close

Вход

Забыли?

вход по аккаунту

?

powerpoint - The Achievement Gap Initiative

код для вставкиСкачать
Black Americans Reduce the
IQ Gap: Evidence from
Standardization Samples
William T. Dickens
The Brookings Institution
James R. Flynn
University of Otago
Forthcoming: Psychological Science
Cognitive Ability Predicts
Achievement
• Individual scores on tests of cognitive ability
are strongly correlated with
–
–
–
–
achievement test scores
grades
ultimate educational attainment
many important life outcomes including income
and job status
• It would be difficult to eliminate
achievement gap if IQ wasn’t malleable
Black Americans have made notable
gains in social and economic status
in the last several decades
• Wages and incomes: Between 1940 and 2000, the average
earnings of black men rose from 43% to 73% of the average
earnings of white men.
• Occupational status: The ratio of black to white managerial
and professional workers has doubled since 1960.
• School quality: Per pupil expenditures virtually equal now in
urban areas.
• Academic achievement: Blacks have gained ground on
whites in both math and reading on the NAEP.
• Blacks have lost ground in some areas (incarceration rates,
out-of-wedlock births), but overall SES has improved.
We would therefore be very
pessimistic about about prospects
for equality if after 40 years of
this sort of progress the blackwhite gap in cognitive test scores
was unchanged.
What We Look At
• Scores on nationally representative
standardization samples for four of the most
widely administered tests of cognitive
ability
• We present average “IQ” scores for blacks
with non-Hispanic white mean set to 100
with a standard deviation of 15 points (50%
of whites will score below 100 and about
10% at 80 or below).
Black Scores on Four Tests
of Cognitive Ability
(white average = 100)
89
S-B
S-B
88
IQ score
87
86
WAIS <25
85
WAIS
84
83
WISC
WISC
AFQT
82
81
1970
1975
1980
1985
1990
1995
2000
2005
Aggregating Across Tests
• We have data for each test broken down by
age group and regress the black IQ by age
on explanatory variables
• We find that we can reject that black white
gap is the same on all tests and at all ages,
but we cannot reject that rate of black gain
is constant across time
• Thus we control for age and test in our
regressions of test score by age on year.
T AB L E 1
An nual R ate of B lack G ain in IQ P oints
C ontrols
S am ple
N one
A ge
T est
T est and A ge
A ll A ges
.195
.187
.183
.188
(.046)
(.025)
(.031)
(.021)
.212
.161
.190
.184
(.057)
(.033)
(.033)
(.025)
U nder 25
Black IQ Gain by Percentile
on AFQT 1980-1997
IQ Point Gain From 80 to 97
7
6
5
4
3
2
1
0
0
20
40
60
Percentile of Black Distribution
80
100
Projected Black IQ by Age
in 1972 and 2002
105
100
2002
IQ
95
1972
90
85
80
75
0
5
10
15
Age
1972
2002
20
25
Possible Objections
• The tests changed to become easier for blacks
– Test designers assure us that no such change was made deliberately
– g loadings of all test barely change
– blacks and whites taking both versions of AFQT at same time show
negligible differences in average scores on the two different versions
• Samples changed due to institutionalization or who is
identified as black
– Most subjects are young people still in school (and therefore samples
are probably comparable)
– Reasonable estimates of magnitude of changes can’t explain more than a
fraction of decline
– Profile of American Youth (AFQT sample) is carefully constructed (with
appropriate sample weights) to be representative of US population
including very disadvantaged minorities.
More Objections
• Did we “cherry pick” only tests with gains?
– We report all nationally representative samples we could find
– In an appendix we report all studies using same tests
with comparable samples (not nationally representative)
at multiple points in time
• most show little or no gain, but …
• most are from later in period and show black-white gap
considerably smaller than 16.5 points
• terminal values for black IQ are very similar to our estimates
for same point in time
• some evidence suggesting little or no gain for adults
– Thus at worst these studies raise questions about the timing of
gains, but not the magnitude for children and young adults.
T est
R ea son E x clu d ed
B la ck
G a in s
K -A B C
A ccord in g to Jen sen early
sam p le co n tain ed ov erly d iv erse
g ro up d im in ish in g B -W g ap .
A lso tests d esign ed to m in im ize
b lack-w h ite g ap.
E arly test g iv en to resid en ts of
o n e H ou ston h ou sin g project,
later to resid en ts fro m sin g le
to w n in A lab am a
S am p les are jo b app lican ts to
em p lo yers w illin g to sh are race
in fo rm atio n . N ot represen tativ e
o f po pu lation o f sam e ag e.
F u ll stand ard izatio n sam p le O K
b u t no t av ailab le. O ur so urce is
su b-sam p le o f research sam p le
w ith IQ . W h ites o v er ed ucated
an d d ata no n-rand o m ly m issin g .
1 983 -2 004
-1 po in t
R av en ’s
W on d erlic
W oo d co ck
Jo hn son
F in a l
B la ck IQ
92
1 985 -2 000
93
9 p o in ts, b u t
for 9 year
ad ju st for ag e dif. o ld s
-1 po in t
1 970 -2 001
all ag es
n o g ain
85
ag es 1 6-2 4
1 6-24
2 .4 p o in ts
87
1 987 -1 999
all ag es
n o g ain
88
< 25
<25
-1 .3 po in ts
89
Документ
Категория
Презентации
Просмотров
3
Размер файла
294 Кб
Теги
1/--страниц
Пожаловаться на содержимое документа