MBA 611 Week 2 Share: To Cause or Not to Cause
... That is the question.
Cause and effect
Is this different from antecedent and consequent? In statistics causation usually means the so-called correct prediction of the impact of some intervention. Is this reality though? We have been using System Dynamics for two weeks now to set up our cause and effect model.
Then we used the model at the end of this week to generate rather shocky data by vibrating key parameters. We only drew ONE sample, ONE possible realization of the causal model! What we generated is very typical of so-called observational models. Does every realization simply come from the cause and effect? Or is there something else at work?
Then we performed some light exploratory analysis with a 2 dimensional density contour map of a scatter plot -- now we are in the realm of cause and effect? or is this just antecedent and consequence?
The question before the house: What does causation mean in business analytics? to the consumer of our analytical product?
McElreath p. 129 talks a bit about this topic.
"Rethinking. What's a cause? Questions of causation can become bogged down in philosophical debate. These debates are worth having. But they don't usually intersect with statistical concerns. Knowing a cause is statistics means being able to correctly predict the consequence of an intervention. There are contexts in which even this is complicated. For example, it isn't possible to directly change body weight. Changing someone's body weight would mean intervening on another variable, like diet, and that variable would have other causal effects in addition. But being underweight can still be a legitimate cause of disease, even when we can't intervene on it directly."
Search your souls, minds, business experience, for an example and argue it and please keep it short and sweet. Do you agree with McElreath? Weirdly, we can use the 4 Cause Analysis to analysis our question.
Enjoy!
Any takers? Is statistical causation any different from our experience and judgment of causation?
ReplyDeleteI think that antecedent differs because it is the behavior behind the cause. I believe that it is a combination of feelings, thoughts and actions that occur before an effect. I think statistical causation is different just because you are implying mathematics. For example, looking at the difference between finding a strong correlation if a drug is the cure for cancer by evaluating patient symptoms and looking at why a girl broke up with her boyfriend is vastly different. This involves someone's feelings and it is hard to generalize a person's emotional response to an event that is bound to have different emotional reactions in humans for any given sample size. I think causation in business analytics is having an assumed variable and solving for statistical significance. Real experiences of cause and effect can equate to millions of possibilities while I believe in analytics, the causation can be narrowed down and solved for mathematically.
ReplyDeleteI suppose the first thing to note is that statistical models, as detailed as they may be, are really just a small slice of our complex reality. In saying "complex", I mean that our reality seems to be the result of combinations of countless causes that happen concurrently, everywhere and all the time. Some cause-and-effect relationships may seem simple, but really anything could also be explained as a long series of interactions that dates back to the beginning of time. For example, I could push a chair and it will fall over. In a small frame of reference, it seems clear that the cause is me pushing, and the effect is the chair falling over. At the same time, we could look at this interaction in a much larger, even universal frame. What caused me to decide to push the chair? what caused the reason for chairs to exist? etc.etc. As far as we know, reality is the penultimate soup of cause and effect, and so far, we have not been able to come up with a computer program to model it in its entirety (maybe someday, but that in-of-itself has far reaching philosophical implications). On the other hand, statistical causation is more simply the fact that a statistical model is capable of outputting different but calculated results when initial conditions change. This type of causation may well reflect parts of reality quite accurately, but certainly not perfectly.
ReplyDelete-Adam P.
DeleteCausation means that one event causes another event to occur. In business analytics, we use statistical causation we make inferences about the model by characterizing the association between variables. These variables can help explain to the consumer of our analytical product, why their product is over or underperforming for consumers. For example, an Athlete's skill can become influenced by intervening on other variables. An athlete's skill can become affected depending on variables such as their training to hone their skills, and the use of teachers to further their skills.
ReplyDeleteCausation: If there is a relationship between an outcome and at least one underlying variable, we can say that the outcome is caused by the variable/variables. For example, the wellbeing of a person, is caused by variables, such as how many hours of sleep this person has slept, how healthy diet did this person eat, and other variables such as relationships to friends and families, personal achievements, and so forth. This is also what we do in business analytics, we use statistical causation to see patterns between variables in our models.
ReplyDeleteWritten by Vetle
DeleteCausation is an important aspect of business analytics, because with so many convoluted factors in business endeavors, it can be difficult to decipher which variables are affecting which outcomes in a business setting. As a result it is important to investigate causation in order to alter processes and understand which processes are leading to consistent outcomes in one direction or the other. In terms of cause and effect relationships, this is one of the fundamental principles of life, because every action has a reaction.
ReplyDeleteImportantly, statistical causation does not always imply direct causation; it simply indicates that there is a consistent relationship that can be quantified statistically. The phrase "correlation does not imply causation" captures this caution because statistical relationships might be influenced by other lurking variables. Experiential and Judgment-based Causation, is based on personal observation, experience, or intuition. It often involves making causal inferences based on what seems obvious or evident in everyday life.
ReplyDeleteThe world is incredibly complicated! There are tons of things happening all the time, and everything seems to be connected in some way. Even simple events, like pushing a chair over, have a long history behind them. We can't possibly model everything with a computer program, but statistical models can be helpful. These models are like simplified snapshots of reality. They can show us how things might change based on different starting points, but they're not perfect reflections of the real world.
ReplyDelete1. Systems thinking encourages stakeholders to view ethical issues not in isolation but as interconnected components of a larger system. This helps in understanding how ethical decisions in one area may impact other parts of the system. Systems thinking helps identify feedback loops where actions lead to consequences that may in turn influence future decisions. This is crucial in ethical discussions to foresee potential unintended consequences of decisions. In essence, systems thinking enhances the depth and breadth of ethical discussions by providing a structured approach to understanding complex interactions, dynamics, and implications within systems. It encourages stakeholders to move beyond simplistic cause-and-effect relationships to consider the broader ethical dimensions of their decisions
ReplyDelete2. I have had fewer issues with Vensim in week 2. I feel i am starting to know my way around the software and am getting used to all the controls within it. Still a little confusing but it's much easier than it was in week 1.
3. I am spending about 3-4 hours on the assignment to make sure i am getting everything down correctly.
John Bernardi^
Delete1. Some ways in which system thinking facilitate the discussion of ethical issues is through the examination and consideration of each component and its values in a certain system. When applying values and parameters to things such as people or groups of people this leads to a conflict where you have to examine people as numbers and factors and vectors and if this is done incorrectly and not thoroughly, it could come across in some ways as unethical possibly. In general, by including diverse perspectives and stakeholders they can be sure that systems thinking ensures a more inclusive ethical analysis.
ReplyDelete2. I struggled a bit with Sensitivity Analysis and SysArch, but in general my grasp on the software as a whole and formatting graphs, causes strips, and changing functions and values improving every week.
3. I spent roughly 2-3 hours on the assignment and this post. This combined with the 2 hour live session on Saturday morning totals to 5 hours a week on this class. I also spend a couple more hours of reviewing other class materials and watching the video lessons.