After the very promising response to last night’s tentative suggestion for a survey, I’m going to carry on thinking out loud about it. My draft title for the survey is above.
Firstly, I’m not going to discuss ways of publishing the survey, and ways of responding to attempts to discredit it. For now I’ll just suppose that the results will published on the internet, perhaps in my blog. And secondly I’m not going to discuss whether it will constitute “science” or not. For now I’m just going to assume that it will be at least as good science as anything Tobacco Control produces, and perhaps even a great deal better. It’s not that I don’t think that either of these questions matters, but rather that I think there’s no need to think about them right now.
Instead I’d like to discuss the aims and methodology of the proposed survey.
Aim and Methodology
In this respect I regard the aim of the study as primarily that of finding how what kind of social impact smoking bans have had on smokers. I want to find out how socially isolated or excluded smokers have become. I realise that smoking bans are quite likely to have had an impact on non-smokers as well as smokers, but for the moment I propose leaving them aside.
The methodology of the study is firstly to decide on a set of questions to ask smokers, and then get enough volunteers to go out and ask smokers that they casually encounter these questions, and record their answers, and transfer the responses into emails that get sent to some central hub for later processing at the end of the study period (2 – 3 months).
In this respect, I’ll probably be the central hub. And surprisingly and gratifyingly we already have 8 volunteers pollsters (9 if I include myself)* in Britain, USA, Canada, Spain, and Greece. And that’s after just one day. So we may be hope for twice that number. If these pollsters manage to get the responses of just 20 smokers, that will produce a sample of 160. We should hope to push it up to 500 or even higher.
What we don’t have are the questions. And in this respect I agree with Iro Cyr that the questions should be kept simple. And I agree with some of her proposed questions as well. And for speed and simplicity, I’m also inclined to adopt a check-box approach. I’d like to think that an individual smoker could be surveyed in the period of a minute or less. This is in part because I want to minimize the impact/delay on smokers, and also minimize the effort for pollsters. I also think the questions should be neutral, and not leading questions in any sense.
So my first draft set of questions and check-box replies is:
Do you approve of the smoking ban? [love it][+2] [+1]  [-1] [-2] [hate it]
What effect has it had on your social life? [+3 better] [+2] [+1]  [-1] [-2] [-3 worse]
How often do you now visit pubs and restaurants? [+3 more] [+2] [+1] [ 0] [-1] [-2] [-3]
How much do you now smoke? [+3 more] [+2] [+1] [ 0] [-1] [-2] [-3]
Please summarize your response to the ban in one word: ………………………………..
These are just draft questions. What I hope is that we can settle on some set of questions in discussion.
So the pollster will go out armed with a pen and a set of printed questions. He/she will ask smokers these questions, and will note their answers. So after surveying 10 smokers, there will be 10 completed forms.
There’s also the possibility of smokers filling in the forms themselves, and maybe taking forms home to friends to fill in and return by hand or by post.
I think that in addition to these questions, the pollster should note down the sex and approximate age of the smokers interviewed, and also the circumstances (e.g. outside a pub).
Afterwards, at home, the pollsters transcribe the responses into an email (perhaps of a standardised format) and send them off, taking care not to send them twice.
In addition to this, the pollsters should fill in the questionnaire themselves, and also provide additional data about their locality, the nature of the local ban in their area, how long it has been in place, how well it is observed, what climate conditions are like (i.e. cold/warm outside) were like, and maybe a few other things. The pollsters’ own responses to the poll would be kept separate from other responses.
So there would be a set of data for each area, and a set of observed data about each smoker, and the smokers’ individual responses.
Some questions arise: how do you identify smokers? I suggest that only people who are seen smoking be questioned, because they are manifestly smokers. That’s the acid test. People who say they are smokers may not actually be smokers.
I’m also a little concerned about being too intrusive. Simon Clark has a piece today describing how some smokers dodged a photo survey. I’m wondering whether pollsters (if they smoke) should set smokers at ease by smoking themselves, and thereby winning their trust.
I’ve also given a bit of thought to processing the data which is gathered. As many of you know, I’m a computer programmer, and if the data in the emails returned are of a standard format, I could write some code to read the emails and store the results. Ideally this should be in a database, but I don’t have much experience of those. And while I have fairly good mathematical skills, I haven’t done very much statistics, and am not sure how to work out statistical significances, confidence intervals, and so on. I’d have no problem producing charts, tables, and graphs.
But I also think that the data should be stored in several locations. And so when data is sent by email, it should be carbon-copied to other people (who would just keep it safe).
Apart from the mathematical data processing, there would also be a task of interpretation – drawing conclusions. I think this should be carried out by several people.
As I’ve been thinking about it today, several other concerns have arisen.
Firstly, I’m a bit concerned about what happens if one or more of the pollsters is an antismoking ‘mole’ who will falsify results to show that all the smokers in his area are delighted by their smoking ban, and want more bans. After all, it could happen, now that we know for sure that the antis are reading this blog. Suggestions please.
Secondly, I’m a little concerned about asking questions which might seem intrusive. After all, we’ll be asking people about their personal lives. One question that I would like to ask is whether the smokers are married or single, because I think this makes a difference. But is that too intrusive?
Thirdly, I’m a little worried that the smokers that my proposed methodology will turn up will only be those smokers who are found smoking outside pubs, or on the streets, or at private social occasions. We won’t find the smokers who never go out any more, and only smoke at home.
Finally, this won’t be my survey. It’ll be a co-operative effort. I’ll do some bits, and others will do others. My aim will be not to overload anyone (myself first and foremost).
Anyway, those are my thoughts for today. Over to you. And remember, the more volunteers the better.
*Current volunteer list: Iro Cyr, R, Waltc, Lecroix, RooBeeDoo, gimper30, garyk, nisakiman, Frank.
Further volunteers: Mentesabiertas, Smoking Hot.