Identifiyng gifted children and dyslexia early diagnosis: risk of cheating on IQ tests
1. Marco Ripà
IDENTIFYING GIFTED AND TALENTED
CHILDREN THROUGH IQ TESTING
12th Asia Pacific Conference on Giftedness
Dubai, July 15, 2012
2. How can we measure children’s IQ?
• Through standardized IQ tests (e.g. WISC-IV);
• Using personalized methods, with a good
norm (e.g. high range IQ tests searching for
gifted children);
• Using a combination of both standardized and
personalized tests/instruments.
3. A brief reminder about IQ
There is not a general/singular definition of human
intelligence, but we can try to evaluate specific skills which are
usually related to the most abstract part of cognitive attitude
(Sperman’s Gf factor).
The so called crystallized intelligence (Sperman’s Gc factor)
affects everyday life and contributes significantly to increase
our chances of finding the solutions to complex problems.
WISC test takes into account both these components, while
Raven’s Matrices is focused on non verbal intelligence
quantification.
4. How is IQ theorically distribute
among people?
Normal distribution of random phenomena
Gaussian distribution and IQ (σ=15)
5. How is IQ really distributed?
A change of perspective...
IQ distribution in populations only approximately fits a bell curve
(in the same way as height), diverging from it rapidly as the
scores surpass 2.5 standard deviations from the mean. There is
nothing that says that IQ has to follow any particular kind of
distribution curve at all.
Conventional (ratio) IQ fits children well enough. However, as
children's physical growth slows during adolescence, the growth
of mental age also slows and then plateaus around the 20’s.
6. IMO…
My personal point of view is that modern
standardized IQ tests, such as WAIS tests or
Standford Binet versions, do not suffer from
substantial variance in the [+2, +3.5] s.d. range.
So, for our purpose to discovery giftedness, we can
easily refer to a normal distribution of the scores.
Based on the previous assumption, we could
subdivide gifted children into four different groups:
gifted or moderately gifted, highly
gifted, exceptionally gifted*, profoundly gifted*.
7. WAIS tests classification
According to the table above, we obtain the following subsets:
[+2,+3] s.d. from the expected mean gifted children;
[+3,+4] s.d. from the expected mean highly gifted children;
[+4,+5] s.d. from the expected mean exceptionally gifted children;
> +5 s.d. from the expected mean profoundly gifted children.
8. A short digression: learning disabilities
Around 10% of Italian children
between 6 and 10 are affected by
dyslexia and this number becomes
higher in middle school.
They find it difficult to read, write, Reading disability is the most common
make calculations. form of dyslexia and it is not correlated
with IQ, it is not caused by environmental
factors nor psychological ones, sensorial
deficits or neurological deficiencies.
It is necessary to discover this
learning disability in time, if we
want to reduce its impact on a
child’s future: the worthy age for an
For this purpose, we have constructed a
appropriate diagnosis is 6 years old,
fast and relatively inexpensive method to
but the screening is convenient until
search for giftedness in children who fall in
the age of 9.
the age range 7-16, looking for a possible
learning disability (in particular dyslexia)
with specific regard to children younger
than 9.
9. Could we measure a +5 sd IQ
from the mean?
To recognize a profoundly gifted child we need to use a
smart strategy, because common standardized IQ tests
suffer from ceiling effects and do not have a ceiling
high enough.
One possible solution could be to use a combination of
two different kinds of IQ tests, such as WISC plus
Raven’s Matrices. Nevertheless this does not allow us
to investigate above 2-4 standards deviations from the
mean (depending on the age group we are
considering).
10. We are able to differentiate between 3 age
groups (7-11, 12-13, 14-16), providing an
increased ceiling for IQ estimation…
• Our method is based on a specific combination of two different types of
IQ tests (a collective spatial test plus an individual one), but it could
easily be adapted to different targets. For example, we could try to
investigate about creativity potential (even if it is not so simple to
achieve a good indicator for this purpose) or underachieving problems
(often related to hyperactivity).
• For children/adolescents in the age group 10-16, we are able to
discriminate deeper and deeper in the IQ screening as the age grows.
This lets us combine a collective IQ test with an individual one (or a high
range IQ test), increasing the ceiling of our research and avoiding false
positive cases. The IQ growth during early age is not linear, and it is not
simple to detect highly (or profoundly) giftedness adolescence.
11. The Raven’s Matrices types that we have to use will be different
according to the given age range we are considering:
• Coloured Progressive Matrices for 7-11 years old children,
• Standard Progressive Matrices for the 12-13 years old subset,
• Advanced Progressive Matrices otherwise.
For every pupil younger than 9, we will add, at least, eight verbal--
memory-arithmetic subtests of the WISC-IV (similarities, digit
span, coding, vocabulary, letter-number
sequencing, comprehension, arithmetic, word reasoning), searching
for substantial variances between the performances achieved on
the two different tests.
This could represent a good hint for more detailed individual and
personalized tests, trying to reveal specific comorbidities such as
dyslexia, ADHD, dyscalculia, etc.
12. In detail (operating procedure)
For children above 9, we are firstly interested in detecting giftedness, so
we will perform a preliminary screening using Raven’s Matrices, taking
into consideration only results at or above the 90th percentile (the
bottom level compatible with giftedness).
For these 1.33+ s.d. IQ children, we will administer the WISC-IV to
confirm the giftedness diagnosis, also giving a more detailed
measurement of the theoretical/potential IQ of the subject.
The ceilings (σ=15) grow-up as the age raises:
7-11→IQ 120-135, 12-13→IQ 125-140 and 14-16→IQ 140-160.
This approach could be switched to specific underachieving research,
adding to Raven’s Matrices another collective tool (e.g. SDAI or SCOD).
Moreover, a 120+ IQ score represents a partial indication of a creative
potential above the mean, since 1.33+ s.d. IQs are positively correlated
with creativity.
13. Further lectures on this topic
Paul Cooijmans’ (founder of Glia and GIGA societies) study about
high IQ development by age is very useful in the spirit of better
understanding the ceiling of our approach:
http://www.paulcooijmans.com/intelligence/iq_development_with
_age_modelled.html
The upper bounds that we have previously estimated are based on
my personal experience in the high IQ community, dealing with
some gifted and very gifted children and teenagers.
14. A particular example of IQ growth by age:
one highly gifted boy who took a wide set
of supervised and high range IQ tests
The next slides refer to a highly gifted friend, previously very
active in the high IQ community, who took a lot of supervised
and high range IQ tests during his adolescence.
His IQ development enables us to understand better the
cognitive ability development of a very high IQ teenager.
15. Taken tests list (converted to σ=15) Month-Year:
• RAPM-Set II: raw 35/36, 08-2009;
• Isis Test: IQ 156, 08-2009;
• Cerebrals International 88 composite Contest 2009: according to Dr. Jouve IQ between 153-161, 08-
2009;
• Advanced Spatial Intelligence Test: IQ 147, 07-2009;
• Hieroglyphica: IQ 146, 05-2009;
• Compactica: IQ 150, 03-2009;
• SLSE48: IQ 154, 12-2008;
• SLSE I: IQ 148, 12-2008;
• Logima Strictica 36: IQ 143, 10-2008;
• Cooijmans Intelligence Test: IQ 143, 09-2008;
• Exactica: raw 49/80 (±158), 08-2008;
• Simplex: IQ 138, 07-2008;
• Nemesis Test: IQ 141, 04-2008;
• Plane&Space&Numbers: IQ 153, 03-2008;
• CFNSE-D: 99.914% (age corrected) 99.89% (adult population), 11-2007;
• GET-γ: IQ 144 – age corrected, 08-2007;
• The Sùnesis Test: IQ 145.45, 04-2007;
• Median nicologic.free.fr: IQ 155, 06-2006.
[Average score: IQ 147 (σ =15)]
The IQ scores are based on norms by the test author, if available.
16. age (years) year IQ mean (σ=15) Number of tests taken
11-12 2003 117,0 1
12-13 2004 130,0 1
15-16 2007 145,5 3
16-17 2008 147,3 8
17-18 2009 149,5 4
[My IQ growth has been quite similar to the one above]
17. An important issue related to very high IQ
performances: risk of cheating
Most of the standardized and supervised IQ
tests are not without risk if we want to use them
for giftedness screening (even if they represent
the best choice for average people’s reasoning
skills evaluation, or IQ deficit diagnosis).
It is regrettable that tests like the WAIS or
Stanford Binet (every version) are sold at the
moment on eBay for €300- €1000.
18. The results of a search for
professional IQ tests on eBay
19. It is a joke to cheat on them, for example, achieving a perfect score
under the supervision of a serious psychologist and in front of the
Media too.
Last year a famous episode of cheating occurred on the Cattell
Culture Fair III (form A+B) with a boy who got a perfect score under
the television eye, but that was not able to reach a 130 (σ=15)
performance on a similar test. One possible solution, to reduce this
risk of cheating is to use qualitative high range IQ tests, even if
episodes of cheating also occur on them (for example the Get-γ test
compromised in 2010).
A new idea to create totally culture free numerical IQ tests, which I
have recently had, is based on a method to construct new integer
sequences starting from a given and explicit set of sequences.
The related solving of problems will be linked to inference and
retro-analytical reasoning, similar to the retrograde analysis of
chess problems.
22. Focal points
• Gifted students’ screening is very important in order to invest in the
future from a meritocratic perspective, gaining individual richness
from youths’ talents valorization.
• An efficient strategy, to perform a preliminary talent screening, is to
look for children’s IQ. For those ones under 10, as a preliminary
indicator, it is abstractly possible to adopt an adult IQ test,
distinguishing every performance above the adult mean on that test.
• The approach previously described assures a cheaper and faster
method than administrating WISC-IV for every pupil: a good way to
search for creative children and gifted ones. In fact, we can easily
discriminate children at or above the canonical gifted level (IQ 2+ s.d.
from the mean) setting the Raven’s test to be bound at 120 (σ=15)
and the WISC-IV cut-off at 130.
23. Last specifications
• A 120+ IQ could be a good indicator for creative potential by
itself and students above 120 would be tested again using
different solutions/tools.
• If necessary, children who suffer from ceiling effects on the
Raven and WISC-IV tests could be investigated again using
specifically designed high range IQ tests. To this end, it is
essential to keep in mind that “There are all sorts of
confounding variables that give rise to differences in scores;
therefore, it is always advisable to be skeptical of the results
you obtain” [Paul Laurent].
• For pupils below 9, we are able to point out related learning
disabilities too (dyslexia in primis), without forgetting the
possibility that a child with disability could be a very
gifted/talented little boy as well.