Thursday Thoughts: Statistics

It's Thursday, so you know what that means...

Thursday Thoughts

If you are new to my blog, or haven't been reading on Thursdays, this is a segment where I post my thoughts about various nutrition and health related topics, or maybe even something else if it is really an intriguing topic.  Most of my current views and opinions come from experience working as a dietitian and as college instructor in nutrition courses.  Most of that learning is science based or observational from the clinical setting.  In general, I hate over-generalizations, so for many of the things I post, I expect there is one or two exceptions because broad sweeping generalizations are always bound to miss out on something. 

For those of you who don't know, I am working on a second Master's degree and last week was my final class, which you can probably guess was statistics.  The way the courses work here are that we have some time before class to do work, then the professor comes to us from the US for a week (Tues-Fri 6-9:30 and Sat-Sun 8:30-4:30), then we get 1-4 weeks after class to finish the final assignments.  Right now I have been neglecting reading blogs, not intentionally, but as a result of the work required for this course.  I have also had less time to get my own blog updated regularly.

With that being said, this post will of course be on the short side, but I do want to address this topic, and what better time than when it is fresh in my head.

1) Stats are often misunderstood.  Make sure that when you see a shocking number or statement using a statistic that you think about it carefully.  These are often used as a way to grab the reader's attention, and while they often share great information, other times they may be used to distort what is going on or may flat out misrepresent what is really going on.  Think about if the numbers can really make sense in the way they are explained.  Having a basic understanding of what statistical terms mean can help you to better evaluate things that you read or hear about.

2) Don't forget to consider the fact that when stats are shown and are not seen in the full context, you lose some of what was going on in the study.  All the variables can't be addressed in one simple sentence bearing a statistic.  You should always question and want to know more information.  I see diet information all the time, but if I don't know the sample size, the control versus experimental groups or the time of the study (among other bits of information) it is like hearing someone make a comment about someone else and have it printed in a magazine out of context.  In other words, it does not capture the whole story and may be used just for shock value or the benefit of whoever is the author.

3) Numbers don't lie, but people can.  It is very easy to manipulate numbers to show what the author wants you to see.  This also goes along with taking things out of context.  It is easy to list numbers, but without more information it may not mean much.  I once saw a stat that showed China would have more college graduates in some specific program (? engineering) than the US would total students.  My first thought was that China has a much larger population than the US, followed by thoughts about what they offer as a country for their majors.  It may be true that the US school system has some issues, but the stat seemed like a huge negative comment about the US school system, but to me I just wondered what other information this was based on because it did not seem like a reasonable comparison.  I have seen others where if you think about it, they stat is comparing apples and oranges and there is information left out.  Again, stats offer dramatic stories, but you need to really think if this stat is accurate in what it is telling you.

4) Whenever you can, if you intend to use the statistic you read about for anything other than your own information, check to see what the stat was really describing.  I know not everyone has access to journal articles and not everyone has the skills to sift through what those articles mean (it does take practice, it doesn't just happen over night), but if you are thinking about using something you read and it involves a statistic, make sure you go back and review the primary source of the information.  It does no one any good to pass on bad statistics.  You may find that it is a great piece of research and that the statistic was used correctly, or you may find that a lot of other information was left out and that was not at all what you thought that statistic was implying.

5) My last thought is mostly a reminder that statistics do not prove things.  They do not show causation.  Research does not provide clear proof or show that one things causes another.  Statistics can however show links or correlation between things.  There may be a relationship present, but not proof.  I think the only ones I remember from my MPH program was that the link between smoking and cancer and the link between asbestos and mesothelioma were proven.  This means there was very clear statistical evidence, and that rarely happens.  There are almost always multiple variables and it is never that clean cut.  As I said at the start, statistics are often misunderstood, and I think often misused as well.

Not all research is good research and just because something gets published does not mean it was good research either.  But it is a start.  Research can always generate more research and it is the role of research to search for links and support for guidelines and policies and for other fields to get support for what they need.  It is up to everyone to critically evaluate everything the hear or read to prevent falling into the trap of letting others think for you by never questioning what you are told.  As a science based person I work hard to find as many reliable sources as I can for subjects I work with, and then for anything else I read, I always question what is really being said.  I am aware of the bias present in almost all writing.  We are only human so it is hard to stay objective, but that is why it is important to not believe everything right away without questioning it, if even just a little.

Well, those are my thoughts for now on statistics, mostly from the side of the issue of misuse.  Statistics play an important role in everything we do because research relies on them and then policy relies on those.  I never said it was a perfect system, but it is the one we use and therefore the stats need to be used correctly to ensure the proper policies and guidelines are in place.  Data is only as good as the way in which is analyzed.  It can provide some very strong information for things we do daily, as long as it is analyzed in the right way.

QUESTIONS:  Have you ever taken a statistics class?  What are your thoughts on statistics?  Have you ever seen a stat used and you really questioned what it was indicating?  Do you believe everything you read/hear or do you question these things.