Is that how I’m really perceived on Social Media?

One of the fun parts of working for IBM is that I get to collaborate with a diverse community of researchers spread across the globe. Most recently I’ve been fortunate enough to participate in a study, run by the IBM Almaden Research Center, using their latest social analytics research technology called System-U which does psychometric analysis of social media content. For more check out a recent Venturebeat article entitled IBM researcher can decipher your personality from looking at 200 of your tweets).

System-U breaks its analysis down into the following groups of characteristics:

  • Personality: Psychological characteristics that differentiate one individual from another; currently modelled on the Big Five Personality Traits.
  • Basic Human Values: A person’s belief and motivators, where someone could hold numerous values but with varying degrees of importance; currently includes 6 dimensions of human values based on the work of Professor Shalom Schwartz.
  • Fundamental Human Needs: A person’s fundamental needs; currently based on Maslow’s hierarchy of needs, as well as Ford’s work on Marketing and consumer-related needs modeling.
  • Emotional Style: A measure of how a person’s emotional states look like and change. This helps us answer questions such as what is the emotional outlook of an individual and how resilient a person is emotionally.

Now before anyone cries “Big Brother” this was a completely opt-in study and, at least for me, was more of a learning tool to help me understand how I may be perceived through the different social channels. Also, since I am active on both public (Twitter) and enterprise (IBM Connections) social networks, it allowed me to examine the differences that may exist between the environments. I’ve always firmly believed that people are different inside and outside the enterprise since the drivers for engagement are very different, therefore I was really interested to see if the analysis validated this.

So in the name of social sharing… I am going to bare all and share my own personal analysis results. And I apologize in advance to my fellow tweeters for being a bit high on the “Self-Enhance” dimension when on Twitter. I’m sure that’s one piece of insight that won’t surprise anyone :-)

Marie Wallace Analysis

Before explaining what this all means, let’s start with some definitions:

  • Hedonism: seeking pleasure and sensuous gratification for oneself.
  • Conservation: emphasizing self-restriction, order and resistance to change.
  • Self-transcend: showing concern for the welfare and interests of others.
  • Open-to-change: emphasizing independent action, thought, feeling, and readiness for new experience.
  • Self-Enhance: seeking personal success for oneself.
  • Excitement: getting out there and living life, upbeat emotions, and having fun.
  • Harmony: appreciating other people, their viewpoints and their feelings.
  • Ideal: a desire for perfection and a sense of community.
  • Curiosity: a desire to discover, find out and grow.
  • Closeness: being connected to family and setting up home.
  • Self-expression: discovering and asserting one’s own identity.
  • Openness: the extent to which a person is open to experience a variety of activities.
  • Neuroticism: the extent to which a person’s emotion is sensitive to the environment.
  • Conscientiousness: a tendency that a person acts in an organized or spontaneous way.
  • Extraversion: a tendency to seek stimulation in the company of others.
  • Agreeableness: a tendency to be compassionate and coorperative towards others.

Now for some observations that I thought were interesting:

  1. All in all, despite being very different data sources, my personality pretty much consistently came across the same, which is what I would have expected. I’m still Marie Wallace, warts and all, no matter where I am. However there were some distinctive differences between my public and enterprise personas.
  2. When inside the enterprise it would appear that I am much more thoughtful about the community, about other people’s needs, opinions, emotions, … and things are much less about self-promotion and more about collaboration. Now I don’t believe it says that I’m a total ego-maniac on Twitter, but it does indicate that I am perhaps less sensitive than on the enterprise network. Something I promise to work on fellow tweeters :)
  3. What was a bit surprising, was that I measure as being more open to new ideas and activities inside the enterprise. Now maybe this isn’t hugely surprising if you think about the collaborative nature of the ESN.
  4. What did initially surprise me was that I measured as being more conservative on the external networks? But again I suppose that inside the enterprise you are in a “warm nest”, among friends where you are incentivized to take risks and openly share opinion; whereas externally there may be an element of “not wanting to get egg on your face in public”.

All in all I really enjoyed being part of this analytics project and firmly believe that this type of analytics can play a role in the broader social media analytics space. It just needs to be managed thoughtfully and not used to exploit people.

7 Responses to “Is that how I’m really perceived on Social Media?”

  1. My only issue with it is the infographic is not really that informative. It reads a bit like a horoscope/cold reading. Also does your internet version of you, really reflect you? I would have set your “Agreeableness” much higher (at least in the Enterprise)


  2. Looking at my own personal analysis, I did feel that it got lots of things right so my sense is that it’s not doing a bad job at least of profiling some aspects of me. It definitely didn’t get all things right, for example the extroversion calculation is way off but that may just be a characteristic of social media as a communication channel. Thanks for your comment about the “agreeableness”, because I would like to think that I was more compassionate than the analysis showed :-) I guess my feeling is that all-in-all its an interesting set of analysis and that over time this type of psychometric analysis will become increasingly useful and unfortunately ;-( increasingly abused by certain parties.


  3. My reaction to this approach is that it needs to be underpinned by hard evaluation. We have offline methods for doing psychometric testing that have been around for decades so it is easy to run an experiment to see how accurate this approach is. I should add – it’s quite possible that the original researcher has done this, I’m not sure. The Venturebeat article mentions “within 10% accuracy” which is super vague.

    The second point I would make is that I don’t think can be too worried about public personas not being a true reflection of a person. It’s just another bias we have to be cognisant of. A lot of these methods once refined at the micro level will be employed at a macro level to identify shifts, trends, patterns etc.. And I would add that there is more honest and non-contrived content out there on platforms like Twitter than people would like to believe.

    All in all I would treat this type of research with a healthy dose of scepticism. Claiming that you can read 200 utterances and say with 90% ‘accuracy’ whether a person is hedonistic or neurotic is a massive claim, potentially even ethically questionable. I am or course willing to be proved wrong by a scientific evaluation :) And for the moment I am going to place it on the continuum between horoscope and psychometric testing – in the middle somewhere near Klout.


    • Thanks for the comment Adam.

      First off, the guys did indeed validate their research through using formal psychometric tests in order to create a ground truth; assuming we believe in the tests in the first place :-)

      And secondly, I totally understand and applaud your scepticism. We should all treat analytics (of any sort, but especially social analytics) with a good dose of it. Too many data scientists believe their own hype which is never good. My perspective is that we are too early in the research (and it is research despite what some vendors may claim) to rely 100% on social analytics. There are many reasons for this, not least of all that we are dealing with very sparse data which contains massive biases and this will not get resolved (if ever) until there is much broader and deeper adoption and wider integration of other communication channels. However as they say “don’t throw out the baby with the bathwater” and I do firmly believe that this type of analysis (and other types of social analytics) does provide valuable supporting insights that can be effectively leveraged, but only if done so thoughtfully.


  4. Here are some links to related research papers.

    This one specifically talks about the personality analysis work: Jalal Mahmud, Michelle X. Zhou, Nimrod Megiddo, Jeffrey Nichols, Clemens Drews. “Recommending targeted strangers from whom to solicit information on social media,” In Proceedings of IUI 2013, Santa Monica, CA, March 19-22, 2013. pp. 37-48. Nominated for Best Paper Award @

    There are several more in the pipeline. And there are a number of papers on their previous qCrowd work @



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: