DAY 93: Heckman on the Economics of Youth Smoking Initiation

April 19, 2005 12:56 pm by Gene Borio

Last Wednesday, April 13, Defense brought in another economist to shoot holes in DOJ expert witness testimony; this time the target was Dr. Michael Eriksen’s testimony that “defendants’ advertising and marketing are ’substantial contributing factors’ to youth smoking initiation.”

But this time the marksman did not seem a tobacco-chewing denizen of the academic backwoods, but was in fact the prominent Dr. James J. Heckman, Henry Schultz Distinguished Service Professor of Economics in the Department of Economics at the University of Chicago. He is certainly a heavyweight–he’s a Nobel Prize winner in economics and is employed also at universities in London and China. Plus, at $1,000/hour, he’s the most expensive witness yet.

During testimony, we heard a bit about the “Chicago School.” The UC is well-known as a haven for “free-market” economists and libertarian thinkers, and over the last 20 years it has had a large effect on academia, politics and the law (at the very least, Mr. Bernick is an enthusiastic alumnus, as we learned earlier in the trial). Dr. Heckman considers himself “left-of-center” at UC, and “right-of-center” in the rest of the world. He proudly declared that his independence of viewpoint had gotten him in trouble at UC several times–and even somehow prompted a 3-year hiatus at Yale. (”Is that considered an exile to Siberia?” Judge Kessler asked.)

So what does an economist have to do with marketing and youth smoking initiation? In recent years, economics has successfully injected itself into many of the behavioral sciences, even psychology (when poets begin citing Milton Friedman, we’ll know the field has grown like Kudzu and overtaken all others, its victory complete). As one might surmise, the economics angle seems to leave out certain other aspects of human behavior, such as, say, emotion. Perhaps this is why Dr. Slovic’s “affect heuristic” seemed so revolutionary to Nobel Prize winner Kahnneman.) Plato’s vision of a charioteer (reason) guiding and controlling otherwise-runaway emotion (the horses) would not seem to affect this school, which seems to ride around in cars anyway.

Dr. Heckman’s expertise revolves around the evaluation of social and economic policy, and he has studied the effects of societal interventions on actual decision-making. As he explains it,

In laymen’s terms, I have employed econometric and statistical techniques to make careful empirical determinations of whether programs or policies that have been undertaken actually work and how well they work. Thus, my work uses theory and methodology to more accurately draw inferences from observed data.

A little hazy on the concept? I sympathize, as does Judge Kessler, I’m sure. In live testimony, she asked him to identify his particular  area of expertise in the field of economics. He said his expertise was in “empirical economics and . . . research and analysis applied to wide range of subjects.”

As you can see from the above, there’s something of the distanced-professor to Dr. Heckman, and you can see it in his person. In his 60s, with short white hair and a thick but not overweight body that moves fluidly, he tended to talk largely out of the right side of his mouth. It was unclear if this was due to some sort of injury or not. Mentally, he can vary from brief minor distraction to intense, concentrated focus. While not exactly the absent-minded-professor type, you can imagine his wife giving up on trying to get him to come out from his study to eat dinner.

Over decades of work, he has developed standards and protocols for studying how people make decisions, among other human activities. Still, while accepting him as proffered by the Defense as an expert in economics, DOJ balked at his proffer as an “expert in the empirical analysis of human behavior.” By the end of his testimony, Defense had changed their proffer to “expert in the statistical analysis of human behavior.” DOJ accepted that.

Dr. Heckman was asked by the Defense to look at 4 of the main studies that Dr. Eriksen relied on in his testimony. Dr. Heckman found major problems that would prevent anyone from drawing “causal inferences regarding youth smoking and tobacco company marketing.” The studies, he found:

–Did not have a principled way for including and excluding explanatory variables in their models;

–Suffered from selection bias; and,

–Failed to measure actual exposure to advertising campaigns


Dr. Heckman’s oral testimony concerned:

1. Uncontrolled “Unobservables”

There has been no study that could isolate hidden variables, or “unobservables” in the relationship between marketing and youth smoking.

Dr. Heckman pointed to studies which used receptivity to tobacco marketing as a proxy for marketing exposure, ie, naming a favorite ad, or having tobacco-branded merchandise. Such proxies are used because of the near impossibility of identifying 2 groups of subjects that are different only in their actual exposure to advertising.

But, a) marketing receptivity is not scientifically proven to be an adequate proxy for exposure to advertising, and b) there seemed no control for “unobservables.” In this case, say, one such unobservable may be a pre-existing “preference” for the allure of smoking. Such a preference, then, would influence BOTH attention to such advertising/merchandise AND initiation of smoking later. The proxy behavior wouldn’t cause smoking, but merely precede it in time; the preference would be the cause of both the proxy behavior and smoking. [Where such a preference may come from, and at what age a child becomes receptive to tobacco advertising was not addressed here. We do know children are exposed to advertising very early in life.]

2. The Term “Cause” and its Effects

Dr. Heckman said there is no scientific definition for Dr. Eriksen’s term “Significant Causal Factor.” There’s not even a scientific quantification of the term “significant.” It’s possible Dr. Eriksen could have assigned a value to “significant” that would allow his determination to be tested by others.

Along the same line, there was some confusion about the term “causal factor.” I believe Dr. Heckman ran off Mr. Bernick’s rails during redirect. Before Mr. Bernick could get him back on track, Judge Kessler queried him about how causal something might be if it is one of many causes. Interesting question. But suddenly, for some reason, Mr. Bernick took to rustling the large pages on his demonstration easel, so I missed some of this testimony. Perhaps Judge Kessler could hear.

The Defense wanted Dr. Heckman to show that Dr. Eriksen’s work, and the work of the studies he based his conclusions on were unscientific endeavors that failed to show cause. Dr. Heckman’s testimony was very strong, and he himself displayed a high degree of integrity and an almost obsessive scientific curiosity and acumen which certainly fed into that sense of integrity. He may have significantly weakened Dr. Eriksen’s testimony, even though he also seemed to confirm strong associations between advertising and youth smoking.

This may be another area where Judge Kessler feels she must carefully weight differing interpretations of the same data and phenomena.


With Ms. Brooker’s near-interminable, convoluted questions, Dr. Heckman’s answers that would start out well, but then take a sharp left hand turn up some obscure scientific alley that appealed to his insatiable curiosity at the moment, and Mr. Bernick’s Jorge Luis Borges turn of mind, it was a difficult testimony to sit through.

The Parenthetical Ms. Brooker

Ms. Brooker seemed to want to get every possible variable nailed down and cited within each question. This often prompted calls –from Mr. Bernick or from the witness–for a clearer version of the question. Yet in restating, once again Ms. Brooker would cram as many parentheticals into the question as possible.

This technique, as time-consuming and clunky as it was, could be useful in several ways, I suppose:

–It could obviate the opportunity of some to mis-cite testimony 2 or 3 years hence, quoting a passage without the context necessary to truly understand it. We’ve seen that from both sides here.

–It could keep the questions so narrowed as to preclude a wider questioning on redirect. However, the one time such a challenge came up–during Mr. Bernick’s questioning, she protested that she had specifically kept her questions focused on “disaggragated, geographical-level data”–Ms. Brooker’s objection was overruled. See how exciting this trial can be?

Dr. Heckman Slips the Leash

Dr. Heckman was actually pretty good at answering the question right off the bat. But then, if allowed, would veer off the subject at hand. A lot of the respect Judge Kessler will give him is because he obviously loves science, the search for knowledge, and his work. But that intense professional curiosity led to some severe left turns in his answers–and no one seemed to want to interrupt him.

Mr. Bernick: Tap-Dancing Inquisitor

And Mr. Bernick is not the most straight-ahead of cross-examiners, either. I imagine a person stretched on the rack in 15th Century Spain saying to Mr. Bernick, “I’m sorry, could you repeat the question? I didn’t follow.” Apply that to a witness like Dr. Heckman, and you have a recipe for disaster.

I felt often Dr. Heckman had no idea what Mr. Bernick was trying to elicit–and I don’t think he would have cared much anyway. Since oftentimes Dr. Heckman took a question as a springboard for scientific inquiry, his answer could well be in discordance with the Defense’s case. But Mr. Bernick is a verbal Fred Astaire, and can improvise the cross (examination)-steps he needs to let his partner shine.

2 Responses to “DAY 93: Heckman on the Economics of Youth Smoking Initiation”

  1. krueger Says:

    I’m reminded of the industry’s decades-long mantra: no proof that smoking causes cancer. No proof! No scientific proof!

    The trick of course was the standard of proof, which the industry conveniently set so high that by that standard virtually nothing could be proven a cause of disease.

    The industry uses the same trick today on secondhand smoke:

    “they establish a standard of proof that is impossible to meet, such as suggesting that the only way to be sure of a conclusion is to conduct an experiment such as a randomized controlled trial. It is simply ethically and practically impossible to randomly assign people to be exposed to controlled levels of ETS for a given length of time, wait for them to die (or kill them), then conduct autopsy to verify the precise cause of death.”

    The great thing about this trick: it’s not a denial. Instead it’s a magnificent obfuscation, dressed up as commitment to scientific rigor. And it accomplishes two things for the industry: doubt and delay.

    It’s pretty easy to see how doubt is the aim here today: the industry wants to persuade the judge that we just can’t be sure advertising causes smoking. No denial, no one says it doesn’t, we just can’t conclude it does. The evidence that it does, doesn’t meet this standard of proof. This standard that’s conveniently high.

    However, delay is an aim today, too. The evidence mounts and mounts that (surprise) advertising increases consumption for this product. At some point the evidence is compelling. The industry aims to put off that day as long as possible. The industry wants today not to be the day.

    “A demand for scientific proof is always a formula for inaction and delay of the guilty…in fact scientific proof has never been, is not and should not be the basis for political and legal action”

    An example of (private) candor from BAT scientist S J Green.

    So, Heckman says the research evidence isn’t proof. No proof! No scientific proof! By this standard of proof.

    I would love to see Heckman apply that same standard to the tobacco industry’s own economic “research” that claims to show that smokefree workplaces hurt business. For instance:

    It seems that when this industry has something that’s in its interest to regard as proven, the standard of proof is very different. Opinions are facts, anecdotes are data, and this constitutes economic research. The industry does not invoke the same standard here. I call this “the amazing shifting standard” of tobacco research.

    A related trick: that third factor! You measure advertising and smoking, you see a relationship, well, could be that advertising and smoking go up together because of it’s a third factor that causes them both:

    “pre-existing ‘preference’ for the allure of smoking. Such a preference, then, would influence BOTH attention to such advertising/merchandise AND initiation of smoking later.”

    This is quite a flashback — industry PR on smoking and cancer for decades pushed a third factor:

    “[tobacco industry] research projects attempted to show that both lung cancer and smoking were caused by some other ‘third factor,’ such as a person’s psychological makeup, religion, war experiences or genetic susceptibility”

    I find the similarity striking. Decades ago, the third factor was proposed to get tobacco product off the hook. Turns out smoking doesn’t cause cancer after all — no, turns out you can’t be sure that smoking causes cancer. Today the third factor tries to get tobacco advertising and promotion off the hook. Turns out smoking ads don’t cause smoking — no, turns out you can’t be sure that smoking ads cause smoking.

    The industry likewise pushes the third factor on secondhand smoke. You measure exposure to secondhand smoke, you measure disease, you see a relationship, well but it could be a third factor. A confounder.

    In this too, the amazing shifting standard appears. Tobacco industry consultant Lee’s “inconsistent application of his standards to papers that agree and disagree with the tobacco industry’s position” has standards shifting so fast and so far, it’s, well amazing:

    Like Lee, Heckman is very good at pointing out possible confounders in research that disagree with his sponsor’s position.

  2. krueger Says:

    Heckman’s deposition in another trial:

    Pag 62 provides a flavor for Heckman’s left turns: “because you are talking about their preferences between today and tomorrow today and between tomorrow’s today and the next day and those can be different”.

    Pages 154 to 157: after explaining his rational addiction theory, where smokers know their risks, and claiming that the public knows the risks of smoking, Heckman is unable to state the risks of a smoker dying from lung cancer.

    Page 205: Heckman gives his expert opinion that advertising has no effect on smoking; its only effect is on brand choice. An opinion convenient to his sponsor:

Leave a Reply

The primary purpose of this site is to provide information in a timely manner. Postings should be informative. The usual rules apply: No libel, no profanity, no personal abuse, keep it on topic, and short.

If you are scheduled as a court witness, CHECK with your lawyer before posting anything here!