Doctors Calls for Testing in Medicine

It takes several years of training and practice for any individual to begin to dispense medicine and care for patients. With this in mind, many patients who receive advise from a physician or a surgeon are bound to believe that that professional makes decisions and chooses the most effective methods for treatment that are available to the doctor. H. Gilbert Welch is far more modest by stating that the practice of medicine today is so complex that sometimes medical professionals choose options but are not certain that they are the most cost effective or even the most useful for patients.

Writing this article in the NYT, H. Gilbert Welch goes through a number of treatment options that doctors have dispensed in good faith but whose overall efficacy was not tested as rigorously as required. he now recommends that the ability to step back and conduct evaluation of one treatment option against another in order to determine relative effectiveness is necessary. One method of going through this is to integrate randomized evaluations as a critical part of decision-making by the profession. This call for more research is interesting because it is not intended to find out new methods and drugs but rather to focus on what works in the repertoire of treatments offered today.

This doctor highlights an important but hardly emphasized factor in innovation and knowledge today. Human welfare could be improved substantially by exploring the utility of existing treatments and knowledge but this is often surrendered to the quest for the new. And if that is applicable in a cutting-edge profession like medicine, one wonders how much so in other areas.               

Why Expensive Musical Instruments Do Not Sound Better

There are several ways in which conventional wisdom and expert opinion is very sensible advice but my reading of Freakonomics has led me to agree that what is the consensus and unquestioned wisdom is frequently plainly wrong and based on no evidence. Ian Sample of the Guardian writes about the results of a modest experiment with persuasive results that tested the claim that antique musical instruments (specifically violins) have a superior sound to more modern versions. Granted that there are limitations to the experiment principally related to the blindfolding of the players together with the small sample of the instruments, I am persuaded that its results are valid.

What the results show is that the instruments that are highly valued as having been created by a master such as Antonio Stradivari do not produce superior sound that it detectable to expert players. As the story states, these unique instruments are very highly valued by collectors and performers and are the prices are often justified on account of their superior quality of sound. What the results imply is that the less valuable instruments are probably not inferior in this respect. To my mind, the value of the Stradivarius instrument is most probably based on their rarity and scarcity. In addition, most of the owners of these antique instruments are very capable musicians whose skill may give the impression that that quality of performance is produced by the rare instrument. Instead, its just that the skill is the real magic.

These findings trace the continuing realization that most claims such as the superior taste of expensive wines, performance of maestros and now the sound quality of antique musical instruments, sometimes fails when subjected to randomized testing.  

Schools and Education

I read Greg Mortenson's Stones to Schools less than a year ago and was impressed by the simple idea that building schools in remote parts of the world was a far effective development tool than many acknowledge. On my part, the theory offered by Greg that building schools in Afghanistan was a far more effective weapon for female child emancipation and longer term peace than the full-scale war that was the alternative, seemed plausible. I thought that while one cannot educate everyone to pacifism, this theory had some merit relative to available alternatives.

Annie Lowrie of the Slate Magazine has put forward this very incisive argument against that theory and one that goes beyond attacking the author. the argument goes that physical construction of schools is not the equivalent of providing an education and therefore the obsession with building schools is itself wrong-headed. The essence of this critique is that while donations are easier to marshal in the name of putting up structures, it is often the provision of education that is a far bigger challenge. Informed by randomized evaluations, the alternative seems to urge that construction should be replaced with payment of superior teachers, de-worming of school children and separating learners into classes that take account of ability.  

  

Keeping Girls in School

A scholar friend sent this to me in a private email:

A colleague here at Wits has received a request to help design a policy intervention that can halve the girl's drop out rates due to pregnancy, by half in two years. He is toying with the idea of a cash transfer to every girl who graduates who gets to fourth form by age of 18 without getting pregnant. He would like to pay the girls 400 rand more than they are paid for child, care at present. Do you think a 400 rand more is a stronger incentive that can keep a class eight girls till form four? Or will it become a perverse incentive that would account for increased abortion rates two years down the line? How and when do cash transfer work as a policy tool?

My Quick Response:

That appears to be a serious attempt to respond to a policy problem and I think it can be designed to work well.

However, it needs to be clarified whether the intention is to reduce the drop out rate alone or to delay early motherhood. The good thing is that both are related because early motherhood does end school attendance. It would be good to review figures to confirm this assumption.

If it is true that early motherhood is a major contributor to drop out rates then we could concentrate on dropping either and hope to get the same result.

My concern though is that incentives are okay but two years are hardly enough to cut rates by 50% irrespective of where the starting line is. There's no reason to make for such an ambitious deadline regarding motherhood because a delay of only a couple of years cannot be used to declare victory. I think that this must be done through a whole primary and secondary school cycle i.e 12 years so that we can determine whether the factors that drop rates at one level necessarily work for the other.

In my view, the design should take advantage of loss aversion tendency. This requires banking the chosen sum say 400 rand now in the girl's name and the sum together with the principal is to be collected upon completion of fourth form. In this way, the behavioral incentive is bound to be stronger because the individual will have to give back that money. This is far better than the mere expectation of the sum at the end of a long cycle.

Regarding abortion, that is bound to happen because of the need to claim the funds by cheating as well. However, it is dependent on what the relative costs of abortion to the final sum are in order to justify that arbitrage. Indeed, this perverse incentive could be made stronger with the sharp two year deadline.

Finally, it is not easy to tell what level of incentive would result in the behavior change required above but I consider that financial incentives are likely to create a decent level of change. So the way to do this is to start it as a randomized experiment involving a significant number of girls in order to isolate the effects of the financial incentives and determine what the relative price of such a delay is. On the whole, I am glad that this being considered and I would like to find out what the final decision is. Keep me in the know.

I will post further information on this.