DATA…GOOD OR BAD? — Part Two: BIASED DATA

Before I go on talking about the pitfalls of dealing with data, I thought I would say a few things about writing  for this web site.   The other day I was thinking that I should really name it “My Own Two Cents dot com.”  Writing is a egocentric act in many ways.  If you keep it to yourself, I guess it falls out of that category, but doing what I do here does at times seem rather self-centered.  So, I looked up myowntwocents.com in a search engine.  I figured someone must have a site by that name.  Turns out that there is no such web site…but I could buy the name for $2495.00.  Now that was funny, to me anyway.  TWO CENTS would only cost me $2495.00!

So, why do I write these blogs for all to see?

I write because I enjoy the flow of what passes for logic in my brain down onto paper, the written word.  We all spend our lives wondering why most people don’t truly appreciate our thoughts about things, and the older we get, the longer the list becomes.  This blog and my novels are my list.  The blog is free and the novels are cheap.  Read what you want.  Then make your own list.

*****

So, back to talking about my logic and how it applies to data.  Feel free to comment.  I refuse to roast criticism, although I may comment back politely.  Don’t worry about embarrassment, The Novel Slide Rule doesn’t get many visitors, but the visitor list does cover a very wide range of countries.  That also surprises me.

*****

“Watch Out!”

If someone says that to you, are your senses heightened?  Now let’s say someone asks you to collect observations for them.  Are you liable to see things that you never noticed before?  There lies the problem.

A number of years ago a government agency that we all know was brought in to investigate a particular set of employee complaints about working conditions that seemed to be causing illnesses in the employees.  It was in an area that was of interest to me, so I read their report.  The report contained one rather silly mathematical mistake, but we can forget that for now.  It is what the report failed to contain that is of interest.

The level of complaints included those recorded both before and after the agency was on site working with the employees.  However, the report never presented the data as a function of time.  And as a function of time, it looked like this.  Before the agency came on site, the level was five complaints per month.  While the agency was on site, the level was ten complaints per month.  After the agency left, the level immediately dropped back to five complaints per month.

Now you can read anything you like into that, but be careful.  It is just data without explanation.  To its credit, the agency did a thorough check of possible causes, and no cause was ever found for any of the complaints.  Guessing is not an option at this point.

I have witnessed similar situations for other types of reported observations at least twice.  In both cases, the level of observations slowly returned to “normal” in about a year.  The observers in question were aware at the beginning that the levels were out of proportion to “normal”.  How did they know this?  Probably this can be blamed on two causes:  word of mouth rumors and the fact that they were asked to specifically look for the problem.  Eventually the excitement of the problem grew dim in the minds of the observers, and it disappeared.

I’m sure the world of psychology has already noticed this effect and written extensively about it, but I’m afraid that general data taking by public or private observations doesn’t always keep that in mind, maybe seldom keeps that in mind.

So, be careful how data is gathered by observation.   Try hard to keep the human mind out of the picture.

Just remember how many times you said in anger to a friend or loved one, “You always do that!”  No they don’t!  You didn’t really keep track.  You just think you did.  Prove it with objective data, or better yet, cut them some slack!

Leave a Reply

Your email address will not be published. Required fields are marked *