Statistics Is Science
The Chivers brothers, Tom and David, authored How to Read Numbers, a book written with a mathematical/scientific bent to aid in the proper and accurate delivery of information in media. Both authors provide a unique perspective. Tom, a journalist with a passion for maintaining scientific integrity within journalism, and David, a mathematician serving as a professor in economics. Their conviction is that we should care about numbers as they undoubtedly make a difference in the world. They write, “we’re going to talk a lot about numbers; about how they’re used in the media, and about how they can go wrong – and give misleading impressions. But along the way we will need to remind ourselves that those numbers stand for something. Often they will represent people, or if not people, then things that matter to people.”[1] In short, they argue that we must approach, use, and understand statistics accurately to make good decisions; failure to do so will result in poor choices that are detrimental to both individuals and society alike.
Within the body of the book, several important concepts were of particular interest to me. I will highlight two. One was the Simpson’s Paradox, the problem of using the same data that can tell two opposing stories. Regarding my work among vulnerable populations, I have witnessed the use of data toward this end. While the argument is factual, it is only partially valid in that the same data can also be used to communicate a very different and contrary message. Going forward, mindfulness of this paradox will challenge me to pause and consider how I am using, and hopefully not manipulating, data to provide a compelling case for my argument.
The second concept was determining what constitutes a big number. While our cohort was in DC, my wife and I walked through the Covid Memorial in the Washington Mall. The day we visited, there had been 676,286 deaths in the US as a direct complication from Covid. Tragic, no doubt. This was especially impactful for my wife, who works as an internal medicine doctor in the hospital; thus, she has taken care of and witnessed the deaths of many Covid patients. Emotions aside, I ask, is this a large number? According to Wikipedia, there were 405,399 US deaths during World War II from 1941-1945.[2] Comparing these two numbers, it does seem that the number of Covid deaths is high. Yet, as we consider the number of deaths in the US in 2019, there were 2,854,838 deaths.[3] Again, we must ask, is this a big number? At the same time, a recent article gives evidence that there was a significant rise in the death rate in the US in 2020, 15-19% in fact, with strong evidence of a direct correlation to Covid.[4] In the world of statistics, the point is that context is everything to best understand and analyze data. To aid in understanding the context, the size of the denominator is critical in determining whether or not a number is big as it provides a framework for comparison and analysis. Concerning this challenging conundrum, the authors write, “When is a number a big number? There’s no such thing, really.”[5]
These particular guide points from the authors will go with me as I consider my NPO:
- “Put numbers into context.”[6] It is paramount to know the denominator of the statistical information being considered. To not put numbers into context is to fail to understand the problem accurately.
- “Check whether the study you’re reporting on is a fair representation of the literature.”[7] Are the assumptions fair assumptions? Did they consider all possibilities? Am I presenting data in a way that is manipulative or not fully vetted?
- “Be careful about saying or implying that something causes something else.”[8] Throughout the book and this doctoral program, I have been reminded to be mindful of my words, absolutes, and strong positioning without a foundational base.
- “Be wary of cherry-picking and random variation.”[9] Good academia, and practice for that matter, is based upon accurate information and consideration of all possibilities.
- “If you get it wrong, admit it.”[10] Undoubtedly, human error will reveal itself, and when it does, the best path forward is honesty and humility – own it, admit it, and learn from it.
[1] Tom Chivers and David Chivers, How to Read Numbers: A Guide to Statistics in the News (and Knowing When to Trust Them) (London: Weidenfeld & Nicolson, 2021), 2.
[2] “United States Military Casualties of War,” Wikipedia, October 26, 2021, accessed October 27, 2021, https://en.wikipedia.org/w/index.php?title=United_States_military_casualties_of_war&oldid=1051878295.
[3] “FastStats,” last modified October 20, 2021, accessed October 27, 2021, https://www.cdc.gov/nchs/fastats/deaths.htm.
[4] J. Hosp Med 2021 October;16:596-602 Published Online Only July 21 and 2021 | 10.12788/Jhm.3633, “Excess Mortality Among Patients Hospitalized During the COVID-19 Pandemic,” Journal of Hospital Medicine 16, no. 10 (October 1, 2021), accessed October 27, 2021, https://www.journalofhospitalmedicine.com/jhospmed/article/242997/hospital-medicine/excess-mortality-among-patients-hospitalized-during-covid.
[5] Chivers and Chivers, How to Read Numbers, 63.
[6] Ibid., 166.
[7] Ibid.
[8] Ibid., 168.
[9] Ibid.
[10] Ibid., 169.
7 responses to “Statistics Is Science”
Leave a Reply
You must be logged in to post a comment.
This is a great summary of the book and it was insightful–especially your discussion on “big” numbers and how it applies to covid-19 deaths. When I read their discussion on big numbers what came to my mind was the U.S. deficit. Those numbers are so huge one can barely comprehend them. But when I think of my own personal monthly budget, I think of principles. I stay within my means, why can’t eh government stop raising the debt and spend responsibly? When I think of it in those terms, the big numbers aren’t intimidating anymore. So the debt is no longer is a big number, just take a breath, be responsible, trust the principles, and then it becomes manageable. This book was so good, glad it was on our reading list.
Eric, wow, you sure got a great deal of application from the book toward your NPO. I relate to your insights about wether or not a number is big. For my NPO, the research (i.e. the numbers) around digital natives is in its infancy, so I wonder how many of the number available now are of much consequence. Thanks for pointing out the importance of not assuming or quickly making a correlation based on certain numbers. It’s so easy to see a statistic and offer a hypothesis about how it came to be or what it means going forward. All the best in your NPO work.
Eric, Excellent summary! Really clear. How specifically are you engaging in statistical analysis within your project?
Eric – I’m glad you were able to glean so much application from this book for both your NPO and the work you do full-time with your nonprofit. Having a history in the nonprofit sector as well, I always found it fascinating that for fundraising, it was critical to provide the ‘heart-string story’ as well as the statistical evidence to appeal to a wider variety of donors. You reference the manipulation of data and I too have watched so many organizations over the years cherry-pick the data in order to inflate their effectiveness. I’m glad you were able to find some tangible guideposts to set as you continue with your NPO.
Eric, thanks so much for using Simpson’s Paradox to challenge us all – especially those involved in Christian community development – to be factual in our reporting and not manipulate numbers for the sake of marketing or mobilizing resources. I think the choices we make in this regard will ultimately reveal whether we actually trust God to provide for the mission He has called us to regardless of the circumstances, or we think we must manipulate the data a little to our advantage. Again, thank you.
Eric, I too found the conversation about what constitutes a big number interesting. It is so important to remember how context shapes the “bigness” of a number. I started asking questions like, is the Trinity a big number? What are the numbers God uses to do a risk assessment? This of course led me to the bargaining Abraham did with God in order to save Lot and the residents of Sodom and Gomorrah. As you continue your work, how can the theology of numbers impact the context?
Chivers’ talk about the deeper problem with the Simpson Paradox being the Ecology Fallacy. How important is paying attention to that fallacy for your work?
Eric!
I love that early on you pointed out that the numbers represent more than just a lifeless digit. It is the very things that the numbers represent that actually have meaning.
I have a question for you, how are you going to remain mindful of Simpson Paradox moving forward? Especially, as you have already found its presence in what you are studying.
Your reflections on the Covid numbers are interesting and are they big numbers? The other question I have are those numbers truly reflective of a deaths that are caused by, impacted by, accelerated by Covid? I think that this whole Covid situation is a great opportunity to try out the 22 possible ways to skew the data.