Data vs Information (Part 4)

As we talk about the dangers of using data instead of information in making decisions, I wanted to highlight how easily it is for data to be taken singularly and completely misrepresent reality if viewed in the wrong context

To do this, I'm going to use something most people are more familiar with than dogs.   So, let's look at crime.

For the purpose of this, we'll look at simple homicide data from the US Bureau of Justice.

If you look at the most basic of statistics, you will note that from the years 1975-2005, 52.2% of all homocide offenders were black, while 45.8% were white (the Bureau of Justice classifies hispanics as white).  Based on this data, one wold assume that blacks were marginally more likely to commit homicides than whites. However, this assumption is made in the void of population data (and completely where the analysis ends when it comes to dog attacks).

If you take population into account, you'll note that there are nearly 6x as many whites as blacks in this country.  So when you look at homicides per 100,000 people, blacks appear nearly 7x as likely to commit a homicide as a white person.  So based on this data, blacks are more dangerous, correct?

This is the danger of looking at only one small segment of data. 

However, when we look at other factors involved in homicides, we find that race has nothing to do with one's likelihood to commit a homicide.  There isn't anything inherantly violent with a being black. 

Through actual research and data-gathering, other factors, such as economic conditions, drug use, gang activity, unemployment and even divorce rates have a much higher correlation with homicides than race.  However, you can never get to those conclusions if you only look at one small data set. 

This is why it's very important to base decisions off of INFORMATION, not small data sets that will not give you an accurate picture of the entire story.  This is why tracking dog bites by "breed" is NOT a valid picture of what is really going on when it comes to dog bites.  We must continue to look at the entirety of data available in order to make decisions. 

Comments

[Caveat]

Excellent points raised here.

I think we have a right to expect that elected officials will use the best available evidence to come to consensus about the most rational, fair and effective approach to any issue.

Some call me naive, but I don't care. I think anything less is dishonest and I wish everybody felt that way.