Overview of results from two recent studies of the profoundly gifted (accepted to a conference), which aim to understand subpopulations and effective educational interventions. Samples include Ruf and Gross populations.
2. ▪ Several recent studies have shown that:
▪ The profoundly gifted seek higher education at
very high rates (~50% achieve doctorates by
their late 30’s).
▪ The profoundly gifted achieve a lot in their
early careers (grants, publications, works of art,
founding companies…).
▪ Some of the profoundly gifted outshine their
intellectual peers (wranglers).
▪ To help educators better meet the needs of
these students, it’s important to:
▪ Study differences between subpopulations of
the profoundly gifted.
▪ Study the role of educational interventions on
later accomplishments.
3. ▪ Aims two-fold:
▪ Unsupervised learning (persistent homology) to understand subpopulations in 2 samples
of profoundly gifted students
▪ Samples simulated from the SAT score distribution given in the Lubinski 10-year follow-up of the
Study of Exceptional Talent sample
▪ Subset of Ruf’s case study population with extant Stanford-Binet LM scores and deviance-based
scores for nonverbal and verbal intelligence (31 individuals)
▪ Subset of Gross’s case study population with extant Stanford-Binet LM scores and early childhood
achievement scores in math and the humanities (17 individuals)
▪ Supervised learning (logistic regression algorithms) on the Gross population (16
individuals) to understand:
▪ The role of IQ, early achievement scores, and early educational interventions in predicting
graduate school attendance
▪ The role of IQ, early achievement scores, and early educational interventions in predicting early
career recognitions
4. ▪ In each sample,
distinct
subpopulations
emerged:
▪ Evenly gifted
individuals
▪ Who separate out from
the general population
early in the analysis
▪ Have high ability
across those abilities
measured (blue box)
▪ Unevenly gifted
individuals
▪ Intellectual strengths
in either STEM or the
humanities
▪ More common
Example results from Gross sample
5. ▪ In each sample,
distinct
subpopulations
emerged:
▪ Evenly gifted
individuals
▪ Early split
▪ High scores in verbal
and nonverbal IQ
▪ Unevenly gifted
individuals
▪ Split into verbal
talents and nonverbal
talents quite early
▪ Further separation
based on degree of
each type of talent
Example results from Ruf sample
6. ▪ Machine
learning
algorithms
agree on many
predictors:
▪ Positive
predictors
include early
English/math
achievement
scores
▪ Early entry
and grade
skipping as
ineffective
SBLM English Math Early
Entry
Grade
Skip
Subject
Acceleration
Radical
Acceleration
MARS 0.26 0.33 -0.49
BMA 0.54 8.68 51.71 -10.07 -7.55 5.70 18.95
DGLARS 2.00 12.76 -7.17 -12.92 20.28
HLASSO 0.05 1.13 2.38 -0.68 -0.76
All 4 algorithms have decent model fit, though the topologically-
based ones have better fit and more consistent estimates.
Table values give coefficients of logistic regression (take
e^coefficient for odds ratio).
Radical acceleration and early identification of talent is important.
7. ▪ Radical
acceleration is
a strong
predictor of
early career
recognition, as
is early entry.
▪ Agreement on
these two
predictors is
remarkable
across models.
SBLM English Math Early
Entry
Grade
Skip
Subject
Acceleration
Radical
Acceleration
MARS 0.44
BMA -6.25 5.79 0.97 1.38 2.41 33.10
DGLARS 2.20 4.66
HLASSO 0.02 -0.26 1.44 3.27
Topologically-based algorithms yield much better fit than other
machine learning models (MARS, BMA) and are fairly consistent.
Table values give coefficients of logistic regression (take
e^coefficient for odds ratio).
Radical acceleration and early entry may be key for talent
development and chances of eminence (becoming a wrangler).
8. ▪ Distinct subgroups exist within profoundly gifted populations.
▪ Evenly gifted (with equally high abilities across intelligence types)
▪ Unevenly gifted (with distinct strengths in nonverbal or verbal intelligence)
▪ Early education matters with respect to educational attainment and later career
outcomes.
▪ Common educational interventions like early entry or a token grade skip are associated
with lower likelihood of graduate school attendance.
▪ Early achievement/advancement in academic areas are associated with a higher
likelihood of graduate school attendance.
▪ Testing students as early as 5 or 6 can predict potential for graduate work later on and propensity
for choosing careers that require graduate education.
▪ Early identification can provide the type of educational intervention that is associated with high
career achievement.
▪ Radical acceleration is key for career development and may be associated with a higher
likelihood of graduate school attendance.
▪ This may be a part of what makes a wrangler a wrangler.
9. ▪ Farrelly, C. M. (2017).Topological Data Analysis for Data Mining Small Educational
Samples with Application to Studies of the Gifted.
▪ Farrelly,C. M. (2018).Topology and Geometry for Small Sample Sizes: An
Application to Research on the Profoundly Gifted.
▪ Gross, M. U. (2003). Exceptionally gifted children. Routledge.
▪ Lubinski, D.,Webb, R. M., Morelock, M. J., & Benbow, C. P. (2001).Top 1 in 10,000: A
10-year follow-up of the profoundly gifted. Journal of applied Psychology, 86(4), 718.
▪ Ruf, D. L. (2005). Losing our minds: Gifted children left behind. Great Potential Press,
Inc..