Everyday Research Methods

People's personalities predict the words they tend to use

This piece in Slate online showcased a study that analyzed the words people use into Facebook. The researchers received brief personality surveys from 75,000 volunteers via a Facebook app. They divided the sample into males and females, as well as into different age groupings. In addition, using people's responses to an online personality questionnaire, they divided the sample into introverts and extroverts, as well as people who were neurotic vs. emotionally stable, and so on.  Using "big data" analytic techniques, the researchers were able to find sets of words that best distinguised these groups from one another.

I've copied a closeup of one of the word cloud figures that the authors used. This one shows the differences between the extroverts and introverts. Certain words were much more likely to be used by introverts--these are in larger font in the word cloud on the top. Other words were much more likely to be used by extroverts--these are in larger font in the word cloud on the bottom.

As you can see, reading the word clouds gives a unique sense of what extroverts and introverts are like.

It's fascinating to see the word clouds comparing men and women, young and old, and agreeable and disagreeable people, too--all are available through the original article at PLoSONE, here.

Questions:

a) What are the words that most strongly separate introverts from extroverts?

b) The word clouds are a great way to represent the study's results. But they're not the only way: How might you represent the data in this word cloud in a scatterplot (or scatterplots?)

c)  These data provide strong concurrent validity evidence for the measure of extroversion/introversion that they used. Can you explain why? 

 

Question wording matters

Chapter 6 discusses the importance of well-worded polling questions. When we look carefuly at the way questions are worded, we interrogate the construct validity of a poll: Is the question measuring what it intends to measure?

My home paper, the Philadelphia Inquirer, ran a story on a recent poll last Saturday (right). It shows the results of a poll that asked people about the national health care law. Half of the sample was asked a question about "Obamacare;" the other half was asked the same question, but about the law's official name, "The Affordable Care Act." You can see how the wording of the question changed the responses.

Here's a link to a web story on this poll, with a link to the full text of the polling questions.

My read of this poll's results suggests that the name "Obamacare" is not necessarily a pejorative term. While calling the law "Obamacare" increased the negative responses people gave, it also increased the positive responses they gave.

The poll also suggests that most people think of the law under the name "Obamacare." Calling the law the "Affordable Care Act" caused the number of "Don't know" responses to more than double. Construct validity is relevant again--how can people give their valid opinion on a law that they don't know about?

In light of these results, how would you word the question in order to get the most accurate picture of people's true feelings toward America's health care law?