DATA LITERACY
Understanding these frequent logic traps will assist you to keep away from making errors in your evaluation
In a latest article, I outlined information literacy by beginning with the overall definition of literacy and adapting it to the info world:
This text follows an analogous thread — it was impressed by the logical fallacies I realized in highschool. If there are logical errors you may make in your argumentative reasoning, then there are additionally logical errors you may make in your information evaluation and statistical reasoning.
This weblog submit by Geckoboard was a useful start line for my analysis:
From there, I dove into just a few fallacies I’ve had probably the most expertise with. The six I picked for this text are frequent errors which can be straightforward sufficient to make. So maintain studying to be taught extra in regards to the logical traps you possibly can fall into when working with information.
So as to add some coloration to the reasons of every fallacy, I included my very own experiences and pulled in some attention-grabbing examples I discovered from numerous sources on-line.
Earlier than you possibly can carry out evaluation, you want information! And if you must acquire the info your self, listed here are some fallacies to keep away from.
Observer Impact (a.ok.a. Hawthorne Impact)
Do you carry out in another way if you end up being watched? I do know I do. And if researchers aren’t cautious, this human tendency can have an effect on their information. The observer impact occurs when the presence of an observer, or the data of being noticed, impacts the info collected.
I used to consider this usually as an Industrial Engineering intern as a result of I used to be requested to gather time examine information on the manufacturing traces. I used to be hyper-aware that if employees knew I used to be timing them, they could carry out in another way than normal (even when I made it clear that my measurements weren’t meant to guage their efficiency in any approach).
Right here’s one other instance in a producing setting:
“The Western Electrical Firm was the only provider of phone gear to AT&T on the time and the Hawthorne plant was a state-of-the artwork plant that employed about 35,000 individuals. The experiments had been meant to check the consequences lighting ranges had on output. The speculation advanced and teams of employees had been studied to see if totally different lighting ranges, ranges of cleanliness or totally different placement of workstations affected output.
The main discovering was that it doesn’t matter what change the employees had been uncovered to, output improved. However, manufacturing went again to regular on the finish of the examine. This steered watched workers labored tougher.” [1]
Sampling Bias
Sampling bias is when the pattern inhabitants you acquire information from isn’t consultant of the inhabitants you wish to make conclusions about. It’s not all the time straightforward to get a consultant pattern — it could price additional money and time. But when the info is getting used to make choices that may have an effect on the lives of individuals outdoors your pattern inhabitants, then you must keep away from this error.
Understanding about this logical fallacy makes me look again on my early faculty days and cringe. Typically we needed to acquire information and draw conclusions, and I’d simply survey my associates. If I used to be making an attempt to make a conclusion in regards to the basic American inhabitants and even the overall faculty inhabitants, my pattern was not consultant of the entire. (Good factor it was only for faculty credit score and never a brand new product or coverage or the rest.)
Right here’s an instance of how a surveying method may fall brief:
“For instance, a “man on the road” interview which selects individuals who stroll by a sure location goes to have an overrepresentation of wholesome people who usually tend to be out of the house than people with a persistent sickness. This can be an excessive type of biased sampling, as a result of sure members of the inhabitants are completely excluded from the pattern (that’s, they’ve zero likelihood of being chosen).” [2]
Survivorship Bias
Some persons are fortunate — they survive powerful conditions the place the percentages are stacked in opposition to them. Pure disasters, financial downturns, dangerous enterprise ventures, and so forth. And after they get by it, they could look again and suppose that success is extra frequent than it’s since they had been profitable. Or, individuals trying from the skin could solely hear the success tales and never the numerous failures or tragedies.
That is survivorship bias: when the group that succeeds is mistaken for the entire group. Because the surviving/profitable group is extra seen, individuals start to suppose that it’s actually the entire group (the inhabitants).
Right here’s an instance of how this bias can have an effect on the info we encounter in class:
“College students in enterprise faculty can recall how unicorn start-ups had been generally applauded inside the classroom, serving for instance of what college students ought to attempt for — an archetypal image of success. Though Forbes reported that 90% of start-ups fail, total levels are devoted to entrepreneurship, with dozens of scholars claiming that they are going to sooner or later discovered a start-up and develop into profitable.” [3]
Hopefully, you accomplished your information gathering with none errors. You continue to must be cautious as you interpret the outcomes out of your evaluation!
Cherry-Selecting
You might have heard the phrase cherry-picking information. It refers to deciding on solely information factors or outcomes that assist your argument and conveniently leaving out the info that gives proof for the counterargument.
Once I consider this time period, I consider politicians utilizing information. You possibly can add a variety of optimistic or unfavorable spin to the outcomes of a examine simply by solely deciding on a part of the findings to incorporate in a speech. It’s of their finest curiosity, and sadly, it’s comparatively frequent (on all sides of the political spectrum).
Additionally, you will see cherry-picking within the media:
“For instance, think about a state of affairs the place a brand new examine, which relies on the enter of hundreds of scientists in a sure subject, finds that 99% of them agree with the consensus place on a sure phenomenon, and just one% of them disagree with it. When reporting on this examine, a reporter who engages in cherry choosing would possibly say the next:
‘A latest examine discovered that there are many scientists who disagree with the consensus place on this phenomenon.’
This assertion represents an instance of cherry choosing, as a result of it solely mentions the truth that the examine discovered that some scientists disagree with the consensus place on the phenomenon in query, whereas ignoring the truth that the examine in query additionally discovered that the overwhelming majority of scientists assist this place.” [4]
Gambler’s Fallacy
“Wow, he’s rolled three fours in a row now — there’s no approach the following roll shall be a 4!”
Have you ever ever heard somebody say one thing like this in a recreation? It makes intuitive sense at first, however if you take a look at it logically, you must acknowledge {that a} cube roll is an impartial occasion (statistically talking). Every roll has no impact on the likelihood of the following roll.
This instance takes us to the gambler’s fallacy: the “perception that the likelihood of a random occasion occurring sooner or later is influenced by earlier cases of that sort of occasion.” [5]
Other than precise playing, this fallacy might be seen in different purposes the place historic information is closely relied on, like in monetary evaluation:
“Gambler’s fallacy has been proven to have an effect on monetary evaluation. Buyers have a tendency to carry onto shares which have depreciated and promote shares which have appreciated. For example, they could see the continuous rise of a inventory’s worth as a sign that it’ll quickly crash, due to this fact deciding to promote. Gambler’s fallacy could also be at work right here, as traders are making choices based mostly on the likelihood of a reasonably random occasion (the inventory’s value) based mostly on the historical past of comparable previous occasions (the pattern in its earlier value factors). The 2 will not be essentially associated. Its previous value trajectory in itself doesn’t decide its future trajectory.”
There are a variety of components at play in figuring out a inventory’s value, however to simplify the info all the way down to the historic value and make choices on shopping for and promoting based mostly on that appears within the realm of gambler’s fallacy.
False Causality
Simply because two variables are correlated doesn’t imply one brought on the opposite. Correlation doesn’t equal causation! The false causality information fallacy happens when there’s an assumption that one variable’s pattern brought on the opposite variable’s pattern — with out taking a look at different potential components and causes.
For some bizarre examples, try this web site with charts on spurious correlations. Hopefully, none of you’ll say that one in all these variables brought on the opposite — Nick Cage doesn’t deserve that:
When you have ever fallen into one in all these information fallacies, you aren’t alone! That is why it may be constructive to have individuals look over your information evaluation, whether or not it’s for work or faculty, to level out the blind spots you may need in your methodology or reasoning.
Let’s all work collectively to enhance our information literacy expertise and keep vigilant in opposition to information fallacies!
