Assorted Notes on Personality

March 9, 2009

Do programmers’ personalities affect how they make code?

To get me started in the area of personality, Jordan Peterson kindly pointed me to three papers.

The first, “Predicting creativity and academic success with a ‘Fake-Proof’ measure of the Big Five,” explains a method of measuring personality that is resistant to bias. This is necessary because people are liable to inflate their Agreeableness, Conscientiousness and Emotional Stability. Native English speakers are able to bias their responses better than others. The key insight is to pit these traits against each other. Instead of asking questions designed to measure one trait, each question forces the subject to weigh traits against each other.

Some highlighted notes from the paper:

Quantifiable measures of personality can be used to predict real-world outcomes. The most predictive of these, Conscientiousness, accounts for 12-25% of the variance in academic performance. Composite measures of self-discipline are twice as effective as IQ at predicting academic performance. Conscientiousness is also the best single personality predictor of workplace performance. After Conscientiousness, Emotional Stability is the best Big Five predictor of workplace performance. It is also a good predictor of job satisfaction and organizational commitment. Extraverts are good candidates for team-based activities. They are effective performers in leadership positions. They are happy and enthusiastic. Agreeableness is a good predictor of teamwork performance. A combination of Extraversion and Agreeableness predicts a transformational leadership style. Openness is a significant predictor of an individual’s creativity.

A quick primer on the psychology of personality: In the early days, theories of personality were mostly opinions. The measures were unreliable and not terribly predictive. Eventually, researchers went through a dictionary, classified all the words related to personality, created questionnaires based on these words, and then performed factor analysis on the result to get five main dimensions of personality. Some analyses find fewer or more factors, but a consensus has emerged around five factors. These are called the Big Five. The Big Five are Conscientiousness, Agreeableness, Neuroticism, Openness/Intellect, and Extraversion. These five are well-studies, valid, reliable, and predictive.

The second paper, “Between Facets and Domains: 10 Aspects of the Big Five,” digs into the constituent components of the Big Five. Each of the Big Five domains can be broken into two aspects. Extraversion can be broken into assertiveness and enthusiasm. Neuroticism can be broken into volatility and withdrawal. Agreeableness can be broken into compassion and politeness. Openness/Intellect is unsurprisingly broken into openness and intellect. Conscientiousness can be broken into industriousness and orderliness. 

There are significant intercorrelations. For example, while conscientiousness is good overall, high orderliness is correlated with neuroticism, which is maladaptive.

The third paper was “Prefrontal Cognitive Ability, Intelligence, Big Five Personality, and the Prediction of Advanced Academic and Workplace Performance.” The main thrust of this paper was to rip into psychologists’ measurement of intelligence.

First and foremost is the formal nature of the psychometric conception of intelligence… [and its] limited grounding in other domains of psychology and neuroscience.

More interestingly for me, it turns out that Dorsolateral Prefrontal Cortex function can be measured and it is significantly correlated with conscientiousness and academic success. Intriguingly, it is strongly correlated with working memory capacity, which suggests we can measure the size of an individual’s working memory. This might prove useful for e.g. a study of program comprehension. 

Another tidbit had to do with self-reports of job performance and personality:

Supervisor-rated job performance was not significantly correlated with any of the personality variables. Self-rated job performance, on the other hand, was significantly correlated with Extraversion, Openness, and Conscientiousness and had near-significant correlations with Emotional Stability and Agreeableness.

Intelligence has a predictive validity of .58 for professional-managerial jobs. The relationship between cognitive ability and performance increases, not decreases, with experience. There is a linear decline in cognitive function over our lives.

One piece of discussion jumped out at me: Those with damaged frontal lobes are susceptible to goal neglect, and this is relatively more likely to occur under conditions of novelty or ambiguity.

Goal neglect = misprioritization of tasks

Novelty or ambiguity = when you don’t really know what you’re doing, e.g. many programming tasks

Could a measure of D-PFCA predict a programmer’s likelihood to shave the yak?

Along the way, Hassam pointed me to “Exploring the underlying aspects of pair programming: The impact of personality,” which purports to study the impact of personality on pair programmers. Alas the study is quite badly done (e.g. it uses the widely discredited Myers-Brigg Type Indicator to measure personality and suffers a general lack of rigour) and gives no useful information.

As an aside, I’m astounded by the vast difference in quality between the psychology papers and most of the empirical software engineering papers I’ve read so far. The psychology papers are of a far higher quality—I hope to learn from them.

%d bloggers like this: