I created a GitHub repo for this project, if anyone wants access let me know: https://github.com/brianmingus/BigFive2MBTI
I created a GitHub repo for this project, if anyone wants access let me know: https://github.com/brianmingus/BigFive2MBTI
This is quite interesting
- Harvey, R., Murry, W. & Markham, S. (1995). A "Big Five" Scoring System for the Myers-Briggs Type Indicator. Paper presented at the 1995 (May) Annual Conference of the Society for Industrial and Organizational Psychology, Orlando. http://harvey.psyc.vt.edu/Documents/BIGFIVE.pdf
I found the Big 5 intercorrelations! (had to dig through a huge number of papers!) These might help me deconvolve the Big 5 / MBTI underdeterminacy problem.
HTML Code:Big 5 Factor N E O A C N -.21*** .05 -.25*** -.49*** E .43*** -.07 -.22*** O -.06 -.04 A .13 C
- Costa, P., McCrae, R. & Dye, D. (1991) Facet Scales for Agreeableness and Conscientiousness: A Revision of the NEO Personality Inventory. Personality and Individual Differences.
Here's a more recent article that specifically discusses that BFI test you're using and reports significantly lower discriminant correlations. As the authors explain:
Originally Posted by John, Naumann & Soto
Last edited by reckful; 09-18-2013 at 10:09 PM. Reason: typo
Here is the R code for deriving the regression coefficients from the published simple correlations
Resulting equations:HTML Code:# Correlation matrix among the Big 5 predictors. From Costa, McCrae & Dye (1991) BigFive = matrix(c(1,-.21,.05,-.25,-.49,-.22,1,.43,-.07,-.22,.05,.43,1,-.06,-.04,-.25,-.07,-.06,1,.13,-.49,-.22,-.04,.13,1),nrow=5,ncol=5) # [,1] [,2] [,3] [,4] [,5] #[1,] 1.00 -0.22 0.05 -0.25 -0.49 #[2,] -0.21 1.00 0.43 -0.07 -0.22 #[3,] 0.05 0.43 1.00 -0.06 -0.04 #[4,] -0.25 -0.07 -0.06 1.00 0.13 #[5,] -0.49 -0.22 -0.04 0.13 1.00 # Correlations between the outcome variables (MBTI) and the predictors (Big 5). From McCrae & Costa (1989) # BigFiveMBTI = matrix(c(.05,-.34,.15,.10, .17,.07,.06,.61,-.27,-.15,-.11,.22,.01,.25,-.16,-.10,.11,.04,-.10,-.34), nrow=4, ncol=5) # [,1] [,2] [,3] [,4] [,5] #[1,] 0.05 0.17 -0.27 0.01 0.11 #[2,] -0.34 0.07 -0.15 0.25 0.04 #[3,] 0.15 0.06 -0.11 -0.16 -0.10 #[4,] 0.10 0.61 0.22 -0.10 -0.34 # Matrix of regression coefficients for Big 5 to MBTI mapping t(solve(BigFive) %*% t(BigFiveMBTI)) # [,1] [,2] [,3] [,4] [,5] #[1,] 0.4247516 0.57887055 -0.51924251 0.071551185 0.41540844 #[2,] -0.3586514 0.03361865 -0.14232090 0.174525355 -0.15672420 #[3,] 0.1956583 0.19209555 -0.20754362 -0.115928690 0.04490257 #[4,] 0.2044586 0.66652262 -0.08037801 *0.005574157 -0.09712008
EI = .42N + .58E - .52A + .07O + .42C
SN = -.36N + .03E - .14A + .17O - .16C
TF = .12N + .13E - .21A - .12O + .05C
JP = .2N + .67E - .08A + .01O - .1C
Note that I still haven't tested these or validated them, but they might work.. give it a shot, and I will dive deeper soon
@Jennifer, here are your final scores in the new equations (the old ones are wrong). The new ones have a zero midpoint.
EI = .42*2.75 + .58*2.25 - .52*3.44 + .07*4.8 + .42*2.56 = 2.1
SN = -.36*2.75 + .03*2.25 - .14*3.44 + .17*4.8 - .16*2.56 = -1
TF = .12*2.75 + .13*2.25 - .21*3.44 - .12*4.8 + .05*2.56 = -.5
JP = .2*2.75 + .67*2.25 - .08*3.44 + .01*4.8 - .1*2.56 = 1.57
As you can see it classified you as ISTP (it also classified me as an ISTP).
Do you mind taking this 11 item version and trying those scores? http://www.ocf.berkeley.edu/~johnlab/pdfs/BFI-10.doc
Holy moly, mingularity. Neither McCrae & Costa (in the 1989 study) nor Costa, McCrae & Dye (in the 1991 study) found any significant correlation at all between Big Five Neuroticism and the MBTI S/N dimension. That is, the 1989 study specifically found no correlation between those two, and the 1991 study found no correlation between Big Five Neuroticism and Big Five Openness (the one that mainly corresponds to MBTI S/N).
And yet... somehow... you're telling us you took the stats from those two studies and arrived at a formula that makes someone's Big Five Neuroticism score the biggest factor (and by a wide margin) in calculating their MBTI S/N preference.
Even assuming your overall approach here makes any theoretical sense — which, for the reasons described in my earlier posts, I don't think it does — you've obviously botched the calculations.
Here's a hint for you as you try to figure out where you went wrong: @Jennifer's MBTI N-equivalent score (Big Five Openness) was her highest score on the Big Five test, and by a substantial margin. I'd say you can safely assume that any conversion formula that turns her into an MBTI S must still have some bugs to be worked out.
The system is underdetermined (!). There are an infinite number of ways to shift the variance around while still resulting in an entirely valid mapping. Trying to use your intuition to understand the coefficients I found is pointless because the solver shifted the variance around willy nilly.
You can look at the Big 5 score and the resulting MBTI score. However, you can't look at the coefficients.
Also, I have not validated this model, nor tested it much, aside from the smoke test of it giving me the right score for my own type.
I find it difficult to believe you're not just blowing smoke at this point. The 1989 study you're using found that, for men, the correlation between Big Five Openness and S/N was huge (61 — the highest correlation in the entire table) and the correlation between Big Five Neuroticism and S/N was all but nonexistent (7). For women, the difference wasn't quite so dramatic, but the Openness-S/N correlation was 41 (the only statistically significant S/N correlation) and the Neuroticism-S/N correlation was 16.
And the 1991 study found virtually no correlation between Big Five Neuroticism and Big Five Openness.
So there's no way (without some kind of large-scale fuckup on your part) to get from that data to a formula where, to calculate someone's S/N score, you multiply their Neuroticism score by .36 — more than twice as high as any of the other four factors — and you only multiply their Openness score by .17.
You either made a careless mistake that you haven't yet woken up to (or aren't willing to admit to), or you truly don't know what you're doing with those numbers.
I don't mind taking it, but I don't have time to run the complicated equations. So here's the data, if you want to use it. Btw, it's the BFI-10 ... 10 questions.
1. 4
2. 3
3. 4
4. 4
5. 1
6. 3
7. 3
8. 2
9. 3
10. 5
E = 2 + 3 = 5
A = 3 + 3 = 6
C = 2 + 2 = 4
N = 2 + 3 =5
O = 5 + 5 = 10
And yes, considering N is the trait that has always been the most obvious and least in question, I'm kind of laughing at coming up as an S. That's... funny.
"Hey Capa -- We're only stardust." ~ "Sunshine"
“Pleasure to me is wonder—the unexplored, the unexpected, the thing that is hidden and the changeless thing that lurks behind superficial mutability. To trace the remote in the immediate; the eternal in the ephemeral; the past in the present; the infinite in the finite; these are to me the springs of delight and beauty.” ~ H.P. Lovecraft