough influences to make the boards poll off a bit,they may well be marginal in which case I'm wrong and I often am we'll see.
No harm in the discussion and questions though,I hope.

The Rooster

gandalf wrote: »

Where I think the parties should take notice of the boards results is the future elections in this country. If they fail to take proper notice of the youth of this country then they will be in for a very rude awakening soon.

The "youth of the country" have always been more left wing than the rest. That was the case 10 years ago, 20 years ago, 30 years ago, 40 years ago.

Plenty though lose the oul "spread the wealth" philosophy when they start paying tax themselves.

Lapin

Darragh wrote: »

At the time of writing, there have been 13,950 votes on the system. Of those accounts, over 11,600 had registered before the /Vote went live on Feb 4. We are aware that people have signed up to vote............
Darragh

Why not simply ban new members from voting, thereby eliminating the possibility of duplicate votes. Or put in place restrictions similar to those in place for joining the soccer forum.

Keep up the good work by the way, I'm enjoying the poll and looking forward to comparing the results after the election.

DeVore

This, unfortunately, isnt born out in the numbers.

The single largest demographic for FF supporters is 18-24.

A lot of people are simply restating their personal beliefs and biases as accepted truths (not having a go, its a generally human thing to do). I'm putting my faith in the law of big numbers.

Also, BB... we released this on Friday at 3pm. By that evening we had 4,000 votes, by the end of the weekend we had 10,000. As it grew it didnt shift at all, in fact it was scarily stable meaning that all the new voters were more or less voting the way the previous voters had (or we would see variations in single party numbers, which we didnt). At this size, moving this sample set is simply not possible without automation which would turn up on our radar like Godzilla.

I'm not saying we're predicting the GE here, but I am saying that this is a perfectly clean and accurate poll of the voting patterns of Boards.ie users.

By the way, the average age of a boards poll voter is about 32 so there goes the "its just the yoof" argument too.

DeV.

DeVore

Lapin wrote: »

Why not simply ban new members from voting, thereby eliminating the possibility of duplicate votes. Or put in place restrictions similar to those in place for joining the soccer forum.

Keep up the good work by the way, I'm enjoying the poll and looking forward to comparing the results after the election.

We felt it was more inclusive to allow everyone to vote but we can seperate out the votes of those who joined before or after the announcement of the vote. So we can have our cake and eat it

Fun stat.... the average post count of an "active user" on boards is about 210ish. The average post count of a voter is 730ish. All our old timers are voting it seems

DeV.

jimaneejeebus

DeVore wrote: »

This IS just a bit of fun, but we take our fun seriously.
Come 26th of Feb we will get a better idea of how accurate we are compared to the only poll which actually matters.

DeV.

That's fair enough. I thought you had it so people could vote multiple times. I guess with 13000+ votes it is quite significant. [as much as polls can be significant i guess] good job!

nesf

DeVore wrote: »

Its one vote per account, not one vote per IP. Proxying wont make much difference. Also, one drop in an ocean isnt much of a problem. We have 13,000 votes now. 11000+ from accounts prior to this being announced.

That works for the national percentages but it doesn't work for the constituency votes and the transfer process. We've quite obvious misrepresentation of certain groups in constituencies where they've absolutely no chance of a seat. This distorts the total seat allocation in favour of certain groups (SF and Independents and Greens to pick three such groups off the top of my head that benefit from such distortions by having more eager member bases and/or campaigns to get their base out voting).

Overall numbers are interesting, some of the transfer data is interesting (e.g. how transfer-phobic FF is at the moment etc) but the actual seats being distributed are extremely inaccurate with the current set-up (not that I see a way around this without active sampling being used and votes being restricted).

gerrycollins

Supposing there is no major trend changes as the real day approaches I am very interested in the end result vs the boards.ie/vote result.

I know of an election area where the grey vote may swing the result against what boards.ie result currently stands and are looking forward to be proven wrong or right.

Rabble Rabble

nesf wrote: »

That works for the national percentages but it doesn't work for the constituency votes and the transfer process. We've quite obvious misrepresentation of certain groups in constituencies where they've absolutely no chance of a seat. This distorts the total seat allocation in favour of certain groups (SF and Independents and Greens to pick three such groups off the top of my head that benefit from such distortions by having more eager member bases and/or campaigns to get their base out voting).

Overall numbers are interesting, some of the transfer data is interesting (e.g. how transfer-phobic FF is at the moment etc) but the actual seats being distributed are extremely inaccurate with the current set-up (not that I see a way around this without active sampling being used and votes being restricted).

This is all moot since DeVore says the aim of the voting is to show the bias of Boards.ie not to predict the election. Also we would have to wait until they have done their statistical skewing of the results with the normalisation for age. That could be done at a local level for actual votes, if the number of first preferences were high enough = multiply the votes of a 50-60 year old bracket by 1.5 and round the results, multiply the votes of the 18-25 * 0.75 and round the results, and then transfer as before.


EDIT:

Just to be clear here, for non-mathematical people, the 1.5 and 0.75 are my guesses. Boards would have to work out how un-represented the age groups are and then apply the multiplier biases.

Dubol

Darragh wrote: »

Amended, thanks.

I've seen suggestions in a number of places and about a number of candidates and was made aware of at least one tabloid newspaper looking at running a story with the allegations.

Darragh.
I have voted in my constituency and my lady friend has voted for her constituency. We have both used the same computor. Does this constitute a breach of the rules? If so I apologise.
I have seen remarks on another forum stating that boards.ie poll is not reliable. I know for certain that the owner and a few of his moderators are ardent supporters of FG and that they would not allow any such posts but as of today they have not deleted them. The fact that boards.ie is one of the leading forums may have led to some jealous, misleading comments. Keep up the good work. This is without doubt the number one social network in Ireland.

DeVore

DeVore wrote:

The single largest demographic for FF supporters is 18-24.

on boards?
Regarding the quantity of them,My memory from when I used mod politics here was that FF posters were very few in number compared to other parties,or at least few to broadcast their number in either posts or polls and that was when times were good never mind now!

Boards doesn't age verify it's users does it? To be honest I take what people put down as their date of birth in web profiles with a huge pinch of salt.

nesf wrote:

That works for the national percentages but it doesn't work for the constituency votes and the transfer process. We've quite obvious misrepresentation of certain groups in constituencies where they've absolutely no chance of a seat. This distorts the total seat allocation in favour of certain groups (SF and Independents and Greens to pick three such groups off the top of my head that benefit from such distortions by having more eager member bases and/or campaigns to get their base out voting).

Overall numbers are interesting, some of the transfer data is interesting (e.g. how transfer-phobic FF is at the moment etc) but the actual seats being distributed are extremely inaccurate with the current set-up (not that I see a way around this without active sampling being used and votes being restricted).

+1 to be honest.

I'm going to shut up now though as it's not my purpose to be seen as dissing the poll or the work gone into organising it.I'm just outlining what jumps out at me about it.

nesf

Rabble Rabble wrote: »

This is all moot since DeVore says the aim of the voting is to show the bias of Boards.ie not to predict the election. Also we would have to wait until they have done their statistical skewing of the results with the normalisation for age. That could be done at a local level for actual vites, if the number of votes were high enough = multiply the votes of a 50-60 year old bracket by 1.5 and round the results, multiply the votes of the 18-25 * 0.75 and round the results, and then transfer as before.


EDIT:

Just to be clear here, for non-mathematical people, the 1.5 and 0.75 are my guesses. Boards would have to work out how un-represented the age groups are and then apply the multiplier biases.

You miss my point, I'm not talking about age distribution distorting the result but variance due to the poll being easily stuffed (not nefariously just by the nature of online polls). Makes the constituency and transfer stuff a bit pointless in some respects (but not in others).

Buceph

DeVore wrote: »

It is a bit of fun but with 13,000 votes we can seriously begin to datamine. A lot of people say they cant believe it because it doesnt match with their expectations (!?) or that it doesnt match with the RedC/MRBI polls... polls which are done on much smaller sample sets, dont have PR/STV and have varied quite seriously from their own prior prediction time they are run... while ours has remained more or less rock steady as it grew and grew.

This IS just a bit of fun, but we take our fun seriously.

On the one hand you say it's a bit of fun. But on the other you talking about the ability to "seriously... datamine" and imply that RedC/MRBI aren't as accurate as this thing? You must be having a laugh, and I don't mean saying "it's just a bit of fun" by that, I mean, you must be taking the piss. No-one in Ireland takes constituency polls seriously, bar the candidates themselves on weird occasions, and I've seen how badly wrong that goes from first hand experience. Secondly, the serious polls wouldn't even try and imply the relevance you seem to be doing from this. Thirdly, you're citing the poll results as changing as pointer to bad results. When it's the exact opposite. How many people change their preference all the time, then look at this and how many people go back and change their vote here without being prompted to.

You can mine data all you want, but the collection of your data is suspect meaning the significance of your data is suspect and any implications you take from it are extremely suspect.

The largest statistical sample size you have is 570 or so, and that's for a 5 seater where 10,000 or so preferences will be needed to get at the quota, with a poll of 30,000, so even in your biggest voting area you aren't reaching anything significant. You have the country broken down into however many constituencies, with multiple seats and an average of about 300 votes in each. That is not anywhere near enough to reach a statistically valid sample size. Then you have the problems of this being a community site, where community "groupthink" for want of a better word is going to come into play. Then you have the problem that boards simply isn't an accurate demo/psycho/ethnographic representation of Ireland. And then you have the problems Nesf said of pockets of campaigners having an effect, especially when people are getting elected with 63 votes.

Sure, say it's a laugh. But to say some of the implications you'd like us to draw are ludicrous. Only the broad sweep of this will have any relevance. Anything else that comes out accurate will be through pure luck and not anything statistically sound.

nesf

nesf wrote: »

You miss my point, I'm not talking about age distribution distorting the result but variance due to the poll being easily stuffed (not nefariously just by the nature of online polls). Makes the constituency and transfer stuff a bit pointless in some respects (but not in others).

In addition to that of course,theres the fact that unlike the real poll,we on boards can see how the thing is going to date and may vote differently accordingly.
Mind you this is all academic,people can do that but if they are, send for these little fellahs

CDfm

nesf wrote: »

You miss my point, I'm not talking about age distribution distorting the result but variance due to the poll being easily stuffed (not nefariously just by the nature of online polls). Makes the constituency and transfer stuff a bit pointless in some respects (but not in others).

But won't it be more representative of its demographic on boards and then the distribution of boards users will influence this.

On the on-line poll transfer thing , is it because people do the tick all the boxes thing or not. Are transfer counts/tally's accurate anyway?

I am very cynical about politics and tune out the other media - if it wasn't for the boards coverage I just wouldn't go there.

Its a pity about the vote stuffing thing & I am glad this thread is here as there is something about boards that you expect integrity -so its nice to see it in action.:)

Sparks43

I must say a big thank you to Darragh and all who made this poll happen.It has been enjoyable to look through different Constituencies to see the trends and compare with other polls and surveys

Well Done guys


I cast my vote in the exact way i will next Friday for the candidate who i believe in.I had not looked at results and even if i had it would have no bearing on my vote.

People who try to manipulate votes like this will be sorely disappointed by the real result ie The Ballot

I have a feeling that the Boards poll will be quite close to the Election in a lot of areas and i look forward to seeing th comparison.

Victor

DeVore wrote: »

Fun stat.... the average post count of an "active user" on boards is about 210ish. The average post count of a voter is 730ish. All our old timers are voting it seems

That just proves mike65 and I voted. :pac:

There is a certain element of regression to the mean with a very large number of votes - one vote now only changes the result by a tiny amount, whereas at the start, it could have had a material effect. I've taken some snap shops over the last 2 weeks. The poll went public just before the first snap shot, so there is only a few hundred votes in the first image. I think the image is rounded to the nearest whole per cent.

Buceph

CDfm wrote: »

But won't it be more representative of its demographic on boards and then the distribution of boards users will influence this.

It reflects the people who voted in the poll, nothing more. And it can't represent anything more unless a much bigger quantity of people voted. And even then you'd have skewed *graphics because of the type of people who use boards.

From my first year statistics, to get something approaching a statistically relevant poll, you need a sample size of 5% of the total, or about 2000 people in the case of Irish political polling. Then you get to decide if you want that sample to be randomly selected (and what form of random, and from which source you select your list from which to randomly sample) or whether you want a representative sample, so people of different ages, genders, income, etc. and you have the same problems in selecting them.

Polls will always have difficulties. The ones in papers are usually criticised for being conducted over a landline telephone, when a lot of people opt out of the phone book, and a lot of poor and young people simply don't have landlines. The same goes for a poll conducted on the street in a city at 11am,
who exactly would be on a street at 11am, office workers, the unemployed or stay at home parents? But the general opinion is that you can get around those concerns by getting a big enough sample size.

And boards isn't getting a big enough sample. If it polled everyone on their party preference, they might do well. In that case you have a lot of people being asked a general question. But with the boards poll, you have a lot of people being asked different questions, which skews it to a few people being asked a specific question. Maybe the constituencies with a lot of voters could be more accurate, but I don't think the numbers are big enough for that, without even getting into the sample selection process.

Victor

Dubol wrote: »

Darragh.
I have voted in my constituency and my lady friend has voted for her constituency. We have both used the same computor. Does this constitute a breach of the rules? If so I apologise.

That isn't a problem, both your votes are likely to count. Its the people signing up with 10 accounts that are being targeted.

Buceph wrote: »

From my first year statistics, to get something approaching a statistically relevant poll, you need a sample size of 5% of the total, or about 2000 people in the case of Irish political polling.

Not so sure about that (and I have a PGrad Dip Stat). The number of people you poll merely determines the margin of error.

Approximately:
1,000 +3% with 95% confidence
3,000 +2% with 95% confidence
10,000 +1% with 95% confidence

Those percentages stay the same whether you are polling Kerry South or all of China - the samples just need to be random (acknowledged as an issue here).

Buceph

Buceph wrote: »

Polls will always have difficulties. The ones in papers are usually criticised for being conducted over a landline telephone, when a lot of people opt out of the phone book, and a lot of poor and young people simply don't have landlines. The same goes for a poll conducted on the street in a city at 11am,
who exactly would be on a street at 11am, office workers, the unemployed or stay at home parents?

Just on a point of information,I know for a fact red c are contacting people on mobiles as thats how they contacted a member of my family very recently.

nesf

Victor wrote: »

That isn't a problem, both your votes are likely to count. Its the people signing up with 10 accounts that are being targeted.

Not so sure about that (and I have a PGrad Dip Stat). The number of people you poll merely determines the margin of error.

Approximately:
1,000 +3% with 95% confidence
3,000 +2% with 95% confidence
10,000 +1% with 95% confidence

Those percentages stay the same whether you are polling Kerry South or all of China - the samples just need to be random (acknowledged as an issue here).

Exactly correct Victor. Also depends on how big or small an effect you're measuring. Something like voting is small, ergo the need for large groups of people.

Victor

Deleted User wrote: »

Just on a point of information,I know for a fact red c are contacting people on mobiles as thats how they contacted a member of my family very recently.

Yeah, some of the companies give the basics of their polls on their sites.

http://redcresearch.ie/polling

Poll Accuracy

Polls can tell different stories because of a variety of factors, such as the method with which they are undertaken, the accuracy of their sampling, the way their questions are asked, the order the questions are asked and who they include in the final sample.

RED C are very careful in all our polls to provide a full disclosure of not only the basic method we use, but also the detail of exactly how we sample, what weights we use, and who exactly is included in the results. This has become the standard in the UK and the more transparency we can bring in for polling in Ireland the better.


Which method is more representative?

RED C conducts polls by phone, because we believe they are more accurate. The main reason for their accuracy is that the telephone interviewer has no control over who they interview. They could be calling into an apartment or a mansion. This makes respondent selection truly random.

This has always provided us with a base sample that appears to be more representative than those achieved using a face to face quota approach. With claimed past vote at the last election within margin of error of how people actually voted on the day, face to face quota controlled polls have historically provided a bias, which has then been “adjusted” by the polling.


How do we make sure the sample is accurate?

To ensure the sample is totally representative we use an approach called random digit dialling (RDD) to create a random database of telephone numbers. By doing this we are able to ensure we reach all telephone households, including those that are ex-directory.

It is also important when conducting a poll that you account of all households, including mobile only households that now make up at least 25% of all adults. Half of our sample are interviewed using an RDD landline sample, with the other half conducted using an RDD mobile phone sample, this ensures 98% coverage of the population reaching landline only households, mobile only households and those with both a landline and a mobile.

Quotas are then set on demographics such as age, gender, social class and region to ensure that the people we speak to are representative of all adults aged 18+, based on the very latest Census statistics.
How do we make sure the questions are asked properly?

Often overlooked, the question order, or way in which you ask a question has a significant impact on the results.

RED C always asks vote intention questions before we talk about anything else. This way the results cannot be biased by other questions on the survey. For instance if we were to ask about satisfaction with leaders before vote intention, there is little doubt that this could have an impact on how people declare they will vote, particularly at a time when many people are unsure how they will vote.

Detailed statistical analysis has also shown that it is very important to read out all parties and rotate the order they are mentioned when asking about vote intention. If all the parties are not read out it has been found that smaller parties can be forgotten.


What else do RED C do to improve accuracy of polling?

RED C also use other techniques, developed over the years to improve the accuracy of their polling. These include, measuring only those likely to vote and weighting by past vote.

In any one election turnout might only be at around 60% of all eligible to vote. It is important therefore that we control our voting intention questions by how likely people are to vote. RED C ask respondents to say how likely it is that they will go and vote in a new general election using a ten point scale where 10 means they would be absolutely certain to vote and 1 means they would be certain not to vote. We then exclude anyone who gives a score of 4 or below from our voting intention calculations, as these people will not vote in an election anyway.

RED C also compares the declared past votes at the last general election, to the actual result. This is a final check on the accuracy of our sample. We then weight to the mid point of the two, as we assume that half of the difference between declared past votes can be attributed to faulty recall, and half to remaining political imbalances in the sample.


The Benefit of Regular Tracking

Monthly tracking polls provide us with a wealth of historical data to analyse and judge our polls by. It is very unusual to see significant shifts in party support when tracking on a monthly basis, with only three occasions since 2004 where we have seen party support shift by more than 5% from one poll to another. This makes sense as people don’t change their mind that quickly unless something pretty dramatic has happened. This regular polling then provides another measure for us to ensure the accuracy of our polls, and will be updated regularly when our next poll is published.

DeVore

So long as you used different accounts you should be fine. We check post history and account details when the same ip turns up and determine they are seperate people.

DeV.

[Jackass]

Old_-_School wrote: »

I received polling cards in 2 separate addresses in separate constituencies and I will be voting in both constituencies so should I vote in both in boards.ie vote as well?

[offtopic]

I know you're joking, but just to warn people, as I know a couple of people who receive more than one polling card, it is a criminal offence to vote more than once in an election, and if the glitch is discovered the double votes will be traced and you could face charges.

As it's unconstitutional, I'd imagine they could be relatively serious charges also.

If you do receive more than one polling card, it is wise to report it to the registrations office, and of course, only vote with one of them.

[/offtopic]