I have been using the SHARE data to measure the determinants of subjective life expectancy - the variable isn't ideal. One is simply asked the probability of surviving to a given target age which depends on one's age (70 if you are 60, 80 if you are 70 etc). I have used a few techniques like stochastic frontier estimation,quantile regression, it doesn't seem to matter.
So most of the effects make sense,if your folks are alive,you are not ill,or never smoked you expect to live longer etc. What is striking is the marginal effect of being female is worth about 6 months at most but we know women survive men by a few years on average so this is a puzzle.
Incidentally former smokers expect to live longer than those who never smoked!
Answers please on a post card.
10 comments:
to the younger and less litigation-conscious members of the group (both male and female), could i ask you to restrain your more instinctive responses to Kevin's provocative inquiry.
Kevin,
a few comments
(i) are women too pessimistic or are men too optimistic? Would be interesting to compare subjective data with what the best objective guesses are
(ii) does this gap apply to the other questions regarding expectations. There is one question in there on weather expectations that i think is intended as a control
(iii) What is the actual situation with life expectancy? Is it still the case that a 70 year old man has a lower life expectacny than a 70 year old woman or has much of the things that kill men more than women gone at that stage.
(iv) How does cognitive ability work in here?
(v) you say you have controlled for health. I assume this is self-rated health. have you tried constructing some sort of morbidity index.
(vi) would be interesting to think of some endogenous angle whereby people set expectations to motivate some sort of behaviour.
(vii) Cross-country differences are useful here to see whether this is universal or some sort of cultural-linguistic thing.
no, nothing got to do with the dummy var coding. the point is that there is a "real" life expectancy for both men and women. We know that this is higher for women but Kevin is showing that men have a higher subjective life expectancy. All i meant with point (i) was that we dont know from Kevin's post whether this is due to men thinking they will live longer than they actually will or women thinking that they will shorter than they actually will. Of course, there are other permutations e.g. both men and women overestimating life expectancy with men overestimating more than women etc., etc.,
Martin raises the point about the connection of this work with the work we are doing on risk perception related to drinking and smoking. We have loads of information on this in the bibliographies on the health risk site Kevin. Would be interested in seeing your results on drinking and smoking. Our results for the students is that drinkers and smokers tend to perceive greater personal risk of dying from related diseases than non-smokers and non-drinkers but that they seem to rate the average chance of dying from drinking and smoking to be less than non-smokers and non-drinkers.
To deal with women's 'pessimism' first - i agree that emn's 'optimism' needs to be fcatored in, and there may be an additional reason for this, but one possible reason may be that women use health services more, and so have more (negative) health info in the form of prescription history, consultant referrals, and so on, which will act as an info resource that's more likely to be chronically accessible than in the case of men.
In the case of former smokers thinking they'll do better in general, I'm inclined to think this is to do with self-justification (cf. Katz, 1960) - what would have been the point in giving up and staying stopped if there was no likely helath gain? The former smoker will arguably use a 'faction': the fact, that health gains do accrue, and a fiction, that they can escape the latent unfoloding of earlier damage.
Thanks for those suggestions.I cannot say whether it is males or females expectations that are out of sync' but it would be interesting to figure that out.The marginal effect of being a smoker was a few months-which I think is less than the true effect so maybe it is a more general problem: we underestimate the hazards of life styles etc.
So one question is why do we care about subjective life expectancy?
My suggestions:
1)Expecting to die soon is not very nice so if people seriously underestimate their life expectancy this is a public health issue.
2) People will make decisions (consumption,investment,pensions) based on life expectancy so if those expectations are way off so will the decisions.
3)The "rationality" of expectations is of continuing academic interest.
just dawned on me Kevin that we have life expectancy and a couple of related measures in the Eurostudent survey so this will be an interesting follow-up to your work with the SHARE data.
I followed up Liam's suggestion about "weather expectations": they were asked the probability of whether it would be sunny the next day.It has a very small direct effect on subjective life expectancy so it does pick up innate optimism but it doesn't change any of the results.
However it does not differ by sex:men and women are equally optimistic in general using this variable anyway.
welcome Ronnie - good angle in the sense that if what kevin is finding applies for all the countries then one would suspect a biological base. That's why I would be interested in cross-cultural differences
Health is wealth, its true!! Thats I take care of my kids health and I found so many health related stores with discounted offers at Couponalbum.com..!!
Post a Comment