hmmm, as someone who has taught statistics and research methods for years too I have to say, it isn't statisticians that do that (most statisticians are well aware of the limits of the methods they use as well as the principle of garbage in-->garbage out), it is the researchers who've had a single course or two in statistics (or less) and then go on to use them.

Statistics can be very illuminating if used in conjunction with good research method since good research methods and the scientific method in general are based on AVOIDING the confirmation bias (finding what you expected to find).

But maybe I'm not understanding the context in which you use that saying, Tom.

Wow. Great post.

My comment.

I used to drink/smoked alot when I was younger but pretty much stopped when my daughter was born in '95. Eight years later I was diagnosed.

My personal opinion was/is.

I beleive that you are pre-disposed to this.

Certainly there are 75 yo men and women sitting in a tavern right now chain smoking and drinking scotch. And they have been doing that for 40+ years. Why not them?

Well, I beleive it's in the gene pool. My mom, uncle, grandmother, and great grandmother (same side) all had cancer at some point in their lives. Uncle died from it.

I would love to see a real study based on people that never smoked or drank and their family history of cancer. I think that would be interesting.

Gotta run
-r

And visa versa Tom. You stated that you had spent 20 years as a clinical director in an alcohol rehab center. With that background and having seen the horrible things that alcohol can do to people, I can understand your abhorrence. However, not ALL people who drink are alcoholics as you imply.

You can pit this study against the one that says a glass of red wine a day is good for you. Which will kill you first? The heart attack you get because you don't drink it or the cancer you get because you do? I'll drink a toast to life with Helen.

Eileen

Nelie - That saying about statisticians is partly tongue in cheek (so fitting on this site), and partly to say: Research invites bias. Most research is as much political as it is scientific - funding sources, accountability loops, peer reviews, referees, publication rights and more. Data about a thing is only a map of the thing - not the thing itself. Understanding the data does not lead to real understanding, just description. Setting a description onto paper forces the author to limit how much is said. Good journal work requires summation, abbreviation, encapsulation. A map does not illuminate understanding, it only gives dimensions. Statistics, powerfully descriptive though they may be, trick us into limited understanding and invite bias.

None of us in this forum listened to the odds of our survival. H/N cancers killed most people until quite recently. Statistics about a particular treatment only describe the experiences of a few - and never the experiences of one. The numbers give us indicators, trends, directions of change, possible relationships between variables, ideas about what is happening. They do not illuminate, explain or describe how one will be. Data from the clinical trials of a new chemo drug do very little to describe or predict what my experience with that drug will be - just a well educated guess. I cannot understand that drug for me from the data, only from my experience with it.

And Eileen - forgive me for implying that all drinkers are alcoholic. They are not. But a powerful number of them are and most of those deny this, and deny that alcohol causes them any problems. In all my years in the treatment business, I never met an alcoholic with a drinking problem. Lots with money, or wife, or job problems, but hey 'there's nothing wrong with my drinking.' Hiding behind the rare but visible "gutter drunk" image, lots of drinkers explain away their drinking behaviors as not that bad. Ok. The National Institute of Alcoholism and Alcohol Abuse estimates that 10% of adults in our culture have serious drinking problems. Subract out the abstainers, and that leaves lots of folks impacted. We all defend our own behaviors. I had an MD tell me (with a straight face)once that his definition of an alcoholic was anyone who drank more than he did. nuff said. Its all relative and I'm ok. Thanks for the clarification. Tom

Tom, This is not the place to get into a debate about what scientific research can tell us but I strongly disagree with your overall view and I'm disappointed you're teaching students to throw out the baby with the bathwater since my experience is what they need to develop is more critical thought to distinguish where the baby begins and the bathwater ends.

That said, if you knew my history in academia you'd know that I couldn't agree more that sometimes research is political. But statistical and research METHODS are not political at all, the social structure around them sometimes is. And in fact when the scientific method, statistics and the peer review process are working correctly they eliminate bias and politics that are inherent in other less formal methods of arguing that ones beliefs are objectively correct based on anecdote or more subjective methods. Furthermore, research results are often about way more than description. At their best, especially in fields like medicine, theya re about causation and, eventually about control (prevention of disease and death).

Whew :rolleyes: It took a whole glass of wine to get through all of the above. Amy

Nelie - Well said. Be assured that you need not protect my students. We teach nothing close to the absolutes of babies tossed with bathwater. Critical thinking is the very heart of the matter.

This began with a few 'stranded' statistics about the "causes" of H/N cancers. My fear is that too many people read a statisitic, or a table filled with numbers and thereby feel fully informed - the romance and importance of numbers in action.

My issue here is this: A study about drinking behaviors in relation to cancer is not necessarily helpful to a survivor trying to make a decision about drinking. So many factors play in this complex issue that real controls for bias would be extremely difficult. Rather than choose a new behavior from this study, I would prefer the reader find additional studies before making a personal decision.

The presence of qualified researchers and lots of impressive tables in one study - any study - is likely not going to fully inform an individual's decisions. There is too much respect given to statistics by people who may not understand them. Proof of this is in the highly specialized process of peer review. Science among scientists is wonderful. Raw science in the hands of a non-scientist is an invitation for mis-interpretation and potentially poor choices.

Thank you for clarifying these issues for me. I learn everyday. Tom

O.K.- not trying to be a smart alec here- but I got to ask this question. If John, who loves beer, abstains from now on, or dies from a heart attach after eating all he can to gain some weight back, or is killed by a drunk driver on his way home from work, where will he fit in the statistics?

The SEER numbers track disease incidence and deaths. (Surveillance, epidemiology, and end results) He would be counted in the incidence (if he is in one of the 13 target areas they collect data from that are then extrapolated to estimate the entire US), but he would be counted in the type of disease that he died from, (if not oral cancer) and not at all by SEER if it got hit by a car. (Though that data is collected by other agencies)

By the by, while this study has generated an interesting thread, I stated early on that I thought it was worthless. And while I disagtree with Tom on some alcohol issues I totally agree with him that anyone would be badly informed assuming that ANY given study was the final and sole word on a particular topic.... and even if it came close, in the world of cancer our knowledge base changes daily, yesterday's study seldom carries significant weight down the road as new research leads us down novel pathways.

"Rather than choose a new behavior from this study, I would prefer the reader find additional studies before making a personal decision."

I completely agree with this .

Nelie