It has been widely claimed that the social networking website Facebook promotes narcissism. There has even been some scholarship towards the effect. The paper Narcissism on Facebook: Self-promotional and anti-social behavior published by Christopher Carpenter in the journal Personality and Individual Difference reported that Facebook users that scored higher on the Narcissistic Personality Inventory were more like to use the website more often.

Using data from the personality tests on this website I am going to try to verify this research and extend it to all three of the dark triad personality traits (narcissism, machiavellianism and psychopathy). I will draw the data from three separate tests, one for each trait –the Narcissistic Personality Inventory, the MACH-IV, and the Levenson Self-Report Psychopathy Scale– and divide it into two groups, one group that was referred to the tests via Facebook and one group that came to the test via Google (this will serve as the control group, Google users would seem to relatively synonymous with web users, sorry Bing).

The data is graphed below (the vertical scales in each case is the full range each measured by each scale).

As can be seen, those that came to the test through Facebook were actually less narcissistic than individuals who came from Google. Ditto for psychopathy and machivallianism.

So, where does this leave the theory that Facebook is a simmering stew of narcissism? Well, this probably does not say that much as Google users who search out tests of narcissism are probably not an accurate sample of Google users. Still, it is not obviously consistent with the Facebook as narcissism hypothesis.

I have taken the data collected from the Narcissistic Personality Inventory on this website (n ≈ 7000) and graphed the average scores by age:

And the distribution of scores by gender:

And finally average scores by age by gender:

College majors by narcissism

For the past while individuals who took the Narcissistic Personality Inventory test on this website were asked to enter what they majored in at college (if they did). Above is a graph of average narcissism across all the majors that had at least 25 data points (histogram of sample sizes) after some reduction (e.g. “Business Management”, “Business Administration” -> “Business”). The sample was 52% male and the average age was 40.03 years.

Nursing and Accounting which had lower end sample sizes are different with p < 10-5.

The individual data points can be downloaded here.

To tell if color cues can affect individuals answers on personality tests, the Narcissistic Personality Inventory was altered so that upon page load each user was served up a version with a random red component to the background color and scores were recorded. The colors in RGB notation, where a component can have a value between 0 and 255, were in the range (R=33, G=33, B=33) – (R=232, G=33, B=33).

Below are the average scores for 1861 test takers grouped by the hue they were assigned (from less to more red).

Scores on the NPI by amount of red in page background.

As you can see the trend line is exactly flat, indicating that the background color of a page does not affect how users answer the personality inventory.

The data used in this post can be download for reanalysis here.