Defining Nerds with Data

There are a million different ways to define a nerd or a geek, but what does the data tell us? In a previous post, I talked about defining nerds using the Nerdy Personality Attributes Scale (NPAS). I mentioned off-handedly that one of the things that I liked about it is that it is validated, which is to say that the results have been tested.

I don’t want to go down a rabbit hole of explaining reliability and validity, but I do want to explain a little more about what makes psychological tests or assessments different from the ones you see on Facebook or in magazines. Some of the early parts oftest  construction might be pretty similar, because for both Facebook quizzes and psychological assessments, one of the first things you do is create a list of items or questions. The people writing the test will probably make a list of questions by:

  • Asking experts on the topic what kinds of questions should be on there.
  • Looking at other similar tests/assessments.
  • Using the literature and their judgement or knowledge to add questions that seem related.

But where the Facebook quizzes stop, a real psychological assessment will actually go out and gather some data on that tool and compare it to some other data to make sure it’s right. For example, a new assessment of intelligence should probably show some relation to academic achievement, some relation to problem-solving ability, give some prediction of future success, and be somewhat related to other measures of intelligence.

Image result for giant head alien
Head size is not a factor

Unlike most other measures of nerdiness, the NPAS tried to do this. Where most quiz authors will just use their gut to say what items define nerds, the NPAS started off with 445 items and tested each of them on somewhere between 600 and 10,000 participants, only keeping the ones that actually related to the subject’s own assessment of their nerdiness. Yes, it’s still based on each subject’s own judgement of nerdiness, but with that many people we can get a real “wisdom of the crowds” effect and feel pretty confident that we’ve actually got a data-derived definition of what makes a nerd.

What are the Ingredients to a Nerd?

So the NPAS ended up with 26 questions that most related to nerdiness. You can take a look at the items yourself, and at first glance they seem to be related. It definitely passes the face validity test, which is to say you look at the questions and yeah, they seem right. The item that most related to a person’s own assessment of nerdiness (i.e., “How nerdy are you?”) was “I am an overall strange person.” The item most related to a person’s nerd identification (i.e., “Are you a nerd y/n?”) was “I would describe my smarts as bookish.” I wouldn’t draw too much from this, because the other items are all pretty close behind these.

For tests like these, one fun (well, if statistical analysis is your idea of fun) thing to do is a factor analysis, where you look at a big set of items and see if certain items tend to cluster together, which would mean that you’ve got some sub-concepts. So we would expect a really nerdy person to score highly on each of these items, but what we might see is that some people tend to score relatively higher on some items and relatively lower on others. That suggests that there are some well-defined “factors” or aspects or parts of being a nerd, rather than nerdiness being a singular, unitary concept.

So if we do that for the NPAS, we get this:

Looks like a weird clock, right?

What you see here is that we’ve got three factors, and we’ve reorganized the items to fit with their respective factors. Some items, like 21, aren’t a great example of their factor. Some items, like 25, can’t really decide if they fit with factor 1 or factor 2, but that’s ok. What you do now is look at the items and decide what to call these categories:

Factor 1 Factor 2 Factor 3
I collect books. I am interested in science. I was in advanced classes.
I watch science related shows. I like to play RPGs. (Ex. D&D) I am more comfortable with my hobbies than I am with other people.
I can be socially awkward at times. I am more comfortable interacting online than in person. I have played a lot of video games.

Of course, one of the joys and sorrows of data is that sometimes it doesn’t come out like you would expect. Looking at the NPAS, you see some items related to social alienation and introversion, some items related to skills or abilities, and some items related to interests. BUT, when you look at the data that’s not the way scores are clustering! What happens sometimes is that with closely related items like these, the nuance of the wording becomes important. If you look at Factor 1, it talks more about behaviors (I collect, I watch). Factor 2 looks more at preferences (I am interested, I like) and then Factor 3 talks more about the intensity of interests (advanced classes, a lot of video games). Interesting!

Switching to another type of factor analysis, we want to incorporate the overall nerdiness factor into our model. Once again, the strength/darkness of the line represents the strength of the relationship.

So, if you ask the data (as interpreted by me) what makes someone a nerd, it’s mostly their nerdy interests. So there you have it, but with such a wonderful dataset, why stop there?

Are Nerds Smart?

Whoever put together this survey was thoughtful enough to include some other variables in there as well. I might mess around with the demographic later, but I was intrigued by a few – namely intelligence, autism, and personality traits.

I’m going to start with the easiest analysis first – are individuals on the Autism spectrum more likely to be nerdy? The data says yes (F(1401) = 15.4, p<0.01), but what’s even the point of data analysis without a completely unnecessarily interactive visualization?

For intelligence, whoever put together the survey added a really quick and dirty estimate of intelligence using a vocabulary test. Fun sidenote – in the gold standard IQ test (the WAIS), the one task that most correlates with overall intelligence is a vocabulary test. Thank your English teachers, STEM nerds! So, are nerds smart? The answer is a resounding “kinda…”

It’s a pretty weak relationship (r = 0.2) but it’s there. I’d be interested to see what a more detailed IQ score would reveal, as I expect that the vocabulary to intelligence relationship is less true for nerds than for others, based on the predilection for STEM careers and interests, but this is the data we’ve got.

Finally, there’s personality. Maybe another time I’ll give a primer on Big 5 Personality theory, but for right now I’ll just say that some of the best literature on the field says that we can describe personality on 5 dichotomous dimensions that go by the acronym OCEAN: Openness to Experience (vs. consistent/cautious), Conscientious (vs. careless), Extraversion (vs. introversion), Agreeableness (vs. argumentativeness) and Neuroticism (vs. confidence).

So using a multiple regression (which there’s not a great visualization for), we can see that nerdiness is most related to introversion (no surprises there) and openness to experience. Agreeableness is also related, although not as much as the other two. And conscientiousness and neuroticism appear to have no relationship to nerdiness.

Sorry Woody, turns out it is just you.

Leave a Reply

Your email address will not be published. Required fields are marked *