From the Economist: "Get a Life. Facebook is bad for you"
In this story, the Economist covers a research program by Kross and Verduyn. Past research has linked Facebook use to negative emotions, as they explain:
Past investigations have found that using Facebook is associated with jealousy, social tension, isolation and depression. But these studies have all been “cross-sectional”—in other words, snapshots in time.
(Sound familiar? Simple bivariate correlations can rarely help us establish causation). The researchers apparently recruited a sample of about 80 college students who...
...agreed to have their Facebook activity observed for two weeks and to report, five times a day, on their state of mind and their direct social contacts (phone calls and meetings in person with other people). These reports were prompted by text messages, sent between 10am and midnight, asking them to complete a short questionnaire.
a) This is a longitudinal study. What makes this a longitudinal study? What are the benefits of a longitudinal study?
b) Here's a quote about the results of the study. Do you think these results can support the causal claim that "Using Facebook makes people more miserable?" Apply the three causal rules to this study's results, which found...
...that the more a volunteer used Facebook in the period between two questionnaires, the worse he reported feeling the next time he filled in a questionnaire.
c) Here's another result they describe. Can you draw sketch of this result (following Figure 8.3 as a model)?
... there was a positive association between the amount of direct social contact a volunteer had and how positive he felt. In other words, the more volunteers socialised in the real world, the more positive they reported feeling the next time they filled in the questionnaire.
d) What's the best way to establish the causal claim that "Facebook use makes people more miserable"?
Suggested Answers:
a) This is a longitudinal study, a type of multivariate study. They measured Facebook use at multiple points in time, and they measured people's emotions at several points in time. By studying the variables over time, the researchers are able to establish temporal precedence. The researchers analyze the data by correlating Facebook use at Time 1 with wellbeing at Time 2, and comparing that to the correlation between wellbeing at Time 1 with Facebook use at Time 2. If one of the correlations is stronger than the other, the researchers can infer which variable influenced the other one, over time.
b) There is clearly a relationship between Facebook use and negative emotions, so covariance is met. We know that Facebook use came first in time (since this was a longitudinal study), so the temporal precedence rule was met. What about internal validity? Are there third variables that could explain this pattern? For example, perhaps women use Facebook more, and are more miserable in general? Reading on, the article reports that the authors investigated some third variables, because they write,
A volunteer’s sex had no influence on these findings; nor did the size of his (or her) social network, his stated motivation for using Facebook, his level of loneliness or depression or his self-esteem.
Therefore, the researchers did consider several important third variables. Are there other third variables that might explain the pattern?
c) You could sketch this result following Figure 8.3. You'd sketch four variable boxes, representing Social contact at Time 1 and Social contact at Time 2, and positive emotions at Time 1 and positive emotions at Time 2. The cross-lag arrow between social contact at Time 1 and positive emotions at Time 2 should show a strong correlation.
d) The best way to establish causation here would be to do an experiment: randomly assign people to use Facebook a lot, or not, and then measure their wellbeing at a later time period.
Comments