## Sunday, April 24, 2016

### Statistical Power by way of The Dark Side of Population Statistics

I would argue that the most important characteristic of statistics is the ease at which it presents and describes comprehensive quantitative relationships between two or more individual entities. Essentially in laymen's terms "Statistics are wicked concise". For example, looking back on my childhood the first statistical memory I'm able to recall is the fact that my father once told me that the chances of me being struck by lightning were one in a million. While he may have been pretty close to the estimate in which any individual could be struck by lightning within a single calendar year (1 in 960,000), he egregiously misled me due to the fact that in any one person's lifetime there is a 1 in 12,000 chance that they will in fact be struck, at least once by lightning. Regardless of the fact that my Dad may have lied to me, I was instantly calmed of my relatively ephemeral fear of being struck my lightning because the numbers said that there was a significant chance (p<0.0001) that I would never be struck by lightning.
So, in order to pay homage to the first "class" of statistic that many people are introduced to I have decided to write a brief summary of some of the less satiable population statistics that are applicable to the majority of todays societies. To begin, "Population statistics is the use of statistics to analyze characteristics or changes to a population." as defined by the US census bureau. Why don't we tackle, unfortunately one of the most important things in life: Money.

This image essentially speaks for itself - that is to say these seven simple statistics unequivocally explain the monetary problem our country is currently faced with. Without the assistance of statistics would you ever be able to compellingly explain the multitude of information disseminated from this image? I think the answer is no.

 The Few, The Proud, the Emory GDBBS Students.

1. I could definitely agree. Math is too abstract and stats definitely gives a nice context so that we can effectively understand data. I just wished that the fundamentals of stats were stressed more in college and high school. Even though the one AP stats course I took and working in labs throughout college was adequate for my basic understanding of stats, I wish that I had more. Perhaps this is where bioinformatics comes in? Now that there is an emergence or stress on big data, the knowledge of stats is more and more crucial for science. More techniques and analyses are being created and it's important to keep up as needed. I am glad that you have stressed to us the application, usefulness and the importance of stats in the STEM fields.

2. Nice post and very relatable topic! Your mention of your first stats "class" caused me to think back to all of those "real world" statistics we encounter everyday. One of my favorites was "68% of statistics are made up"...which is fun to say, but looking back I realize how much emphasis we put on stats but yet so few of the general population can look beyond the flashy statistic. I agree with Danny in the need for more statistics education, because, as your post proves, statistics come up in everyday life. We need to be able to understand how a statistic is calculated, what population it draws from and whether it is believable or not. Otherwise, we are just drawn in by its conciseness without understanding how we got there.

3. Although I agree that statistics are often required to help people frame mathematics in a way they can relate to, I do not think it always helps us really understand an issue.
As we talked about at the beginning of the semester, the cold, hard numbers of probability are often misaligned with people's intuitions. Whether it be temporal discounting (acting as if a future payout is worth less) or the gambler's fallacy, often the numbers just won't jive with what we feel to be real. Because no matter how many stats courses I take, rolling a die 40 times without seeing a 6 almost guarantees the next roll will be a 6, right?.

In short, by-the-numbers economic theories fall short because we do not act rationally. Even if our irrationality makes sense evolutionarily - since a feast in four months won't matter if you don't eat this week - it does not change the fact that numbers aren't as convincing as our guts.

4. I think the ease with which statistics shows "data" is also a very dangerous thing. In the days of buzzfeed and youtube and other such immediate gratification "news" sources, it's almost removed the ability of the population to work through and understand more complicated graphs and representations of data. Yes, its cool to show the percentage of something via a colored in image, but this really limits our society -- people get used to such easy to interpret images, and no longer care to look into more complicated displays and therefore lose a substantial amount of the information they may have otherwise gathered.

Beyond the laziness this type of data representation fosters, another problems with this sort of presentation is that people take it at face value. While I understand that there is a distinct lack of fact checking done by most humans in this world, these types of info-graphics support that. Even in the two you posted, there is not a single source for where the data was acquired from. For all we know, just like your father (:-P), these are lying to you, and us.

5. Dominika, I agree with your comment that "quick" data representation has been totally abused by the general public and can lead to lazy opinion-making. We are definitely inundated with quick graphics like this all the time in the news, on social media, etc. It reminds me a lot of how people interpret political polls. We are often given the quick and dirty breakdown of the results with no regard for how polls are conducted or the nuances of political issues. We'll hear stuff like "58% of X state is voting for Trump" without them telling us that 80% of the poll respondents were retired white people, totally skewing how representative the sample is. Or polls will say stuff like "52% of voters are for legalizing marijuana", as if such issues were yes or no only with no intermediate opinions or compromises. I wonder if even such abuse of quick statistics has lead to Americans becoming so much more polarized in our political opinions? f we are force-fed bad statistics with polarizing data without checking the details, then we begin to make quick, polarized decisions on big issues ourselves.

6. I agree that statistics are a great way of understanding the message that the numbers are trying to provide. But I believe that 'pop data' that is shown in the figures you provided are a bit misleading. For example, the message from the second fire regarding the PhD graduation percentage is misleading. There might be a higher percentage of people with PhDs from one country compared to another. But this data does not interpret the quality of the degrees being earned. Not all PhD training is the same in every country and the quality of training may be low. So which is better: a bunch of PhDs that don't know what they are doing or a handful of PhDs that are truly masters of their given topic?