Sunday, April 21, 2013

(In Search of...) VARIABLES

In Search of Understanding Variables

What are variables” Are they some form of molecules, differing thoughts, concepts, ideas, behaviors, emotions, feelings, personalities or people? I guess the answer would be yes to all. Now wouldn’t it? Each one of us aspires to be different. We want to achieve, plan, create, modify, build, manufacture, harness or sometime merely express our inner most feelings. The variability in our expression and desires changes with the nuance of the day, the shades of reflections, the brightness of the sun or the silvery calm of the moonlight. Each moment bestows it’s own separate and distinct milieu that drenches thoughts and changes passion.

There are many variables, for instance; heat, volume, filters, position etc. each has a lasting value and impact on the ultimate result. And don't forget Lorentz "strange Attractor" in this wild and funky world. Yes, Indeed we are surrounded by variables. So when, some list one or two and forget the others without reason, I say bolder dash!

So how in the world can we as individuals, groups or organizations cater to the needs of the individual? How so?

In the days past, it was easy, there was little to go around and what was created was utilized. Today the world is awash with variability. There is literally something for everyone’s desire. You don’t portrait, we have a landscape for you. You don’t like desktops, we have a laptop for you and if you are a curmudgeon about laptops, why then we have a tablet for you. Every color of the rainbow and those other colors conspired by humans in between, is represented in clothes and objects. And yet with all this mass of commerce there is still a need for selective difference. That tiny wisp of individuality.

Big Data is here to fill in some of the salesman shoes that used to walk up to every door. Ah the computers are busy accumulating this information in large swaths of binary codes and storing them in large servers that hum cooled with the needed refrigeration to keep them from liquefying under enormous generated heat or worse through that heat, transmuting into a different frequency of the signal. Thus far it is the human that sits and uses his or her analytical smarts to decipher messages from these non-linear swarms. It is the analytical skills filled with Boolean Logic that determines the IFs that lead to the THENs. But soon we in our ever logical mind will be abrogating that decision-making process to the whir and whiz of spinning or solid state hard drives.

But lately Artificial Intelligence also a product of, you guessed it; human intellect is making its overtures. AI as it is called is slowly transforming our life.

It is!

You want some real life example? How about the driver-less cars zooming around the Google complex and soon to a neighborhood near you? Or, and this is near to the human Achilles Heal, the ubiquitously well known 16-terabyte phenom housed in New York, affectionately called “Watson?”  This machine is a giant repository of resource. But relegating the soft, yielding, compliant, argumentative, rebellious mind of the human to the even, linear, programmed, multiprocessing, cold, calculating 1 and 0s, residing within the set parameter machine, bodes a difference in the fortunes of the future is the difference between one that is naturally created through trial and error or the other crafted by the mechanized “soul” of a cold hard entity of yes and no and only calculated probabilistic in-between(s). One can only be reminded of the echoes from the past, “Dave, I am afraid, I can’t do that.”

You see, transformation in a society is a slow ooze of intellect, baked in the kiln of understanding, hardened to the cold biting air of the real world and then utilized. This is not meant to be pejorative by any means. There is a delightful repository of data that resides in Watson’s terabytes, indeed there is! But there is also the self-wielding algorithm couched by the human mind, the bromide, which would confuse Confucius himself by its selective bias. In all it is the number and kind of variables one uses to create such self-replicating chains of evidentiary information. It is in the intersection of the variables that real truth exists. Can we then know all the variables? Yes, If you can know the mind and body of the 7 billion people! Other than that we are playing with the play-dough of uncertainty and probability.

So what is a variable? It is a logical set of attributes. It may be dependent and thereby change as a result of other values within the system or strictly independent, messianic, authoritative and deliberate. But these nuances change the meaning of the meaning.

You might have heard the term “Multivariate” analysis! This implies that in doing the analysis many variables were considered so that the results were robust and not likely to be confounded by artifact. But just naming “multivariate” analysis does not release the chain of burden on the author nor the reader. For in that term the variables so chosen have to be specified and it is up to the reader to determine if from the reader’s perspective the utilized variables are enough to warrant the validity of the results. However no matter the result they will always be bounded by the confines of the "Confidence Intervals" called "CI." This CI has limits to about 95% in most scientific and medical literature because a probability is never the absolute reality and therefore suggests in a polite manner that outliers outside the confines of this 95% boundary do exist! For instance, I came across a well-referenced scientific paper that suggested that the benefits of a drug were superior to the standard treatment in a particular cancer. Okay, that is good. But reading further through the jungle of the graphs and the Tables and Boxed algorithms, I realized that the variables did not include ethnicity breakdown within the trial methodology. Now that would seem nit-picking wouldn’t it? But dare I say that there was a relevant piece of information about the disease suggesting that the ethnicity had a major impact on the survival of patients. So if one were to artificially bunch the ethnic details on one “Arm” of the study, the results would pleasingly benefit the "other" treatment arm, now wouldn’t it?

Another “brilliant” study suggested a human anatomical difference suggests the risk of a certain “disease/illness.” But here the variable used was a questionnaire handed over to the subject to see if he thought there was an apparent discrepancy between the anatomical regions. These subjects were everyday lay people and some with known medical illnesses and others without any. Given the filled in forms by the two cohorts of with and without the specified medical illness were cross tabulated with the subjectively-filled forms of anatomical differences. And voila there was a correlation determined and then using the probability function, a risk was derived at and the paper was published in a “peer-reviewed-high-impact” journal. An utter travesty, don’t you think?

So you say, “Variable, What Variable?” Slow down Watson, there is more than meets the eye. Be careful in your variable evaluation, in your analysis and thus in your thoughts for future actions. And you know what the “experts” dearly love to avoid? The “confounding variables!” These confounding variables create substantial aberration in results and thus can negate the premise being expounded, exploited, expressed or otherwise being shoved down the throats. Oops! Too strong eh? I am reminded of the Challenger disaster where the cold rings were never cross-tabulated and it took a Feynman to figure it out (watch below).

You see from a mathematical perspective, the additive values of the variables change the downstream results. They do. The formula is 
Here “i” is the summation variable that calls upon each of the integers and “n” is a non-varying parameter.

In Physical phenomenon, the Institute of Physics defines investigations under this light, “When investigating physical phenomena, there are two strands to thinking about variables: the difference between independent, dependent and control variables; 
the intrinsic nature of the variables, i.e. continuous, discrete, or categorical variables.”

Countering physical phenomena, Algebra however defines Variables as, “A Variable is a symbol for a number we don't know yet.”

However in medicine variables loom large (from California State University, Fresno),

Demographics – Examples: Age, sex, race, religion, income, etc.
Physical Characteristics – Examples: Height, weight, etc.
Physical and Mental Health – These concern the presence (vs. absence) of physical and mental disorders, along with their symptoms.  Examples: medical diagnosis, psychiatric diagnosis, number of headaches per week, number of panic attacks per week, etc.
Personal History – These concern experiences that people have had.  Examples: number of lifetime sexual partners,
Individual Differences – These include standard personality traits (e.g., the Big Five), along with other relatively stable psychological characteristics.  Examples: Extroversion, intelligence, financial responsibility, self-esteem, etc.
Beliefs and Attitudes – This category also includes knowledge, opinions, and judgments.  Examples: attitude toward divorce, opinion of the President’s job performance, beliefs about the seriousness of the AIDS epidemic, etc.
Affective Variables – These include emotions, moods, and feelings about the goodness or badness of a stimulus.  Examples: depression, subjective well being (i.e., happiness), anger, how much one “likes” a photograph.
Performance Variables – This refers to people’s performance on all sorts of physical and cognitive tasks.  Examples: number of points on a final exam, number of errors on a memory test, time taken to recognize a stimulus, etc.
Naturally Occurring Behaviors – These include essentially all other behaviors that people might engage in and that might be of interest to psychologists.  Examples: whether or not one person helps another, number of hours of television watched per week, whether or not use public recycling containers, etc.
Treatments – These are actions that are taken (vs. not taken), which are intended to have a positive (i.e., beneficial) effect on another variable.  Although the term treatment comes from medicine, neither the action nor the effect needs to be medical or biological.  Examples: taking a drug (vs. not taking it) to reduce anxiety, using cognitive therapy (vs. psychotherapy) to treat depression, using a new method of studying (vs. the standard method) to improve exam performance, etc.
Other Situation / Task Variables – This is an enormous catch-all category that includes anything about a person’s situation or the task he or she is performing.  Number of people in the room, difficulty of a test, noise level, time allowed to complete a task,  
And others dependent on the stated premise being evaluated.

So before you take the next step of doing a trial, evaluating a premise, exploiting an idea, understanding the world around you, contemplate your methodology, before critics start running you out of town after your 15 minute of fame.

1 comment: