Posts by Andrew Robertson
Last ←Newer Page 1 2 3 4 5 Older→ First
-
I've come to this discussion rather late sorry, but just want to add that (to my knowledge) none of the flag polls have actually been designed to predict the referendum result.
They've been measuring public support for each option, certainly.
If I'd been asked to predict the referendum result using the poll, I would have done a few things differently.
There. Now I've covered myself :)
-
Reply to Sacha.
No it’s not him (he seems like a very reasonable fellow though)
-
Polity: Poll Soup, in reply to Mr Mark,
Thanks for the re-cap Mark.
For the record, the Andrew R posting here is not the Andrew R from Colmar Brunton :)
When posting here I use my full name.
-
Hard News: Decidedly Undecided, in reply to Mr Mark,
Hey Mr Mark/Swordfish
I mentioned this on DimPost but I don't know if you saw it. If you ever want to catch up to talk polls and undecideds, or if you have any poll-related questions I can help answer, just flick me an email at grumpollie@gmail.com
I've got some undecided-related analysis that I've been working on, but haven't had time to refine. I'd be interested in your thoughts on it.
Cheers
Andrew -
Hard News: Decidedly Undecided, in reply to ,
Well politicians are always concerned about public perception.
Tell ya what though - I'm so sick to death of these continual and obvious beat-ups on David Cunliffe. Here he is speaking about something incredibly important, and bloggers and the media choose to focus on a single sentence, and to report it out of context. I guess I hoped we were more mature than that.
I could never be a politician. I just wouldn't be tough enough for it.
I'm no DC fan, but these beat-ups are increasing my support for him.
-
Hard News: Decidedly Undecided, in reply to tussock,
Okay, thanks.
-
Hard News: Decidedly Undecided, in reply to Ian Dalziel,
How would you ever know if the people ‘polled’ are telling the truth…
You don’t, and some people may lie. It would have to be a lot of people though to really make much of a difference in a random sample of 1,000. People are much less obsessed with politics than those who comment on politically-orientated blogs. Not that many people will have a motivation to lie.
In each poll I do chuckle a wee bit at some of the responses we get which help to illustrate people’s (dis)interest in politics. For example – I’d vote for that fellow Key – can’t remember what party he’s in and I’d vote for Winston First and What’s the blue party? and The one that supports farmers.
-
Hard News: Decidedly Undecided, in reply to tussock,
Tussock – it’s not that simple.
The results for each party are non-independent, so if the number of seats for one party changes, this changes the number of seats other parties will get (irrepective of what their range is). Parties with a range crossing the 5% threshold would then add to this complexity and would shift entire ranges altogether. Then to add even more to the complexity, the total number of seats will shift depending on whether parties have ranges that might create an overhang. The number of seat combinations would be ridiculous, and impossible to report.
So no, a little bit of maths won’t do it unfortunately.
-
Hard News: Decidedly Undecided, in reply to steve black,
Just to add to that - 'brand Key' is an interesting one.
I think it works for National at this election because the PM has committed to running and seems fairly healthy. If I was advising a client about their brand though, I'd say don't let it be tied to something that may be temporary.
-
Hard News: Decidedly Undecided, in reply to steve black,
Which might raise some interesting questions about whether campaigning works and actually shifts voter behaviour (versus say, awareness).
Yeah - interesting question. Polling is actually only about 2-3% of the work I do (well, maybe a bit more in an Election Year). Quite a bit of what I do involves measuring the effectiveness of social marketing/behaviour change campaigns. Behaviour change is such a gradual process (unless it's legislated!), and it's near impossible to determine causality.
I see the election campaigns more as brand campaigns, and the internal polls a bit like brand monitors - the parties are trying to find a 'key message' that aligns with their brand and increases credibility among potential voters.