Visualizing Personality part 1

Last modified date

This morning I replied to an email from a dear friend of my. Part of the conversation was about finding out who you are and I shared a link to Jordan Peterson & team’s The Big Five Aspect Scale assessment test. This was by mistake as I intended to share Self Authoring Suite. But as it goes in life there is often a reason for those kind of mishaps.

Why not do the Big Five test I asked myself, that could be fun. And so I did and so I got the results. Those are spot on I have to say. I did the 100 questions and got a lengthy story after I filled in the last question. The Big Five assessment is based on assessing 5 main dimensions divided in each 2 sub dimensions:

  1. Enthusiasm (spontaneous joy and engagement) and Assertiveness (social dominance, often verbal in nature) for Extraversion.
  2. Withdrawal (the tendency to avoid in the face of uncertainty) and Volatility (the tendency to become irritable and upset when things go wrong for Neuroticism.
  3. Compassion (the tendency to empathically experience the emotion of others) and Politeness (the proclivity to abide by interpersonal norms) for Agreeableness.
  4. Industriousness (the ability to engage in sustained, goal-directed effort) and Orderliness (the tendency to schedule, organize and systematize) for Conscientiousness.
  5. Openness (creativity and aesthetic sensitivity) and Intellect (interest in abstract concepts and ideas) for Openness to Experience.

from https://www.understandmyself.com/personality-assessment

The lengthy feedback is divided in 5 main sections each describing the main dimension e.g. Openness to Experience and it’s 2 sub-dimensions e.g. Openness and Intellect. Then there is a paragraph or more about the difference between man and women and how it translates into political preferences. The latter deals with a kind of binary left-right perception. This is not too important to me.

In addition to this level of detail, Big Five assessment also shows how a dimension relates to another dimension and how this can sometimes significantly change it’s meaning. For example on the Openness to Experience dimension I score at the 92nd percentile which is classified as “very high”. Sounds good but then read this:

Open, unconscientious people tend to be “under-achievers” (particularly if also above average in neuroticism). Such people appear to have the capability to succeed, can learn quickly, and are creative, but they seldom implement their ideas.

On the Neuroticism dimension I score at 53rd percentile which is classified as “typical or average”. I am not sure if I now belong to the under-achiever category but I would not be surprised looking back at my school time.

Thus it is important to consider the relationships between the 5 dimensions and it’s sub-dimenions and not fare blindly on a high score for one of the five main dimensions.  And this made me think, would it be possible to create single graph or symbol or visual representation of the total picture? Let’s examine this…

The easiest next step is to combine and order the existing graphs:

We do not need the human symbol on the right top of each card. Let’s remove that first.

 

 

We can go into various directions at this stage already but one that jumps out at me is the green bar with a red marker. In the text there is often reference to “low on…” or “high on…”. This makes the green bar likes something binary (low and high) or as a scale: lower than average and higher than average. The latter, “than average”, is also mentioned frequently.

I already know that improving the green bar will produce something better but it is not the result I have in mind. I want to create 1 symbol for ALL the 15 cards. The following is a a throw at what all can be done with the information available and restricting myself to the card format.

Is the use of a gradient bar clarifying things?

Is the distribution a bell curve?

What color to give those higher in openness and experience?

I show these examples because I know they are not good but do communicate some of the problems one encounters when translating data into visuals. When I look at the text I see quantifiers. Words like very high and high tell me something about how to construct a value system. So what do I find:

  1. very low
  2. low
  3. moderately low
  4. average
  5. typical
  6. typical or average
  7. moderately high
  8. high
  9. very high
  10. exceptionally high

Now my assumption is there is also “exceptionally low” and “typical” equals “typical or average”. The complete list of quantifiers therefor could be:

  1. exceptionally low
  2. very low
  3. low
  4. moderately low
  5. typical or average
  6. moderately high
  7. high
  8. very high
  9. exceptionally high

This would give me 9 value brackets to work with. We use a percentile system of 100. We have 9 brackets thus it might seem reasonable to divide 100 by 9, which gives us 11,1 …  Luckily we have 9 brackets to work with so the “typical or average” sits snug in the middle

 

The problem is that my “politeness” is at 9 which is rated “very low” and not exceptionally low. Thus the first bracket of “exceptionally low” can only run from 1 to 8 maximum. Perhaps there a gaussian distribution at play after all? When we visualize my 92nd percentile score on a 9 bracket value system it looks something like this

 

When we zoom in closer we see clearly how the “very high” score borders at the “exceptionally high” score. Bear in mind this is not a pissing (I score low on politeness) contest. “Exceptionally high” might be unfavorably when combined with some other dimension.

An aside

In my feedback to the team I wrote how great it would be to have this above mentioned symbol that communicates the Big Five traits at a glance. I also wrote how interesting this would be as a basis for a dating or matching service. I have in mind dating as in dating but also match making as in employers finding personal, start-ups finding co founders and so forth. I used to work on a dating app last year and together with my daughter we made an interactive demo with Framer.app. Looked cool and can be combined with what I got in mind here.

Source: http://bl.ocks.org/bbest/2de0e25d4840c68f2db1

Instead of trying to optimize the horizontal scale I jump into some unknown territory. There are various graph styles that inspire me, like the aster plot/graph seen on the right. From my understanding the quantification of the Big Five assessment is accumulative in some form.

Openness + Intellect → Openness to Experience

In my case it is:

96 + 72 → 92

Obviously this is not the average of 96+72 as that would make 84. Maybe Openness to Experience is a category on it’s own? It is hard to say at this point without the information about how it is calculated. But for this exercise it doesn’t really matter, I make up something. The goal is that the end result clearly and visually communicates my personality.

Another goal is to lay 2 symbols or graphs on top of one another and deduce possible areas of agreement and conflict. With these requirements in place I continue and see what happens on my screen. Do I have a ready answer? No I do not. This is an adventure, this is Design Thinking at work.

Here the base graph, 5 colors and the same desaturated to quickly check for color vision deficiencies.

The desaturated version. Obviously there is much room for improvement.

I know that each of the 5 segments need to be divided into 3 sub-segments. Also, the total of these 3 sub-segments most likely is less than 100%. Following is an example of one of those segments representing 80%

The blue segment represents 80%. One problem is I measure it from the center. Why not use the surface as a measure?

The circle diameter is 500 pixels thus r=250. The surface area of one segment is about 39 200 px but to make working easier I use 40 000px. 80% of 40 000 is 32 000px. I wonder if using surface area instead of the linear approach actually solves anything. It is an interesting exercise but I decide to stick to the linear division.

The percentage signs are removed as it was causing too much clutter.

If you squint your eyes it almost looks like a mid 20th century abstract painting. In fact, the cover of Peterson’s book Maps of Meaning shows lots of similarities I think.

The color version and I have not yet been able to trace the creator of this image. There was no reference in the book itself as far as I can tell.

Each one of the main five dimensions has 3 segments. If the segments decrease systematically in size it would be easy to make a graph like this.

However, we have no such luck. The main dimension might be of a smaller percentile than on of it’s sub-dimensions. And there will be a lot of overlap and obfuscation if we go down this route. As the five main dimensions are probably more important or easier to communicate than the total of 15 dimensions I decide to keep the main dimension more coming to the front.

This might actually work. The total Extraversion includes Enthusiasm and Assertiveness. Visually this is communicated. Any of the dimensions can be larger than the other. Further exploration will show if this is going to be communicated clearly but my experience tells me it is going to work. Is the overall graph an improvement of the 15 separate bar graphs? I still think this it going to be the case. Let me continue and find out.

I am not sure how useful it is to have the main dimension cover the whole pie and thus I made the B version.

 

Technically this might work better but I do not get the feeling of “personality” from a graph like this. It takes some time to understand what is happening in this graph, like in most graphs by the way.

About the meaning of the word personality from wikipedia:

Personality is usually defined as the set of habitual behaviors, cognitions and emotional patterns that evolve from biological and environmental factors

It is definitely about people and indirectly this is about the interaction between people. To place an icon of a human in the center of this graph is not going to make it any better or more informative. I sincerely doubt that rounding the corners, to make it “softer” is going to work either but it could lead to something else. I am gonna test this. First I create a vector outline with nodes in the center of the segment’s outer boundary. The handles roughly line up with the borders between the segments.

Next I fit the amoeba shaped mask such that the sharp corners are hidden.

And bring the color back

In a way I feel less distracted when I look at this graph compared to the sharp edge one. We have a amoeba shaped graph that with some effort can be understood as the representation of the long Big Five assessment document. Is this ideal? No. The personality assessment site 16personalities uses human figures in a cartoonesque style to represent personality traits. The Big Five assessment quantifies heavily with “above” and “below”. These two observations give me a few ideas. Up to the next post.

This is part 2 of a 3 part series on visualizing the Big Five personality assessment. Head over to part 2 for more.


If this interests you and you like to hire my services as a consultant, teacher or designer then contact me

Jeroen