@Greg Callus Hey all,

I need to introduce myself – I’m the student at the Columbia Journalism School who recently joined the Tobacco Policy Working Group (self-invited). I’m the person behind the poll.

I was less than happy with the way the previous consultation had gone (small surveys, self-selecting students etc), and as someone who has been on student governments in the past, I wanted to bring a little more rigour to the way in which student opinion is represented generally – not just for smoking, but for issues that will be even more important in the coming year, like financial aid.

Thus I offered to help the Working Group (chaired by Health Services) conduct an opinion poll.

I have been honest with the Working Group from the start – I am generally ideologically opposed to a complete smoking ban, which is why we’ve built in mechanisms so that each step of my working is plotted out in advance, and will be overseen by someone who is very much in favour of the ban to show I’m not fixing it. In truth, I do care about the proposed smoking ban, but nowhere near as much as I care about the integrity of polling.

To the most recent comment – “just a guy with a statistics textbook and a list of UNIs” – you’re almost there, except I don’t have a statistics textbook OR a list of UNIs.

I’m Deputy Editor for one of the larger political blogs in the UK (PoliticalBetting.com), focussing on opinion polls and betting markets, and where the pollsters come to defend their methodology against our commenters. I have run a political campaign in which I was commissioning polls from a major pollster, and my blog has just started commissioning our own national polls as well. I’ve recently written a book with the CEO of another of the polling companies looking at the next British General Election. I can’t claim to be Mr Gallup, but I have as much familiarity with this sort of stuff as the most interested amateur.

The poll will be of a randomised sample of the student body, with questions and methodology signed off in advance. Like any poll, it will not have a 100% response rate, and the responses profile will be non-representative, so it will then be ex post facto re-weighted by school, gender and smoking status to match the population at Columbia. Once conducted and reported back to the Working Group and the VP, the full results, methodology and data tables will be made available to all (prob online), including the weightings applied. All of this is standard methodology for political opinion polls. The response rate and the weighted sample size will dictate the margin of error, but even 500 responses would give a reasonable MoE.

The delay (the poll was due to be sent out on the 4th) was due to a delay in getting a randomised sample of UNIs from CUIT. I am not in charge of this process, as data privacy laws apparently mean that I cannot be given all the UNIs. Similarly, because the poll is being conducted using subscription to a polling site (subs paid by Health Services), they have to manage the distribution process according to the methodology I’ve written. I take over once the poll closes to reweight and write up the analysis.

There’s nothing different in what we’re doing with this poll than what is done for political polls all over the democratic world. Everything will be made transparent as soon as I’ve reported back in a week or so.

If you have any questions, email me (gfc2111) and I’ll get back to you as soon as I can (allow that quite a few people care about this, and suspect black magic or duplicity, so give me a chance). If you’d care to meet me to discuss it, I’m happy to do that too.

Hope this has cleared up some of the questions.

Best

Greg

@Honesty If the University unilaterally implements a smoking ban tomorrow, will we really give a shit? Sure, there’d be anger and OMG FASCISM, but that would die down in a day or two. It’s not like anyone would go on a hunger strike or take over a building.