Humanity has an inherent need to know: how things work, about our origins, what is our purpose. It was easy to ascribe thunder to a god, rain to another and so on, and even create an hierarchy among them, complete with conflicts and histories of same. Mythology for sure. But a few brave souls dug deeper. Euclid developed geometry, Ptolemy understood his circles and developed a means to "explain" the motion of the planets "around" central earth. Copernicus observed inaccuracies and went on to propose a solar-centered method. The world has never been the same since--thanks to Galileo, Kepler, Newton, Einstein. Unlike astronomy, biology had to wait for Darwin, Mendel, Watson and Crick to parallel in biology the progression from Euclid to Einstein.
The basic nuts and bolts provided by Jacob Bronowski in a BBC / Time-Life documentary in 1973 puts science in a human perspective:
|"We are always at the brink of the known; we always feel forward for what is to be hoped. Every judgement in science stands on the edge or error, and is personal. Science is a tribute to what we can know although we are fallible."|
Truer words were never spoken. So let us look behind the scenes for what we mean by science. First of all it is a valid description of how things work within measurement error. The smaller the error, the more solid the science.
- Observation and Measurement: Using generally accepted measurement systems. Our data and conclusions and generalizations therefrom are unambiguous.
- Repeatability: Unless something is repeatable, it is not predictable, and therefore is not science.
- Economy: The simplest formulation that yields to most and most accurate results qualifies as new science. It is elegant.
- Consilience: Each new scientific finding must be consistent with other branches of science. If not, both cannot be true.
- Heuristics: Deepened insights give rise to new questions, at least in the current stage of development of science. New questions in turn lead to new research, and new science.
In practice, scientific theories must be testable and falsifiable. Otherwise they are pseudoscience. Pseudoscience may have a place in the minds of many with certain psychological needs. Pseudoscience may become science, but far more typically, research overturns pseudoscience when measurements means become sufficiently precise and repeatable. Negative results are important and are part of the game of falsifiability. To find floaters, one must wade through a lot of sinkers.
Research first of all involves descriptive codification of whatever subject. It is at this point that hypotheses are created as to the mechanism. Some theoretical basis must underlie each observation. Each effect observed must obey a law, even if it must be codified as problematic--Heisenberg's uncertainty principle for example. Once a situation is known with accuracy, relationships can be refined in terms of what causes what. Observe, hypothesize, test and analyze: (OHTA) is shorthand for the researcher.
This page is provided to illuminate some of these issues. Richard Feynman, scientist "extraordinair" of the 20th Century, captured the essence of science in his book The Meaning of it All thusly:
Science has three parts:
- Facts or bodies of knowledge,
- Methods or processes that we use to establish those facts, and
- Applications of science, that is to say, technology.
Observational science is the kind most of us are familiar with, like tracking shadow lengths daily for a year to determine the sun's apparent motion relative to earth. It was natural for one early observer, Ptolemy, to assume the earth was fixed and the heavens moved around it. It fit ordinary observations and significantly, it fed the human ego.
Observational science includes compendia of the plant and animal species, descriptions of an outer solar system moon, Titan, where it is so cold methane rains like water on earth to form a moonscape similar to that on earth, how plants grow, the details of how metals deform and so on.
Thus it was that Copernicus found improved explanations over the Ptolemy system which Kepler improved upon, and Newton explained. Newtons theory, accurate in his times, failed over extended times in tracking the orbit of Mercury. Michaelson and Morely observed that the speed of light is an absolute, Hendrik Lorentz and others provided a mathematical structure which Einstein used to show is part of a more accurate description of how gravity really works. Einstein's theory describes gravity as a variable geometry such that space is inherently curved and time is relative to the observer. No departures from his theory have yet been found; its accuracy and apparent precision are immense. Nevertheless, it is still being tested in ever more sensitive ways.
Experimental science usually follows observational science. In modern times, establishing cause-and-effect relationships have become the name of the game for experimental science. Proving cause and effect is high science, and not at all as easy as it sounds. Experimental science has given us the technologies enabling modern communications, travel and the space age. It has also given us modern medicine and extended lifespans. Experimental science involves controlling conditions to be studied in search of true cause and effect. Science 101 is first of all about the scientific method; designing experiments enabling the study of observable, empirical and measurable evidence gathered and subjected to logic and specific principles of reasoning. Scientific methodology embraces experimental design involving at least one subject (something tangible with a measurable feature) and at least one controlled variable (such as temperature, humidity, composition, organization and the like}. At least two quantities must arise from a valid experiment, the magnitude of any effect and its attending error. Knowing the error of measurement, it is possible to estimate the likelihood that the effect observed is false. This probability is known as the significance of the effect(s). If an experiment is repeatable with a likelihood of error less than about one in a 3.5 million, its results become part of scientific lore. Science is falsifiable. Just as Einstein overturned Newton's theory, so could his theory be overthrown by yet more precise measurements. 2012 saw the emergence of a particle in physics in the high energy range expected for the Higgs boson. But the announcement on 4 July 2012, was simply that the evidence was strong enough to publicize but not yet complete enough to conclude that it fits the standard model; it could require new physics. This is how the high drama of Bronowski plays out.
Inferential science follows experimental science. The question is: What do our experimental observations tell us? Is our working hypothesis correct? If not what is? Causality is inferred and a theory is born that can then be tested and proved within experimental limits or falsified. From the time of Ptolemy, about 1500 years elapsed before Copernicus improved upon his earth-centered belief. It was another century before Bruno, Galileo, Tycho, Kepler, and Newton persisted to the point where a law of nature was found -- a law that had predictive power and was far more accurate and inclusive than the earth-centric scheme. Bruno and Galileo ran into immediate trouble with an Inquisition pope; Bruno was burned at the stake. Galileo escaped with his life, but suffered house arrest during his last years. Human ego, driven by fear of the unknown, invoked command and control, sent Bruno to an awful, fiery death -- by a man of God!
This problem is still with us today. Our own government, until 2009, withheld vital scientific information from its female citizens for reasons of religion. Science states what is. It is only in the abstract that morality enters. Although animals can behave altruistically, morality is a human invention related to our ability to think and codify in abstract terms, like: Our family will survive the winter if we can store enough food: or, if we nurture that orphan, s/he will add to our numbers, not perish. If we refrain from killing our brother, we will have his hands and head when we hunt, and so on.
There is an important limitation of observation in science: an association, even a strong and reliable correlation, is not proof of cause and effect. Neither can an association become a theory (the simplest possible explanation) until it has been demonstrated repeatedly to high and well-defined levels of certainty with an underlying reason for the correlation. The cause and effect issue relies on its consilience, how well it fits the unity of all knowledge. Since the body of knowledge is ever expanding in our times, opportunity for experiments expand apace. In fact, each expansion is exponential with time, at least up to our age.
The well-known laws of mechanics and thermodynamics qualify as science. So do the laws of chemistry, genetics, and evolution.
Today, observational science is alive and well, astronomy and geology being two important branches. They too increasingly depend on theoretical relationships proven to high precision. As measurement accuracy and precision progress, so do our insights into how nature operates.
Experimental science, where the subject and testing conditions are controlled by the scientist, has become the mainstream of new discoveries. When discoveries fit into science as a whole, they become compelling. That physics begets chemistry became compelling with the advent of quantum mechanics that rationalized why chemical reactions occur. Physics and chemistry together give rise to astronomy and geology. And chemistry begets biology, as became vividly evident with the discovery of DNA and its template, RNA. Reproduction in all living matter employs one or both of these chemicals, which in turn are constructed of just 20 amino acids and a punctuation mark.
The previous paragraph bears on a present-day, US-based, movement to deny the validity of biology, which is to deny the whole of science. For if biology is not true, then the other branches cannot be either. Since the others are undeniably true, why has such a conflict arisen? This answer resides in one's individual psychology and background. For example, misunderstandings about science arise when a person:
- is untrained and thus unaware of what science is;
- fears the unknown, or the unknowable, and cannot move his/her thinking with the times, unable to adjust meaning as new events and discoveries preclude old explanations--falsified. Such people may prefer consistency even when they know, logically, that it is wrong;
- rationalizes purpose and meaning for existence that is not compatible with science, evolution in particular, to reach a comfortable but unhealthy state of denial lest conflicts arise in the mind.
The last point needs a bit of discussion.
Denial is a defense mechanism that protects its owner from fear in some way. Before Galileo, Ptolemy's system worked passably well for humanity as a whole. The problem was, that denial became collectively unconscious to the point where individuals and societies at large, were unaware of the denial. In this way, what started out as a healthy coping mechanism by humankind, became discordant with reality beginning with Copernicus, continuing through Bruno, Galileo, Newton, Darwin , Einstein, and their many cohorts continues down to our times. Mass denial affected the monotheists especially. It still does. Schooling in the literalness of Genesis gives people a needed consistency and feeling of safety. Throughout history, it has been very hard, often impossible, to give it up--even though its literalness, like Santa Claus, cannot be true.
Nevertheless, when new discoveries can be repeated, when they serve a purpose, when they work and prove useful in daily affairs, they become accepted as truths by broader audiences, even the most religious ones. When the fruits of science (logos), conflicts with dogma (mythos), problems naturally arise. Ironically, the nation leading in scientific discovery, also counts among its citizens the most vocal in rejecting some of the most certain discoveries when they appear to threaten their belief system or power over others. That issue is dealt with elsewhere. Our purpose here is to remove the mystery of science, especially for those who might be contemplating careers in a discipline that has become central to modern existence,
There is a further aspect: Scientific observation and experiment overlap with the human factor, notably in medicine, where everything may be controlled with precision except the human factor which arises from our very real genetic, experiential, and temperamental variations. One important result is scatter in the data. This makes small effects difficult to see, a special problem dogging behavioral science.
To make sure-footed moves towards peace we need to be sure of our ground. Bad science is worse than no science. Some of the issues in avoiding bad science are developed next. What separates good science from bad, is:
- Controlling all the variables in terms of:
- Measurement accuracy and precision,
- Sample size representative and valid,
- Random sampling methodology,
- Effective experimental designs and matrices,
- Double blind procedures in health research,
- Using the appropriate distribution function for analyses,
- Applying the proper inferential statistical procedures,
- Analyzing actual data and reporting valid levels of effect, their powers (discussed below) and error bars, and
- Interpreting results in terms useful and meaningful to humanity.
Good experimental science will be repeatable by others who impose the same experimental conditions and perform the experiments in the same way. Good observational science will be repeatable in terms of observations and descriptive parameters. Good science results from observations when the observations made can be used to predict new events within the precision of observations. Good science provides predictions with an error bar associated with each parameter determined and a "residual error" reflecting only the limits of measurement accuracy. However. just because two variables correlate, does not mean one causes the other. Each may be caused by a third variable not considered. The root cause of observations is the deeper and truer realm of science. Science involves logical connections underlying a mechanism by which phenomena occur.
Best science avoids bias introduced by the experimenter, sample sequence, and time of day as well as chemical, mechanical, physical or numerous other human or natural factors. Best science removes the human factor completely by automating sampling, controls and measurements to the fullest extent possible. Physics, chemistry, astronomy, geology and biology all go by these rules and results of one discipline must be consistent with the results of the others. And so they are.
But scientists must always be on guard against drawing false conclusions. They must also refine their logic as they refine their measurements. They must always look for exceptional results in order to refine their theories on a sound basis. The advent of DNA is a modern example of this process. It cross checks and corrects taxonomy, and quantifies the degrees of relatedness among and within all species.
For another example, an experimenter, and the public as well, can be led astray, especially in survey-type studies where dozens of factors are being searched for. Public health studies are especially troublesome in this regard because it is rarely possible to reduce the experimental error (also known as the alpha error) to the point where a single experiment is so decisive that repeating it becomes formality. The alpha error is the probability that any given result would happen by chance alone. It has two forms: standard deviation, and standard error--which is the standard deviation of the mean or average of the data. The former is useful when comparing an individual result with the average, the latter is used when comparing averages between two or more groups of data, tallness in people for example.
In either case, alpha represents the probability that any given result occurs by chance alone. Scientists set alpha to levels appropriate for the situation. 0.05, or 5%, is a common level when exploring data for possible relationships, but it can be set at any other level. The level set is said to be the "significance" that will nominally determine whether a cause-and-effect relationship might exist. The level for new science is 0.00000035 or five sigma one tail.Alpha cannot prove a relationship, only scientific experiments with a confirming theory can do that. Then, the proper test must be selected to provide a best estimate of probability.
There are practical problems too often overlooked in the use of alpha. For example, if we set our alpha error to say 5%, and survey 100 potential causes for a given result, we will find five or so results on average that appear to be significant (at 0.05} when in fact five will typically occur by chance alone. To be right 95% of the time means being wrong 5% of the time, even when there is no effect whatsoever. However, if you find 20 "significant" responses in your 100 tests, then you can have some confidence that 15 or so of them will turn out to be real while five or so are not. For this reason alone, experiments must be repeated until the joint probabilities of error among experiments for the real effects approach zero--if the experimental conditions for each experiment are the same. This concept embraces the concept of power, the math of which is of no concern here. Power is gained by large samples and independent experiments.
The valid sampling and the "test-wise" alpha error limitations are rarely accounted for in the so-called meta-analyses where experimental results by various authors are combined to give the results more "power." It should be obvious that the 5% rule, being wrong 5% of the time, places "significant" constraints on any conclusion from a meta-analysis. How does the meta-analyst weed out the meaningless 5% claims? How many were statistical flukes? Were all the experiments done in the same way, with valid-and-representative samples from the same populations? The short answer is that as many a half such meta-analyses do not pass muster. This problem is especially acute in the medical and social sciences.
Beyond that, data can be manipulated during the analysis to come to a result opposite to what the data really say. For more on this see: "Odds Are, It's Wrong."
Knowing how to calculate alpha does not guarantee its proper use. It can take many forms mathematically. And its accompanying distribution function, like a histogram for example, must be known.
Sometimes, an experimenter cannot afford the added expense or time needed for such proof. Aggravating that condition is a general lack of awareness on the part of officialdom, bureaucracies and the public about just what alpha means. Too often, the experimenters are little better off. And all too often, half-baked science gets into the media, even sold over-the-counter. These problems have resulted in not only bad science, but ineffective, even bad, medications reaching public use.
So how sure can we be? Repeating a study that produced an alpha of 0.05, or one in twenty, that reaches alpha = 0.05 a second time improves the odds for an effect being real to one in 400. If this is still not good enough, we can run it again. If the third trial produces an alpha = 0.05 yet again, the odds that we are looking at a chance event improve to one in 8000. This is how scientists attach numbers to results people may feel intuitively. Nothing is ever absolutely certain, but it can still be very useful.
So how sure must we be? Casino owners in Las Vegas know the odds exactly and shade the pay out to ensure themselves a profit. Why should we do less with issues of health and peace? We must do what we can afford, and do it well with known levels of confidence.
This brings us to a four-way intersection when testing an hypothesis to a selected alpha:
- accepting a false positive
- accepting a false negative
- accepting a true positive
- accepting a true negative
Two of these are wins, one becomes theory and adds to science. The other two have the potential to become myth and too often do. Setting alpha takes some discipline and experience to do right. Biased experimenters have been known to fudge on this point. Otherwise good experimenters have failed when they did not take "experiment-wise error" into account. "Surfing variables" often leads to this error. A rule of thumb here is that if the number of variables look at is n, then no conclusion can be drawn safely unless 1/n is at least a few multiples smaller than the standard error or sigma. Even then, the results need to be tested by independent scientists and confirmed by a theory of cause and effect that explains results found.
Even in our scientific age, junk science arises every day. True science is testable; that is it can be tested. Junk science fails upon rigorous testing. But the fact is that many people come to believe in the junk and hold fast to their beliefs. They are mostly innocents if the society doing the molding is led by authoritarians or sociopaths who put themselves ahead of their brethren, no matter how numerous.
Settling for political platitudes associated with special interests will only invite continued violence and poor preparations for natural disasters.
The all too-human emotion, hubris, has discredited many otherwise notable scientific careers. In the 20th century, a beloved man, a winner of two Nobel prizes, fell into the trap of hubris in his later years. He pursued vitamin C as a cure for the common cold, finding but ignoring the very real effect it has on severity. Because of his hubris, he earlier missed discovering DNA by an eyelash. Of course in the end, Linus Pauling really proved he too was human. By any measure, he was one of the greatest and most-remarkable people who ever lived. He was living evidence that it is not the mistakes we make, but how we handle them that counts. Pauling never flinched; he kept stepping up to the bat.
The above is only for the flavor. If we want to do serious research or be able to interpret works of others for ourselves, we will need to know not only how to calculate alpha but also what it means in any particular situation. Using alpha properly to its full extent can be tricky; even experts can err. Alpha has four companions we also need to know about if we want to interpret our experimental results properly--number of samples, variance, beta and power. Then there is design of experiment. Their detailed renditions are beyond our scope here.
One thing is sure from written history: the human condition only began to improve with the advent of the scientific age: When facts, and the science arising therefrom, became more useful and powerful than fancy, myth, or faith.
Recognizing and interpreting facts requires an open mind, like those of Bruno, Galileo, Kepler and Newton. And that is why a near universal Internal Locus of Control in each generation is vital if we are to make effective progress against terror and other self-inflicted human ills. Of course, there are numerous economic and political inequalities that need to be corrected. And justice must prevail.
A person can be a highly competent scientist without being creative. But the great ones are. To Bruno, Galileo, Kepler and Newton, we can add, Pasteur, Darwin, Mendel, Einstein, Boltzmann, Bohr, Heisenberg, Freud, Feynman, Watson and Crick, and a great many others. Conversely, Creativity does not ensure a career just in science; it is more global than that. Creativity is a signature of the animal kingdom. Humanity sets itself apart in that creative events often rely on abstract thinking and not a little logic and/or mathematics.
In democracies, change can only come about from concerted action by informed voters able to think for themselves. Natural History highlights some of the problems, as does the Authoritarian Personality. For the core personality beneath violence at the organized level, see: Sociopath Next Door by Martha Stout. For how extremists operate and co-opt Authoritarian systems see also:
- Adorno - Authoritarian Personality, AP,
- Altemyer - AP in Politics,
- Burger - Confirmed Milgram's Study,
- Dean - AP and the NEOCONs,
- Frank - Bush on the Couch, Freudian,
- Hare - Psychopaths,
- Milgram - Americans are blindly Obedient,
- Radicalizing a Neighbor - Israel in Gaza,
- Roadmap via Links - Roadtopeace.org and External,
- Stern - Interviewed terrors,
- Stout - The Sociopath Next door,
- Wakefield - The Hijacking of Jesus, and
- Zimbardo - The Lucifer Effect.
Adorno was first to popularize the AP. Milgram, Burger and Zimbardo highlight American behaviors. Each study was scientifically sound. Together, they arrived at sound conclusions. Altemeyer reduced the complex AP to a simpler construct consistent with Evolution. Altemeyer, along with Hare and Dean, shows the dangers to society when obedient APs, led by extreme psychopathic APs, take over a nation. Stern, alone among these references reported the actual thinking of terrorists and their leaders. Frank, Babiak and Hare, and Stout demonstrate the extreme AP psychology from three different viewpoints. Stern and Wakefield address the effects of monotheism.
How all this applies to social systems is not exactly self-evident. The devil is in the details. The scientific methodology for assessing social factors lagged those for mechanics by a couple of centuries and are still being refined. Adolphe Quetelet, a Belgian scientist and statistician, was first to begin making sense out of this most difficult subject. His story, and that of Karl Pearson are worth close study for they provide bases for optimism that a solution to social factors exists: "Statistics On The Table." In each case, it is ironic that the difficulties encountered by Quetelet and Pearson were more with people than with their mathematical approaches, though only the latter were new. Their stories illustrate the frailty of the human personality, and are therefore of central interest to RoadtoPeace.
If you are interested in further rendition of the most exciting development of the human mind in history, one place to start is with the Scientific Method itself. Wickipedia discusses most of the common buzz-words and concepts that go with the discipline.
Experimental design is both the strength of great research and the Achilles' heel of poor research. Rigorous interpretation also goes with great research, especially in life science. See: Experimental Design for the Life Sciences for an introduction.
For a comprehensive overview, visit Science for All Americans. If your interest is in science as a career, see: Science Careers. The Center for Science Diplomacy has a most critical mission:
|"The Center is guided by the over-arching goal of using science and scientific cooperation to promote international understanding and prosperity. It approaches this goal by providing a forum for scientists, policy analysts, and policy-makers through which they can share information and explore collaborative opportunities. We are particularly interested in identifying opportunities for science diplomacy to serve as a catalyst between societies where official relations might be limited, and to strengthen civil society interactions through partnerships in science and technology."|
MACHINE TRANSLATE: This or any other page: Download Babylon
Posted by RoadToPeace on Monday, September 12, 2005.