Mark Gilbert's Blog

Science and technology, served light and fluffy.

Psych & Human Research 101

John Scalzi (a sci-fi novelist and blogger that I’ve followed for years) recently wrote an amusing post earlier this week titled "Okay, Well, That’s Nice to Know".  In it, he finds that a site titled reports that 89.7% of the Internet has a positive view of "John Scalzi".  This intrigued me.  I mean, don’t get me wrong, I’m part of that 89.7%, but what exactly is this site evaluating to come up with that number?  Without any extensive digging, I found that the footer contains this line:

"The engine is still in beta and Bing’s results may vary per day. Last update: january 25, 2013"

It seems likely, then, that what is happening is that your search term gets run against Bing, and those search results are parsed and analyzed for articles that are positive toward that search term, ones that are negative toward them, and a third group that are "indifferent".  Tally up the articles, and you get your three percentages.

What was really interesting about this was how various search terms would be evaluated.  For example, searching for "obama":


Turned up a mostly negative response, but if you searched for "president obama":

President Obama

The response was overwhelmingly in the other direction.  That got me thinking that perhaps articles that were written and referred to the President as merely "Obama" were more likely to be negative in their view towards him, but if they referred to "President Obama", then they were more likely to be positive.  Let’s try another polarizing name – "clinton":


Ooh, but wait – there are TWO Clintons in politics in the news.  Perhaps their results are being mashed together by the internets:

Bill Clinton


Certainly not definitive, but it looks possible that the results for "clinton" were an average of these two.  Now, let’s add their titles:

President Clinton

So far so good for my theory.

Secretary Clinton

A "No results" message was also intriguing, and the "Why’s This?" link explains it:

Secretary Clinton Why No Results

So, clearly they have some more content curation to do.  At any rate, this got me thinking about the possible ramifications. 

To be fair, I don’t claim to be making some unbelievably wise observation here: not only was this discrepancy picked up by one of Scalzi’s commenters, and I’m sure this is Psych & Human Research 101. 
I also am not claiming a causal relationship here – there simply appears to be two phenomena that coincide – presence or absence of titles and ratings.

If you were conducting a political survey, it would appear that how you referred to the target person – either with their title or without it – could skew the results in one direction or another.  The title conveys a sense of accomplishment and honor, things that when they are used in a question may get you to phrase your response a little differently:

What do you think about Obama’s health care plan?
What do you think about President Obama’s health care plan?

Again, probably Psych & Human Research 101, but still interesting.

And just for the record, the jury is still out on me:

Mark Gilbert

Don’t rush.  I’ll wait.


February 6, 2013 - Posted by | General

Sorry, the comment form is closed at this time.

%d bloggers like this: