In the twentieth century, experimental psychological shifted from the study of the mentally ill to research on the cognitive and personality development of basically normal people. Soon there was a plethora of theoretical models on the formation of personality, with different ways of rating personality–too many. In 1961, Ernest Tupes and Raymond Christal sorted through the major theories to see what factors were used repeatedly and settle on five of these named surgency, agreeableness, dependability, emotional stability, and culture. 
This model has been tweaked frequently and the factors renamed; they are typically referred to as Extroversion, Agreeableness, Conscientiousness, Neuroticism and Openness to Experience.  But the theoretical model was only the beginning. Other psychologists developed questionnaires to test for these factors, and performed research to determine the norms where the average for a population lies. Then, they started performing experiments based on the factors. More than half a century later we find many of these experiments cannot be replicated.
This does make sense if you look at how these experiments are constructed. According to a 2013 study conducted by Southern Methodist University in Texas, creative ability and creative accomplishments are higher for people with higher scores in Openness to Experience and Extroversion, and lower for people scoring higher in Agreeableness. So it would seem that you can identify creative people because they enjoy a greater variety of things, but are not necessarily as nice as the average person. Those results are not new and agree with E. P. Torrance’s research in the 1960’s.
But it isn’t that simple. First, you need to know research was based on 75 subjects in Texas universities. Also, creative ability was based on a test measuring verbal creativeness, and creative accomplishments were self-reported, which may have skewed the results to show extroverts as more creative.  Creativity based on by personality assessments depends on population being tested, as well as the criteria for being called creative. Would extroversion still have a positive correlation with creativity if ability had been measured by visual creativity rather than verbal? What would occur if creative achievement was judged by peers rather than self-reported?
Recently, I took the Five Factor Model test on a website. My scores on the Five Factor Model have remained fairly stable for decades. When I was compared to other people taking the same test on the internet, rather than existing norms, my score changed. I discovered that I had suddenly become much more extroverted and agreeable and a lot less neurotic. Comparing myself to average person who takes a self-reported psychometric tests on the Internet is not the same as an intentionally random group selected across the United States. However, selecting populations from the internet is a new method to find people for research projects.
I should not be surprised at the fact that I now appeared more extroverted and agreeable. My children did warn me that introverts and trolls tend to hang out on the Internet.