by James Manning
I am inundated with statistics, data, research, and quantitative analysis every day. The subjects vary from buying power in the black community to people who believe aliens have visited earth. During election season, the news media reports polling on every hot button issue. There are weekly polls showing who is leading, who is gaining, and which politician fell out of favor over a stupid comment made the previous week.
Most of the time, the polls and surveys die on the vine. But some of them become part of a narrative in certain circles that can lead to misconceptions, misunderstandings, and even outright ignorance of the facts. If that narrative survives long enough, it can create a cognitive dissonance among those who formed a world-view based on the misleading narrative of the data.
It is a difficult and frustrating process to convince people that their understanding is wrong, and they often reject new information that run contrary to their ideology. This is especially dangerous when people use misconceptions or a poor interpretation of statistics to generate a narrative and that narrative drives policy.
A great example of a faulty narrative is “Fifty percent of black children dropout of high school.” Based on this statement one could only conclude a crisis of astronomical proportions is happening in the black community. However, the data in calculating dropout rates and graduation rates are different and tell vastly different stories. It is not to say that there are no issues with how we educate and prepare African-American students, but that the narrative doesn’t depict an accurate reflection of the problem and that can lead to bad policy.
Moreover, there is the data where one must ask themselves, “How relevant is that information to the larger issue?” Take for example the yearly report showing the buying power of African Americans. For marketers, retailers, and others looking to attract consumers, this information is important. To the larger issue within the black community, this is neither a true reflection of power nor does it accurately reflect the status of the black community. Considering unemployment rates, poverty rates, incarceration rates, and academic performances, the distribution of spending power within the community is such that it is necessary to drill down on the demographics to determine what policies and where to target those policies. The application of Group Economics will look much different in Jackson, Mississippi than it does in Prince Georges County, Maryland.
Of course, there is the made-up data that exist only to push an ideology. There isn’t much one can do about this because it is impossible to prove a negative to people who have a specific agenda. When a situation like this arises, it is best to go back to the basics of statistics: where did the data come from, what are the variables, what was the sampling methodology, what is the standard deviation? People can make numbers say anything. So rather than concentrate on the narrative, a deeper analysis of the data is more important than conclusions containing problematic biases.
This leads me to two concepts. The first concept is Confidence Interval. Without going too deep in the definition, Confidence Interval is basically the reliability that an estimate will fall within a range. For example: If I have 1000 buckets of shrimp and measure 30 buckets to get their average weight. If the average weight of that sample comes to 8 pounds with the heaviest being 13 and the lightest being 6, my confidence interval is a number that can tell me the likelihood of my remaining buckets being within 7 to 9 pounds.
People conflate homosexuality with child abuse and pedophilia. Obvious bias notwithstanding, the question simply is “What is the data to support this and how confident are those who believe this that within any sampling size of homosexuals that most or even many are pedophiles or have abused children?” Of course, no data exists to support this ideology, but it is a starting point to refute such a claim; although changing someone’s mind on this is an exercise in futility.
The second concept is the rudimentary concept of statistics: “correlation doesn’t imply causation.” A variable’s relationship with an outcome does not mean that the variable is the cause of said outcome. This happens a lot which is why our political system produces so many inept policies to address poverty, racism, economic disparities, and educational gaps. Issues that evoke strong emotions from competing social/economic ideologies are prone to faulty analysis.
It is more important than ever to get “just the facts.” Analysis and interpretations do not suffice. Numbers may never lie… but those who collect them and tell the rest of us what they mean, absolutely do.by