Working Paper: Phillips-Nissen-Rodu, Understanding the evidence about the comparative success of smoking cessation methods: choice, second-order preferences, tobacco harm reduction, and other neglected considerations

[Update 22feb15: A new version is now available: Phillips-Nissen-Rodu Smoking or quitting Neglected considerations (pdf). There is a major change in title, to “Smoking or quitting: choice, true preferences, tobacco harm reduction, and other neglected considerations”, but it is still an improved version of the same paper.

Acknowledgements for helpful suggestions that contributed to the new version to Oliver Kershaw and other participants in an ECF discussion ( and Frank Baeyens.

Comments are still welcome. We will be submitted this version but should have a chance to incorporate new suggestions.


Our new working paper is available for download here: Phillips-Nissen-Rodu Understanding the evidence about cessation methods. [Update: link to obsolete version removed]

Abstract:  The extensive literature on methods people use to quit smoking is almost always interpreted in naïve and unhelpful ways. This is partially due to treating smoking cessation as if it were medical disease treatment, despite the fundamental differences. The main problem, however, seems to be a failure to recognize what it means when someone indicates they want to quit smoking. An understanding of the preferences that motivate smoking and cessation allows us to categorize would-be quitters, particularly identifying the difference between first- and second-order preferences for quitting. This demonstrates the absurdity of attempts to determine what cessation method is “best” or even “better”, as well as explaining the frequent failure of medical interventions. This analysis offers advice for both readers of the research and those who wish to quit smoking.

We believe this is a very important paper. Catherine and I have been mulling over the crux of it for literally five years and the three of us have been working on this version of it for about a year. It potentially explains a lot about why smoking cessation efforts are generally failures and smoking cessation policies are even worse. If this were taken seriously, it could really make a big difference.

En passant, the analysis has some other interesting implications. It shows that NRTs are not nearly the failure they appear to be — so long as you properly understand what is reasonable to ask of them. (I gave a talk on this material emphasizing that point to a pharma industry associated audience last week; slides are here if you are interested. Note that this does not include all the key points from the working paper.) Similarly, the analysis points out other disconnects between what happens in practice versus what happens in experimental models or other research.

Comments welcome, either here in the comments section or via email or other media.

25 responses to “Working Paper: Phillips-Nissen-Rodu, Understanding the evidence about the comparative success of smoking cessation methods: choice, second-order preferences, tobacco harm reduction, and other neglected considerations

  1. Pingback: New Phillips-Nissen-Rodu working paper | Anti-THR Lies and related topics

  2. Pingback: Working Paper: Phillips-Nissen-Rodu, Understanding the evidence about the comparative success of smoking cessation methods: choice, second-order preferences, tobacco harm reduction, and other neglected considerations | Norbert Zillatron

  3. In various discussion of this, I have noticed a lot of comments that find the characterization of tobacco use as being about preference and choice to be the novel part of this. We probably need to make it clear in the paper that this is not the key to this paper, so as not to distract from the new core of the present paper.

    See, for example, this where I wrote an entire paper about the economics of preference and choice in the THR context.

    That is not to say that this is not an important aspect of the paper, of course. It could not exist without that as a basis. But what is novel here is the typology and the important role of second-order preferences

  4. You no doubt already know this, so please accept my apologies – I guess I am in need of a rant. But I cannot help but feel you have made a category error.

    Your paper appears to assume that survey’s into smoking habits and desire to quit are done for insight and education. As if providing a deeper, more intelligent view of behavior will assist those professing to “help”.

    But clearly this is a mistake. These survey’s like most of what comes out of Tobacco Control are conducted for political purposes and for the benefit of those in its employ. These survey’s are clearly a political tool for those seeking funding – ‘look at all these smokers who want to quit! you need to give us more more money to provide better cessation services. Or – ‘look at all these people who want to quit, we need to put in more restrictions and legislation to help them.

    Your paper on the other hand, by providing a much deeper,more insightful view of human behavior would, if taken seriously, merely reduce the need for intervention. It places greater emphasis on the autonomy of the individual which by the same measure reduces the role of the expert.

    As if the experts are going to accept this!! You are threatening their funding and employment.

    Still a very well written and thoughtful paper – something that could only emerge from outside the profession of Tobacco Control.

    As an aside, I am impressed by the self reinforcing logic of TC. They spend 30 years in a well organized and Govt supported campaign of dehumanizing and shaming smokers. Then they do surveys that reveal little more than most smokers are ashamed of their habit and then proclaim this as evidence that they need their help? It reminds me of the same self-reinforcing logic of witch burning. Somebody should do a comparative study…lol

    • No doubt that most such research has evil motives. Nonetheless, if you know what questions were asked (and how) and you know the answers, then you can make use of the data. You just have to ignore the ill intentions and political spin (i.e., all the prose that surrounds the reporting of the data) and look at the numbers. In this case, it does actually show that a lot of smokers, almost certainly a majority, will respond to fairly straightforward “do you want to quit” questions with “yes”. That is sufficient for our point.

      You are probably right in your implicit point that many of the “yes” responses actually represent an inclination to conform to the prevalent view that they should want to quit. (It is very common for oppressed peoples to come to internalize the notion that they really are inferior. That is why oppression can succeed.) This complicates things because maybe they are not even expressing a second-order preference, but merely a “yes, Massa”. While I recognize this, my inclination is to not complicated the present paper by trying to deal with it.

      And, thanks.

  5. Being an experimental psychologist, my first and main reaction to these extremely valuable ideas boils down to the following question: How to “measure” the 1st/2nd order preferences in a reliable way, such that the validity of the 4 (or 5) Categories could be assessed in a fair way?

    More particularly, how to identify/distinguish between people in Category 2 vs. in Category 3 (or even 4), when the choice is between proposing/offering short-term combination NRT (let’s say patches/chewing gum; aiming at tobacco smoking cessation and then NRT cessation) vs THR (let’s say ecigs; aiming at long-term substitution of tobacco-smoking by vaping)?

    Finally, how to deal with all this when the smoking cessation “intervention” aims at chronic hospitalized psychiatric patients (many of them suffering from schizophrenia and/or other psychotic disorders)?

    • Carl V Phillips

      Well, being an economist, my main reaction is that first-order preferences are measured by watching what people choose to do. Second-order preferences are a far more abstract concept since (a) they cannot be measured except subjectively and (b) since people do not have any incentive to think enough about these preferences to get them “right” (i.e., unlike preferences that affect choices, these have no material impact), the reported subjective measures are dubious. But for present purposes, the mere existence of the concept, even if completely abstract, is explanatory. The relevance exists primarily at the level of “I want to quit! No you don’t because if you did you would have already done it; what you want is to be someone who wants to quit.” The explanation for the economic disconnect might actually be technically a bit different from that, but I don’t think it matters. Arguably, it is sort of like the everyday understanding of what gravity is, in the sense that even if it is not technically right it is practically useful and close enough. Based on that, I am not sure there is any particular value in measuring them.

    • Carl V Phillips

      More answer:

      I think the key to distinguishing here is getting people to think (perhaps aided) about which type they really are. The proof that they are 3 or 4 rather than 1 or 2 is that they try NRT and they keep smoking. But it would be useful for them to be able to predict that in advance, both to formulate a better strategy and to warned that it may not work as advertised.

      Frankly I think that smoking cessation interventions for psychiatric patients are sheer cruelty, perpetrated by the same evil ANTZ who cheered on Islamic State for banning smoking (temporarily). ANTZ do not care about people at all, only about tobacco, and so don’t care how much suffering they are inflicting. If an interventions that would be a good substitute for smoking for this bereft population were identified, that would be worth recommending to them. But just trying to take away what is often their only joy in life?? What the hell is wrong with these people???

  6. Some further reflections.

    1. With regard to the “diagnostic/measurement” issue – first of all I mostly agree with your thoughts on that. On the other hand, maybe if you asked smokers to reflect upon and then make a choice between the following two hypothetical scenario’s, maybe you would get close to a distinction between categories (1 or 2) versus (3 or 4)?

    Imagine that scientists have developed two kinds of drugs:
    – if you take drug 1, you will really want to quit smoking and be happy with that desire, you will have no difficulties to do so, and you will easily remain abstinent from smoking for the rest of your life;
    – taking drug 2 will make sure that you will be free of the negative health effects of smoking, also when you continue smoking for the rest of your life;
    If given the choice, which drug would you take?

    In the above, in scenario 1 the “really want to quit smoking and be happy with that desire” might capture a 1st order preference not to smoke, and a 2nd order preference that prefers that 1st order preference. The rest of the scenario describes a situation with no/low transition cost, gives a long-term perspective, but at the same time avoids to suggest that one will obtain any of the benefits that were previously obtained by smoking (that is, it is not suggested that the person will “return” to the state of being a non-smoker who has never smoked).
    In scenario 2, the focus is on avoiding health effects (not on avoidingsocial pressure, economic cost, or a desire to be “not-dependent on something”) and on continuing “smoking” (not using snus, or long-term NRT), because it focuses on ecigs as THR, but it might be re-phrased using any of the other terms for different purposes.

    2. A more general remark/question about your characterisation of 1st and 2nd order preferences, from a functional-psychological perspective is the following. How is it possible/brought about that a second-order preference and a first-order preference may be ultimately opposite/conflicting (for example, cat3, 1st order preference ranking smoking over abstinence, but 2nd order preference for a reversal of this 1st order preference, that is, for preferring abstinence (over smoking))?

    Is it because 2nd order preference takes into account a larger or a different “database” of personal concerns/values/information than 1st order preferences? If not, what is the “source” or “mechanism” of the 2nd order preference (and why is that source/mechanism not taken into account when making up the balance between costs and benefits resulting in the 1st order preference)? Or still, in what sense is the 1st order preference the “true” or “fundamental” preference?

    Also when emphasizing the fact that 2nd order preferences are higher-order or meta-preference about one’s 1st order preferences, the theory should still (at least, for a psychologist) identify the elements/source/mechanism/process that result(s) in that 2nd order preference (“what are preferences about one’s preference based upon?”).

    • Carl V Phillips

      Re 1: I agree that suggesting to smokers that they reflect on their true preferences is promising, and indeed that is a lot of the significance of this. I will add at least a passing reference to that to the paper. I will not attempt to go into detail about what the questions should be (a future collaboration, maybe?!).

      I am not sure the ones you suggest are quite the right questions, but I see what direction you are going. Anyone who declines drug 2 seems like a very strong case of someone who is “addicted” in a bad way — indeed, it might be a candidate for a definition of addiction. You are offering someone a drug that lets them keep doing what they are choosing to do now, but with the costs of it dramatically reduced. Anyone who would turn that down must really hate what they are choosing to do. I think that is far too strong a condition for establishing the second order preference to want to want to quit: Someone might want to want to quit given the high health costs, but if giving the chance to eliminate those costs would not want to want to quit under those circumstances. Thus someone declining drug 2 would be a Category 3 but not all Category 3s would decline it.

      Re 2: We should not think of having a second order preference as opposite or conflicting. They are different constructs, and so cannot really conflict — it is kind of like saying your preferences about having dessert conflicts with your preferences about having snacks. There are combinations of those that might seem a bit disconcerting or self-defeating or whatever (e.g., avoiding sweet snacks for health reasons but relishing sweet desserts), but they are not opposite or conflicting per se. This is another point I will clarify in the paper (and thanks for those!).

      Second order preferences should not be thought of as higher order, and especially better, deeper, more enlightened, or anything like that. It is just simple economics:

      It is perfectly reasonable to have a preference to change your preferences. In fact, it is inevitable. We allude to that in the paper by mentioning such things as wish that we liked working as much as playing computer games and such. Everyone should wish they liked eating squash more than they like eating bacon. Perhaps I need to make clearer in the paper that any smoker should wish that his utility from not smoking would increase to being greater than his current utility from smoking (assuming it is not already there: Category 1 or 2) — such a change could only make him better off.

      I think what makes this different (again, something to add to the paper — thanks again, this is really helpful to me — would you mind if I added you to the acknowledgments after I make the changes?) is that most of the time there is no risk that someone will confuse his second order preferences for being first order. Indeed, he will not really ever even think about them. So, yes, if pushed, he will think “yes, I really do wish I liked the taste of squash better than I like the taste of bacon because it is cheap, healthy, and involves no cruel factory farming” but it does not even occur to him to think that thought unprompted, because it is so obviously not something he can change. But when it comes to a behavior that has huge benefits and huge costs, and that he thinks about a lot (e.g., smoking), he might think about his second-order preferences. Of course, what makes *confusion* so common (when it is almost absent for other preferences) is being *told* that he really does not want to smoke, that he gets no benefit from it, that all he has to do to not want to anymore is take a magic drug, etc., as well as being asked questions like “do you want to quit?” (people are not badgered with the question “do you want to quit eating bacon” nearly so much — at least not when I am not around :-). This primes him to mistake the second order for being first order, a mistake that is normally very difficult to make.

  7. Pingback: Working Paper by Phillips, Nissen, and Rodu - Understanding the evidence about the comparative success of smoking cessation methods

  8. Pingback: Understanding the evidence....smoking cessation.....Phillips-Nissen-Rodu

  9. Re re 2. The most important thing first: excellent clarifications of your position – I think not just for me, but also for the bigger audience of the paper as well. I would definitely include them in the manuscript!

    Re re 1. Your comments make it painfully clear, that it is not easy to “ask the right questions”, or at least, that my attempt failed miserably :) For one thing, I had in mind that a preference for/choice of drug2 (and hence rejection of drug 1) would either indicate that she/he is already in category 4, or that it is somebody in category 3-shifting-to-category 4, after/due to reflection on true preferences; whereas somebody rejecting drug2 and expressing a preference for drug1, might either be in category 2 (because now the cost of transition is avoided), or might be in category 1 (the drug just “reinforces” his/her 1st&2nd order preferences).

  10. Nice paper.
    Too bad the FDA couldn’t care less about any of it.
    Enjoy vaping while you can.

  11. Pingback: New version of Phillips-Nissen-Rodu working paper | EP-ology

  12. Pingback: Working Paper: Phillips-Nissen-Rodu, Understanding the evidence about the comparative success of smoking cessation methods, ETC...

  13. Pingback: Understanding the evidence about the comparative success of smoking cessation methods: choice, second-order preferences, tobacco harm reduction, and other neglected considerations | Vape Studies

  14. Pingback: What is peer review really? (part 8 – the case of Borderud et al.) | Anti-THR Lies and related topics

  15. Pingback: FDA Call to Action: Submit comments on FDA’s Third Workshop Regarding E-Cigarettes | Shutting the Gates

  16. Pingback: Utter innumeracy: six impossible claims about tobacco most “public health” people believe before breakfast | Anti-THR Lies and related topics

  17. Pingback: Sunday Science Lesson: 13 common errors in epidemiology writing | Anti-THR Lies and related topics

  18. Pingback: Economic innumeracy in public health, with an emphasis on tobacco harm reduction | Anti-THR Lies and related topics

  19. Pingback: That shady link between Big Tobacco and nicotine gum — Vape Directory

  20. Pingback: Sunday Science Lesson: Debunking the claim that only 16,000 smokers switched to vaping (England, 2014) | Anti-THR Lies and related topics

  21. Pingback: “Dependence” and the danger of adopting the language of your oppressors | Tobacco harm reduction, anti-THR lies, and related topics

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s