Do liberals earn less money that conservatives? Nope - liberal's family income averages 6 percent MORE than conservative families.
Do conservatives give less of their income to charity than liberals? Nope again - on average, conservative headed households give 30% MORE!
This surprising (to me) statistic throws new light on our ongoing L/C discussion. For the details, see Washington Post columnist George Will's recent column at: http://www.washingtonpost.com/wp-dyn/content/article/2008/03/26/AR2008032602916.html
Perhaps "compasionate conservative" is not an oxymoron? Perhaps liberals like to talk about helping the poor and downtrodden, but when they take action it is by taking higher taxes from the rich and giving it to the poor while keeping a bit for themselves in the form of bigger government with more social services jobs for them and more votes "bought" with government help programs paid for by us taxpayers?
The data in Will's column are from a Syracuse University professor's book and include the following:
- Conservatives give more blood and donate more time.
- Do you REJECT the idea that "government has a responsibility to reduce income inequality"? If so, you belong to a group that give FOUR times more than those who accept that statement!
- People who live in the reddest states give nearly twice the percentage of income to charity as those in the bluest states.
Also mentioned is the strong correlation of altruism with being associated with an organized religion. Perhaps religious belief is the "The God Delusion" (Dawkins) http://tvpclub.blogspot.com/2008/03/god-delusion.html and "god is not GREAT" (Hitchens) http://tvpclub.blogspot.com/2007/12/god-is-not-great.html, but that type of faith leads to actual, personal giving while the opposite leads to talking about it and getting "good vibes."
Ira Glickstein
I don’t trust George Will’s objectivity or his statistical knowledge, but I did expect Ira to be more skeptical. Statistically speaking, conservatives go to church much more than liberals. Church giving is greater than giving to all of education, health, human services, and the arts combined. Ergo, it follows that conservatives give more than liberals. What’s the big deal?
ReplyDeleteI think Brooks, a Wall Street Journal contributor, is out to sell his book. Why did he send it to George Will? The book will not be published for months. He admits he cut out a lot of qualifying information. He says he started the book as an academic treatise, then punched up the prose to make his point forcefully, when his colleagues and editor convinced him it would sell better.
In any case, look on the Internet and you will find plenty of critiques.
Howard - Thanks for commenting! I wrote that the statistical claims surprised me. As my series on this Blog about Lies, Damned Lies, and Statistics indicates, I tend to be quite skeptical about statistics.
ReplyDeleteDo you have any fact-based arguments that contradict the claim by Brooks or detract from his reputation? I don't know anything about him, other than his status as a professor. I agree with you he sent an advanced copy to Wills to get postive publicity.
However, after reading the linked material, it does make sense that those of us who think the government is not our national "nanny" would take greater personal responsibility to help the poor and downtrodden. Those who think it is a government obligation are like some waitresses who say "not my table" when you ask for a refill on your iced tea.
I agree with you that one factor in greater giving by conservatives is that they tend, more than liberals, to belong to religious organizations and therefore contribute to their activities, some of which are used strictly to promote that religion and have nothing to do with real charity.
However, it can be argued that faith-based charities are more efficient than non-faith-based charities because more of the people involved are volunteers who work for free. Furthermore, non-government charities in general are more efficient than government welfare programs because of volunteer labor and a sense they are spending their own money, a sense absent for some civil service employees.
What about the statistic that C-minds give more blood than L-minds? That would seem to have nothing to do with religious affilliation or lack thereof. In fact, Joel, a declared atheist and a C-mind, volunteers his time to tutor public school children. I would be interested in any evidence that, even if you stripped away money used to promote religion per se, L-minds actually give more time and money to compassionate causes than C-minds.
Ira Glickstein
This is an interesting factor in the difference between L/C minds. I think that it may be related to top-down-bottom-up thinking. I have a some personal observations which may reinforce or elaborate the point.
ReplyDeleteWhile living in France, a highly socialized country (not only politically, but also psychologically), I saw certain anomalies. On a cold winter's afternoon I was walking from my apartment in public housing to the shopping center. A busy street runs between the high rise housing and the shopping center. In order to avoid pedestrians blocking the traffic or installing a traffic light (crosswalks don't work n France), they built a pedestrian bridge. Anyhow, I was walking toward the shopping center when I came upon an old lady trying to pull her shopping cart up the stairs of the bridge, while crowds of people pushed past her as they headed home from work via the shopping center. I reversed course and helped her up the stairs and down the other side as I believe most Americans would do. She was shocked and thanked me. I responded with the typical French expressions. "De rien. C'est normale." "It's nothing" and literally "It's normal." She responded, "C'est pas de tout normale aujourd'hui." "It's not at all normal today."
My analysis based upon this and several other situations I've observed, is that people educated under a socialist system come to expect that someone else (the government) will take care of the situation. The normal concern for one's fellow human becomes abstract. People might be concerned enough to vote for a national fund to install escalators everywhere in order to help all the little faceless old ladies who might need a hand, but they no longer will personally get involved and give her a hand. "Non-socialized" people will vote against such a fund assuming that the government would install escalators in crazy places such as the stairs on the high dive at the swimming pool. They operate under the assumption that anytime aid is needed someone will simply extend a helping hand. Obviously both points of view have their flaws. However, it might explain differences in blood donation. Blood shortages are a national problem, so the government should do something about it, thinks the L-Mind. If only we'd put enough tax money into developing artificial blood, we wouldn't have this problem! With respect -Joel
I don’t have any solid evidence one way or the other about personal values and motivations. Even in my own behavior I find that it is just too complicated to explain by any consistent trend; so why expect to evaluate how “compassionate” some unknown classes of people are?
ReplyDeleteMy first reaction to Will and Brooks was that their choice of giving money and blood as the measure of compassion tells more about conservative values than about the variety of human feelings and motivations.
My guess is that finding a statistically significant difference in mean values of such a personal characteristic between such ill-defined classes as L and C minds is irrelevant, because the variation of motives and behaviors within each class is simply too large, and because the values and motives are too complex to be quantified by such simple measures.
But we still have our opinions, in spite of statistics!
Thanks Joel and Howard for your comments. I agree with Howard that statistical correlation does not necessarily imply cause and effect, or the cause and effect might be in reverse or due to some other common cause. For example, it is true the more firemen you see at a fire, the larger the fire is. Does that mean firemen cause fires? Well, anecdotally, some rogue firemen do cause fires to get a shot of excitement and importance, but, generally, the cause is that a larger fire results in a call, by the fire authorities, for more firemen.
ReplyDeleteThat said, Joel's anecdotal example of his experiences in France carries more weight with we humans than statistics regarding thousands or millions of people.
Conservatives are generally portrayed as cold-hearted, selfish, rich people who cut school lunch programs and throw poor mothers and children out of slum apartments in the middle of the winter. So much so that *I* was surprised by the statistics cited by Will and Brooks.
Indeed, *I* would "cold-heartedly" oppose a government program to provide escalators precisely for the reasons given by Joel. Some civil servant would insist all new high diving boards be so equipped.
I am sure there are a few highly qualified high divers with lame feet who need such assistance. The "fairness" argument would require all diving boards be so equipped to serve these needy citizens equally and not embarass them by requiring someone to carry them up. After all, even those with lame feet are taxpayers and human beings entitled to equal treatment!
The "effectiveness" argument would recognize government only has a certain level of resources. The "Laffer Curve" proves (at least to me) that taxation over some percentage (around 30% + or -) produces a reduction in net yield because it leads the well-to-do to seek tax shelters and make political contributions to get loopholes into law. Thus, every escalator at a diving board, every stringent environmental law, every social welfare proposal, is in competition with other valid government expenses. The fact is that big industry, big labor, big financials and other well-funded special interest groups will always distort any government program to their favor and against small labor, small business, consumers, and the poor.
Get over it! Perfect "fairness" is not possible in this human world. Our only hope is more market-driven efficiencies. The next time you curse us "cold-hearted" proponents of market-driven solution, look at the statistics that show we "small government" C-minds, on average, tend to give more blood and more of our personal time to actually helping others than many L-minds who talk about helping the poor and vote for candidates who propose big government programs.
Ira Glickstein
This topic of compassion has reinforced my impression that the entire C-mind/L-mind discussion is largely an artifact of C-minded over-simplified terminology. Of course we are all trapped in our terminology. [The Sapir-Whorf hypothesis says the grammatical categories of the language a person speaks determines how that person understands the world.] An L-mind would at least talk about a distribution where C-minds talk about a dichotomy.
ReplyDeleteFrom comments about government vs. private management I find that this C- and L-mind difference holds for politics as well as personality. C-minds apparently simply judge private as good, government as bad. L-minds say it’s not that simple.
Ira says I should look at the statistics of market-driven conservative policies. I don’t have statistics, but I know that within my memory there have been many free-market disasters.
Thousands of banks, savings & loans, railroads, airlines, businesses of all kinds, like US Steel, Lockheed, Chrysler, General Motors, Enron, and now Bear-Stearns have been grossly mismanaged or corrupted to the point where we taxpayer have been forced to bail them out to prevent an even bigger catastrophe.
This conservative policy is too simple. It says, reduce our taxes and let the market decide how much risk we take to maximize our profits; but when our management totally screws up be sure the taxpayers can bail us out and provide for our ex-CEOs with millions of dollars. Maybe I’m oversimplifying, but I call this hypocrisy.
The above conversations strengthen my view that a basic L/C mind difference is that liberals are more interested in foolproof justice and conservatives in pragmatic realities (it is what it is...).
ReplyDeleteAnd now for something completely different from an Amazon review:
"Moral Politics: How Liberals and Conservatives Think (Paperback)
by G Lakoff (Author)
5 of 5 people found the following review helpful:
4.0 out of 5 stars Why people are liberals or conservatives, 8 Mar 2006
By Martin Akiyama (Slough) - See all my reviews
Lakoff's theory is that people instinctively see the state as being like a family. Because liberals and conservatives have different views of how families should operate, they have different views of how states should operate. Liberals have a "nurturant parent" worldview, in which the right way to bring up children is to love them unconditionally and to allow them to "be themselves". Conservatives have a "strict father" worldview in which the right way to bring up children is to teach them obedience and self reliance. People might not realise it, but they have unconscious emotional reasons for holding the political views they have, and a big influence on this is how people are brought up. Because liberals and conservatives have such deeply held and fundamentally different views of how the world works, they will never understand one another.
Lakoff has put a great deal of thought into this book. He shows how his theory applies to many different policy issues and to people with differing political views. Lakoff himself is a liberal and at the end of the book he explains why he thinks the liberal worldview is objectively superior to the conservative worldview.
There is a great deal of interesting food for thought in this book. I give it four stars rather than five, because Lakoff gives the impression that EVERYTHING can be explained by his theory, and ignores other reasons why people might hold political views. "
I seem to remember from the preface of the above book the author saying something like, "if you want to determine whether a person is a liberal or a conservative you need ask him/her only this question: " You put your baby to bed and it continues to cry after you shut the door. What do you do?"
Not a bad test, n'est pas?
Stu
At the end of Stu's posting the following question appears as THE test of L- vs C-minded:
ReplyDelete"You put your baby to bed and it continues to cry after you shut the door. What do you do?"
I would double check for a wet diaper, pins sticking (in the olden days when diapers had pins), fever, and other signs of rational reasons for crying rather than sleeping. I would fix whatever I could immediately (change the diaper, correct the sticking pins, give aspirin or other medicine when indicated ...) and then hold and kiss the baby briefly. I would then close the door and let my child cry itself to sleep, no matter how long it took. I might have to knock my head against the wall to keep myself from going in again out of a feeling of empathy and perhaps guilt, but I would force myself to let the child resolve the issues that I could not fix.
What I think the L-mind would do would be to check the obvious (as I did) and then cuddle the child and kiss it for an extended period and then leave the room. If the child continued crying, the L-mind would wait a short time and then go in and repeat the cuddle/kiss, perhaps adding a touch of honey or a toy or something else the child would like. This would go on and on and on, as long as the child cried. This teaches the child that he or she is the King or Queen of the world and that the Universe revolves about him or her and exists for his or her sake alone. The L-minded parent would tell her co-workers how she served her child all night when, in reality, she was teaching her child to cry for longer and longer periods, and developing his or her lung power to do so.
I believe if you timed the minutes the child of the C-mind cried vs the total minutes the child of the L-mind cried, and the number of nights this act was repeated, you would find the child of the C-mind suffered less. (The L-mind in this case is like the guy who had to cut his dog's tail off and, because he was "compasionate" he did it an inch each week.)
You would also find, I believe, the child of the C-mind ends up less selfish, more self-reliant, more successful at school and the world of work, and more loving of his or her parents.
For example: Someone like me! My parents lived with my grandparents and my aunt when I was a baby. They read a C-minded book about how to best raise a baby and they let me cry and cry and cry. My aunt would bury her head in the pillows out of sympathy and anger. One evening my grandmother barged into the bedroom of my (sleeping) parents and asked how they could be so cruel!
Ira Glickstein
Thanks, Stu, for bringing up Lakoff. I had read him some time ago, but forgot about his theory. As I recall, it is more than just a theory. There are studies that support it. The conservatives’ self-reliance should certainly be taught in childhood and should carry over to adult life, but fairness must also be taught, otherwise self-reliance can become selfish and even greedy.
ReplyDeleteIra says, “Get over it! Perfect "fairness" is not possible in this human world. Our only hope is more market-driven efficiencies.” Spoken like a strict father. But market efficiency is not the same as fairness, and just because it is not perfect, does not mean it can be ignored.
For example, Senator McCain said, "I have always been committed to the principle that it is not the duty of government to bail out and reward those who act irresponsibly, whether they are big banks or small borrowers." Liberals can agree with that ideal. But the conservative administration quickly bailed out the big banks and ignored the homeowners. That is not fair.
It is the liberals Rep. Barney Frank of Massachusetts and Sen. Christopher Dodd of Connecticut that propose the Federal Housing Administration guarantee refinanced mortgages that make help homeowners avoid foreclosure.
I think Ira’s description of how he would treat his child sounds like a liberal. In any case, his routine is now close to the standard practice recommended in most parenting books. There is also evidence that liberal and conservative minds have a strong prenatal genetic component. See http://www.psych.nyu.edu/amodiolab/Amodio%20et%20al.%20(2007)%20Nature%20Neuro.pdf
ReplyDeleteThanks Howard for the link indicating a neural component to political orientation. here it is in clickable form if others wish to visit that site:
ReplyDeleteNeurocognitive correlates of
liberalism and conservatism.
The paper indicates L-minds do better on tests where they must break a pattern. Examples are not given, but I used the following in some of my classes. Answer the following immediately!
Q: What comes out of a chimney?
A: Smoke
CORRECT!
Q: What is a funny story?
A: Joke
CORRECT!
Q: What do you call the white of an egg?
A: Yoke
WRONG!!!
Here is another one, answer immediately!
Q: What is on a roof in the winter?
A: Snow
CORRECT!
Q: What may you stub when you walk barefoot?
A: Toe
CORRECT!
Q: What bird says "Caww, Caww":
A: Crow
CORRECT!
Q: What do you do when you see a red light?
A: Go
WRONG !!!
It has long been recognized that L- and C-mindedness has genetic and pre-natal and early childhood components.
Even the lowly sentry guarding Parliament knows it as W.S. Gilbert's immortal words in Iolanthe attest:
SONG--PRIVATE WILLIS.
When all night long a chap remains
On sentry-go, to chase monotony
He exercises of his brains,
That is, assuming that he's got any.
Though never nurtured in the lap
Of luxury, yet I admonish you,
I am an intellectual chap,
And think of things that would astonish you.
I often think it's comical--Fal, lal, la!
How Nature always does contrive--Fal, lal, la!
That every boy and every gal
That's born into the world alive
Is either a little Liberal
Or else a little Conservative!
Fal, lal, la!
When in that House M.P.'s divide,
If they've a brain and cerebellum, too,
They've got to leave that brain outside,
And vote just as their leaders tell 'em to.
But then the prospect of a lot
Of dull M. P.'s in close proximity,
All thinking for themselves, is what
No man can face with equanimity.
Then let's rejoice with loud Fal la--Fal la la!
That Nature always does contrive--Fal lal la!
That every boy and every gal
That's born into the world alive
Is either a little Liberal
Or else a little Conservative!
Fal lal la!
Ira Glickstein
As you know, I'm in favor of the notion of hard wiring in L/C minds. However, before assuming that the reference Howard cited is any kind of evidence, you should also read http://neurocritic.blogspot.com/2007/09/liberals-are-neurotic-and-conservatives.html and the references it contains. Amodio and many other workers in this field seem to be guilty of politicizing the subject. The same is true of Stu's reference to "Moral Politics: How Liberals and Conservatives Think, by G Lakoff Both Ira and I attended a lecture about Lakoff's "frames."
ReplyDeleteLakoff was hired as a consultant by the Democratic National Committee after their failure in the 2004 elections, in order to teach them how to frame issues in such a way as to make conservatives look bad. Lakoff is often presented as a scientist when he is in fact a paid political operative. His imaginative tale about conservatives coming from paternal authoritarian families and liberals coming from maternal caring families is unsupported by any hard evidence. Social science is an oxymoron. With respect -Joel
Thanks Joel for your link, repeated here in clickable form: Liberals are Neurotic and Conservatives are Antisocial.
ReplyDeleteAccording to that source, the subjects of the study linked to by Howard were college students and there were only 7 self-described conservatives to 26 (or 36) liberals. There was no data on the gender of the participants. If the liberals were mostly female and the conservatives mostly male, the data, if meaningful, might be related to gender differences alone. There was also a mention the two biggest outliers were not excluded, implying that had they been excluded (as is normal in statistical studies when a few data points are excessively far from the mean) the results would have been different, perhaps null. (However a later coment denied there were outliers.) Also, the error rate differences, while perhaps statistically significant, were 44% for conservatives and 34% for liberals, not exactly earth-shaking.
In any case, it makes sense to me that conservatives, who tend to want to conserve current customs and patterns until and unless there is good evidence that changing them would be beneficial, might tend to follow the pattern and thus incur greater error rates. Liberals, who tend to go for anything labeled "change" or "new and improved" etc., might be more likely to change the pattern of responses and thus get better scores.
But, there is a danger in assuming such biases. For example, I always assumed conservatives wanted to conserve current customs because, under those customs they were successful and rich. The Brooks study refered to by George Will indicates conservatives are not as rich as liberals. I also assumed conservatives were less compasionate, since we are often described as "hard-nosed" or "hard-hearted", but the study shows the opposite when it comes to actual actions, such as giving blood and time and money, where conservatives give some 30% more.
Finally, as to Lakoff - I remember the presentation by Rob't Ridgard, one of the stalwart (and reasonable) liberals in our local Philosophy Club attended by Joel and me. The fact Lakoff is a political consultant and has strong political opinions favoring the left does not, IMHO, devalue the truth of his ideas concerning proper "framing" of arguments. We conservatives can learn much from his ideas.
Ira Glickstein
I agree with Ira that being a “paid political operative” does not discredit Lakoff’s theory. Joel also says Lakoff’s theory is an “imaginary tale.” Of course, that is the way all good theories begin. It is a scientific theory because its validity could be decided by a statistical survey. Lakoff is one of the most brilliant scientists that I know. His study “Where Mathematics Comes From” (with Rafael Nunez) is a psychological study of the foundations of mathematics. I presume Joel would not object to this work because he a “paid political operative.” Joel concludes (“respectfully”): “Social science is an oxymoron.” I would like to know why he thinks that. Is it because it appears too statistics-based to be a science?
ReplyDeleteI would like anyone’s opinion of Max Weber’s ideas, a founder of modern sociology. For example, his best-known book is about the foundations of capitalism. Weber traced the origins of the Protestant material success ethic to the Reformation.
Briefly, the Roman Catholic Church assured salvation only to individuals who accepted the sacraments and submitted to the clerical authority. The Reformation removed such assurances. Psychologically this created insecurity and lack of self-confidence. Weber argued that Protestants had to look for other sources of salvation. Some form of self-confidence had to replace priestly assurance of God's grace, and worldly success became a measure of that self-confidence. According to the new Protestant religions, an individual became religiously justified striving for worldly success with as much zeal as possible with the result people tended to accumulate personal wealth.
By the time Weber wrote this essay, he believed that, unfortunately, the religious values underpinning this Protestant ethic had largely been lost. He cited Benjamin Franklin who emphasized frugality, hard work and thrift for their own intrinsic worth, not for religious value.
Thank you Howard for the nudge which has forced me to set out my skepticism concerning social "science" in more rigorous terms. Let's look at another subject area that uses statistics and contrast them. I taught statistical thermodynamics for many years and although I've forgotten most of it, I still am aware of the general principles. Generally speaking, the goal in stat thermo is to derive macroscopic properties from microscopic properties via a statistical model of the process of interest. For instance, one might examine the dynamics of energy transfer in a gas and the statistics of collisions at various speeds and angles for some gas of a particular molecular weight. The overall microscopic energy transfer is then equated to the heat transfer which in turn is set equal to the macroscopic thermal conductivity times the temperature gradient. One can then solve the equations for the thermal conductivity, a macroscopic property, as a function of the microscopic constants of the molecules being examined (molecular weight, van der Waal constants, etc.). The statistical nature of the process disappears in the details. One finds out if the statistical model holds via experiments which check limiting cases.
ReplyDeleteIn social "science" there are flaws which make statistical methods almost useless. Besides the mathematical ones mentioned by Ira in his first article, there is the element of biased speculation. A theory in a social treatise is not provable or disprovable. Society is just too complex and uncontrollable. Statistical or population studies is just a method for trying to smooth out a few bumps in the road. A particular speculation may be brilliant and insightful, but it ain't science. Statistical data is used to support prejudice masquerading as scientific theory. The statistics do not tell us about macroscopic properties. Let's take an invented example of good social science in order to illustrate my point.
Suppose I'm a city manager and I care about changes in my city's population (a macroscopic variable). I theorize from my general observations of cities and anecdotal evidence, that population flux has a simple relationship to living conditions. I believe that people are driven to migrate because they can't bear some negative factor that exists in the place they live, not because they are attracted to another place. Each city is assigned a elaborated "misery factor" (invented by an assistant to LBJ and popularized by Jimmy Carter). The flow of people out of any city to any other city is proportional to the difference between misery factors, a threshold misery factor and some other stuff.
I can take historical data on migration and establish a few necessary constants and voila! I'm ready to test my theory against all the relevant cities in the US. The statistical model for the movement of individuals has disappeared into the final equation. If my statistical model is wrong, I will soon find out, because the results will not fit reality. That's a testable theory and science in action. Most of what we see in the social science is mere untestable speculation. In case you think I'm being mean to social scientists, let me add that I think that political "science" and most of psychology falls into the same realm. As in the infamous case of Joseph Ellis, we find too often that political scientists become contaminated by partisanship. (Is that related to the Dark Side of the Force?) With respect for all readers and contributors, but not authorities -Joel
"Science, in the broadest sense, refers to any system of knowledge which attempts to model objective reality. In a more restricted sense, science refers to a system of acquiring knowledge based on the scientific method, as well as to the organized body of knowledge gained through such research."
ReplyDeleteLong ago in the archives of this Blog I mentioned my dog figuring out that a newspaper is best carried from the middle rather than the end. I don't know if he could have explained why, or if he even realized he had done a series of experiments to model the reality of gravity, but he was doing science, and, IMHO, it was real science without the "scare quotes".
That said, I have to agree with Joel that certain branches of science are what I would call soft science (and I use that term non-pejoratively). WikiPedia summarizes it thus:
"Within the natural sciences, research which depends upon conjecture (sometimes called hypothesis), qualitative analysis of data (compared to quantitative analysis), or uncertain experimental results is sometimes derided as soft science. Examples are evolutionary psychology or meteorology. When soft science refers to a natural science, it is usually used pejoratively, mainly due to the term's association with social science, implying that a particular natural science topic described as 'soft' does not belong to the field of natural science."
By their very nature, fields of knowledge such as meteorology are based on rather indirect observations of extraordinarily complex interactions. As a result, it seems it always rains on the weatherman's picnic - of course that is because it is only news when that happens!
Yet, I would call meteorology a science. It is just not as precise in predictions as other natural sciences.
When we come to evolutionary psychology we are even further out on a limb because the motivations behind the theories and predictions are often tinged with personal biases and different world views and different attitudes towards religion.
Social science and political science are even further out on the limb - but it is not necessarily due to any lack of skill or honesty by the scientists who pursue those areas, but rather to the complexity of the situations they study and their inevitable philosophical implications.
Ira Glikstein
Just because some people make bad choices of observables does not invalidate the sophisticated techniques of statistical modeling. Complex systems are everywhere, so we have to try to understand them and predict or control them even if we are “out on a limb.”
ReplyDeleteAll scientific models of simple and complex systems advance by peer review and competitive testing. Statistics is a highly developed modeling technology that allows empirical comparison of models, including many checks of experimenter bias.
By what other method would Joel and Ira propose to make objective comparison of models of complex systems, as in genetics, evolution, medicine, economics, psychology, management, politics, and education?
Howard asks "By what other method would Joel and Ira propose to make objective comparison ...?
ReplyDeleteI can't speak for Joel, but I agree with Howard that sophisticated statistical analysis is the best *available* method.
However, as Howard also points out "...some people make bad choices of observables ..."
The problem is not the statistical tools, but the quality of the observables and the liklihood of inadvertent (or purposeful) bias in data selection. "Garbage in, garbage out!"
Howard, as Chairman of my PhD committee knows I used the "student-t" test to prove the hierarchical span of domains from RNA folding and written languages to the Egyptian army of 400BC to the modern US Army and corporations was within the range 5..9. In some of those domains I proved statistical confidence of 90% or 95%
I tried to meet the statistical requirements as to number of *independent* data points and so on and I am confident in my conclusions. However, another scientist might claim the fact humans can reliably sense only 5..9 categories tends to lead them to publish data that falls in that range. Perhaps I inadvertently selected data to fit my pre-ordained conclusions.
Nothing is wrong with the sophisticated techniques of statistical modeling. We just need to recognize that some domains of natural science are inherently more suitable for unbiased experiments and larger sets of data points. Other domains, for example where small groups of college students are the subjects and are asked emotionally-loaded questions by experimenters who definitely have their own political biases, are less reliable.
Howard, Joel, and I are doing what Howard called "peer review and competitive testing". Howard questioned Brooks's results because he is a WSJ writer trying to sell a book. Joel pointed to a website that questioned the study Howard linked to on the basis of the small number of subjects (only seven conservatives) and the fact they were college students. We each tested the results against our anecdotal experience with L- and C-minded people. If this was a simple physics issue, we could all agree on a relatively simple experiment that we could each conduct and get close results. Not so easy with political science!
Ira Glickstein
So the hard question is: Do you think we can attain objectivity, or are we stuck with unrecognizable (unconscious) biases? This would require total detachment or loss of ego. Then what would motivate us?
ReplyDeleteHoward asks "...are we stuck with unrecognizable (unconscious) biases?"
ReplyDeleteYes IMHO. If a person doesn't have an opinion (another word for bias) about important and controversial issues of the day he or she is either mentally challenged or simply not paying attention. The key is to try to overcome biases using things like the scientific method, where applicable, or other more or less effective techniques for objectifying the subjective.
For example, in my System Engineering classes we do what are called "trade studies" where we evaluate different design approaches or different items of hardware and try to pick the best overall item. (In ordinary life we do the same when we choose a car or TV set or PC or where we will retire, etc.)
I've designed a tool that runs on the Excel spreadsheet that somewhat objectifies the process.
http://polaris.umuc.edu/~iglickst/mswe603/Decision3.2%20BLANK.xls. (Free, anyone can download and use it.)
It is particularly effective when an engineering team must work together to make the decision (or a husband and wife :^) because it forces each participant to state and quantify the importance of each criteria (price, speed, style, reliability, safety, maintenance, ...). Instead of shouting "I want the green one!" against "You cheap idiot, the blue one is better" you all have to agree (or at least agree to disagree) on the weight given for cost vs performance vs other factors.
I tell my students an engineer is not ready to do a trade study unless he or she already knows the answer! Try to be objective in inputing the data and if the trade study confirms your opinion, go with it. If it does not, check the data and assumptions for error. If you can't find an error, you have learned something - your initial opinion was wrong!
Ira Glickstein
Howard said, "All scientific models of simple and complex systems advance by peer review and competitive testing. Statistics is a highly developed modeling technology that allows empirical comparison of models, including many checks of experimenter bias."
ReplyDeleteJoel responds: I respectfully disagree with Howard concerning the above. The reality is that experimenter bias often goes undetected or is passed on, if it corresponds to that of the reviewer or editorial board. The situation had gotten so out of hand that in 1999 a committee of the American Statistical Association had to formulate guidlines to try to correct the situation. The Ethical Guidlines say in part; All statistical practitioners are obliged to conduct their professional activities with responsible attention to:
1. The social value of their work and the consequences of how well or poorly it is performed. This includes respect for the life, liberty, dignity, and property of other people.
2. The avoidance of any tendency to slant statistical work toward predetermined outcomes. (It is acceptable to advocate a position; it is not acceptable to misapply statistical methods in doing so.)
3. Statistics as a science. (As in any science, understanding evolves. Statisticians have a body of established knowledge but also many unresolved issues that deserve frank discussion.)
4. The maintenance and upgrading of competence in their work.
5. Adherence to all applicable laws and regulations, as well as applicable international covenants, while also seeking to change any of those that are ethically inappropriate.
6. Preservation of data archives in a manner consistent with responsible protection of the safety and confidentiality of any human beings and organizations involved.
There are many other surprising and interests caveats in this document. The complete ethical standards can be found at http://www.amstat.org/profession/index.cfm?fuseaction=ethicalstatistics
A good part of the disinformation of statistical studies must be laid at the feet of the media. Amodio complains in his rebuttal that there would not have been such an adverse reaction to his study if the newspapers had not gobbled it up and distorted it as a hit piece against conservatives. (It might be better said that his methods would not have come under such scrutiny.) It is an unfortunate fact of life that sensationalism sells better than truth. With respect -Joel
Joel, thanks for the reference on statistical ethics. I read the entire document.
ReplyDeleteI agree that my ideal of unbiased statistics is not attainable, but the purpose of my post was the question: By what method other than statistics, imperfect as it is, would you propose to make objective comparison of models?
Howard said:
ReplyDeleteBy what other method would Joel and Ira propose to make objective comparison of models of complex systems, as in genetics, evolution, medicine, economics, psychology, management, politics, and education?
Joel responds: I've tried for several days to come up with a response to your challenge. It's a very big question and I've become mired in lots of material about statistical frauds. I'll try again in a few days (after taxes). In the meantime, have you ever heard of "Benfords Law." I can't figure out whether or not this thing is a giant hoax by the statistical in-crowd or a reality for a limited domain of data. With respect -Joel
I’d never heard of Benford’s Law. I looked it up on Wiki, and still don’t know what to make of it. Could it have to do with how we quantify how our observing senses detect differences? I think pressure, light, and sound detection is logarithmic.
ReplyDeleteI had been thinking about my own question about the use of statistics, and I decided the answers range from the obvious to the obscure depending on how long you want to think about it.
The obvious answer I first had in mind is that there are all degrees of confidence in our knowledge, from the certainty of the value of pi, or that the sun will rise, to the uncertainty of throwing a die or radioactive decay. So, I would say we have to use statistical comparisons whenever we are ignorant of details. In other words, the obvious answer to my question is that mathematical statistical comparisons of models are useful only to the degree that aspects of the model are empirically statistical (i.e., unknown).
But then, Wiki reminded me, statistics began with games of chance and random numbers where ignorance is assumed to be total. The obscure problem is that this assumption is never true except as an abstract ideal. In any empirical model, or even in our wildest imagination, we have inherited and learned a vast amount of knowledge that influences how we think about chance events and probabilities.
For example, Aristotle, Laplace, Einstein, and Ira do not believe randomness appears except as a result of our ignorance of detailed causes. They believe that every event must have a cause. Other great minds nowadays believe that all detailed events are random and that determinism appears only as the result of good statistics. These two incompatible points of view are models of the world created in our brains, and our choice of which one we favor depends on how our brains work and our experiences. They are two metaphysical beliefs for which no one has found a definitive empirical test. Probability theory is the basis of all statistics, and these two metaphysical views disagree on the foundation of probability theory, specifically, how to define equiprobable events.
The incompatibility gets worse when the models are formalized logically and mathematically. In physics there has always been a problem relating the microscopic laws to statistical mechanics. Max Planck concluded, "For it is clear to everybody that there must be an unfathomable gulf between a probability, however small, and an absolute impossibility . . . Thus [time-symmetric] dynamical laws and [irreversible] statistical laws cannot be regarded as interrelated.”
Nothing is simple if you think too much!
Thanks Joel and Howard for bringing up Benford's Law (which he conceived in 1938, the year *I* was conceived - I was born early in 1939).
ReplyDeleteBenford's Law is discussed in Wolfram's Mathworld and WikiPedia.
Benford looked at many lists of data and concluded the numeral "1" was likely to be found as the first digit about 30% of the time, with the other digits less likely: "2" appears about 18% of the time and poor "9" less than 5%. At first glance, you'd think each numeral from 1..9 would appear about equally, around 11% of the time.
It is easy to see why it works for the street addresses of the first 342 people listed in the book "American Men of Science" in 1938. When street addresses were assigned from "1" to however many houses were on the street, *every* street with nine or fewer houses had a "1", most had a "2", but few had a "9". In addition, *every* street with between ten and a hundred houses also had a "10", and most had an "11", "12", "13", "14", "15", "16", "17", "18" and "19". Many had houses in the range "20..29", but fewer with "30..39" and so on to very few with "90..99". Similarly with streets with more houses.
Does this work with the new mile-based numbering systems used in Marion County FL and many other areas? I'm at 7073 SE 173. That number means I am about 7.073 miles east of the center of our County and the start of my street is 17.3 miles south of the center. I think it would work!
Say all counties (at least in FL) are between 10 and 100 miles from the center to any edge. Thus, all will have house and street numbers indicating 1 mile from center as well as 10, 11, 12, 13, 14, 15, 16, 17, 18, and 19. Most will have 20..29, fewer 30..39, and very few 90..99.
You could probably figure out why it also works for other diverse groups of data, "including electricity bills,... stock prices, population numbers, death rates, lengths of rivers, physical and mathematical constants, and processes described by power laws (which are very common in nature)." [Wikipedia]
An interesting application of Benford's law is discussed in NY Times,1998. There are many numbers in an income tax return and, if they are honest, they should follow Benford's Law. However, when a tax cheat makes up those numbers, they may not follow that Law! The IRS could use that simple test to identify returns for further investigation.
On a related statistics vs cheating note, the NY Times story cited above says a math Prof. asks his students to flip a fair coin 200 times and submit the list of "head" and "tail" results. He tells them they may, if they wish, cheat and fake the list. He detects most of the fakes by looking for a run of six or more "heads" or "tails". If a list lacks such a long run, it is most likely fake!
Ira Glickstein
Hi Ira, Joel, Howard,
ReplyDeleteI have been following your articles on the L/C mind discussions going back to 2007. Your examples on L/C minds, have definitely given the readers a lot to think about. Your diversity of examples gave me a lot to mull over. As I thought about each example or situation you discussed, I would try to reason out whether I leaned toward a liberal or conservative mind. Please note that I used the word “leaned”. That word was carefully chosen because as I examined my intellect, emotions, and knowledge to respond to your examples, I found myself leaning to and fro in the middle, not quite L mind, not quite C mind. Note: I could understand why the L Minds responded, and I could understand why the C Minds responded. Different or opposing information gives real dimension to knowledge! Reading the statement by Joel, when he said, “My objective is to get people to think a little harder about the point of view of others i.e. to show respect. My definition of "to respect" is to act as though there is a possibility that you are wrong and your intellectual adversary is right, no matter how totally farfetched that might seem on the surface”. This blog is proof of how to get the “readers” to think harder about the point of view of others! Thanks to Ira. With respect as always, Deardra
TVPClub is way "ahead of the curve" thanks to Joel who brought up BENFORD'S LAW last April in this Comment.
ReplyDeleteAccording to analysis using Benford's Law posted today it appears that some of NASA's GISS data for Global Warming has been hand edited!
As Howard and I (Ira) discussed on this Blog, Benford showed that numbers have characteristic distributions of digits, where the first digit is more likely to be a low number than a higher number and the last digit is likely to show an equal distribution. Raw temperature data should follow Benford's Law unless some human has tampered with it.
The linked Blog discusses some possibly innocent explanations. In some cases converting from Fahrenheit to Celsius may introduce distortions that cause false positives from the Benford Law test.
This is all very preliminary and I will follow this issue and keep you all informed.
Ira Glickstein