Upload
salome
View
235
Download
0
Embed Size (px)
Citation preview
8/19/2019 Monogrph longitudnal study.pdf
1/129
Society for Research in Child Development and Wiley are collaborating with JSTOR to digitize, preserve and extend access to
Monographs of the Society for Research in Child Development.
http://www.jstor.org
A Longitudinal Study of Moral Judgment
Author(s): Anne Colby, Lawrence Kohlberg, John Gibbs, Marcus Lieberman, Kurt Fischer andHerbert D. SaltzsteinSource: Monographs of the Society for Research in Child Development, Vol. 48, No. 1/2, ALongitudinal Study of Moral Judgment (1983), pp. 1-124Published by: on behalf of theWiley Society for Research in Child DevelopmentStable URL: http://www.jstor.org/stable/1165935Accessed: 21-01-2016 10:48 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/ info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of contentin a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.For more information about JSTOR, please contact [email protected].
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/http://www.jstor.org/publisher/blackhttp://www.jstor.org/publisher/srcdhttp://www.jstor.org/stable/1165935http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/1165935http://www.jstor.org/publisher/srcdhttp://www.jstor.org/publisher/blackhttp://www.jstor.org/
8/19/2019 Monogrph longitudnal study.pdf
2/129
MONOGRAPHS
OF
THE
SOCIETY FOR
RESEARCH
N
CHILD
DEVELOPMENT
SERIAL NO. 200, VOL.
48,
NOS. 1-2
A
LONGITUDINAL
TUDY
OF MORALJUDGMENT
ANNE
COLBY
RADCLIFFE
COLLEGE
LAWRENCE
KOHLBERG
HARVARD
UNIVERSITY
JOHNGIBBS
OHIO
STATE
UNIVERSITY
MARCUS
LIEBERMAN
HARVARD
UNIVERSITY
WITH
COMMENTARIESY
KURT
ISCHER
HERBERT. SALTZSTEIN
AND
REPLY Y
LAWRENCE
OHLBERG
AND
ANNE
COLBY
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
3/129
MONOGRAPHS OF THE
SOCIETY FOR
RESEARCH
IN
CHILD
DEVELOPMENT,
SERIAL
NO.
200,
VOL.
48,
NOS. 1-2
CONTENTS
I. INTRODUCTION
1
II. METHOD
14
III. RELIABILITY
AND
VALIDITY
OF STANDARD ISSUE SCORING
19
IV.
RESULTS
28
V.
DISCUSSION
57
APPENDIX
A
77
APPENDIX
B
85
APPENDIX
C
87
APPENDIX
D
90
ACKNOWLEDGMENTS
3
REFERENCES
94
ILLUMINATING
THE
PROCESSES OF
MORAL
DEVELOPMENT:
COMMENTARY
BY
KURT W. FISCHER
97
CRITICAL ISSUES IN KOHLBERG'S THEORY OF
MORAL
REASONING:
COMMENTARY
BY
HERBERT D. SALTZSTEIN
108
REPLY
BY
LAWRENCE
KOHLBERG AND
ANNE COLBY
120
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
4/129
ABSTRACT
COLBY,
ANNE; KOHLBERG,
LAWRENCE;
GIBBS,
JOHN;
and
LIEBERMAN,
MARCUS.
A
Longitudinal
tudy f
Moral
Judgment.
ith
Commentaries
by
KURT
FISCHER
nd
HERBERTD.
SALTZSTEIN;
with
eplyby
AWRENCE
KOHLBERG and
ANNE COLBY.
Monographs
f
the
ocietyor
Research
in Child
Development,
983,
48(1,
Serial No.
200).
This
paper
presents
he results fa
20-year ongitudinal
tudy
fmoral
judgment
development.
he
study
epresents
n
attempt
o
document he
basic
assumptions
f
Kohlberg's cognitive-developmental
heory
f
moral
judgment.
Subjects
were 58
boys
aged
10,
13,
and 16
at time 1
and were
approximately
qually
divided at each
age
by
social class
and sociometric
status. ociometric
nd socioeconomic
roups
were
qualized
for
ntelligence.
The
study
ncluded
ix
testing
imes-the
original
nterview
nd
five
ollow-
up
interviewsdministeredt
3-4-year
ntervals.At
each
testing
ime
sub-
jects were individuallynterviewed n theirudgments bout nine hypo-
thetical
moral dilemmas. nterviews
were
stage
scored
according
o
Forms
A,
B,
and C
of
the Standard
ssue
Scoring
Manual. All
scoring
was
done
blind
by
individualdilemma. Data
are
presented
n
test-retest,
lternate
form,
nd
interrater
eliability
or
Standard ssue
Scoring.
Validity
of the
instrument
s
discussed.
It was
found hat
ubjects
proceeded
hrough
he
developmental
tages
in
the
hypothesized equence.
No
subject
kipped
stage
n
the
sequence
and
only
4%
(6)
of
the
adjacent
testing
imes
showed
downward
stage
change.Thispercentagewaslessthandownward hangeontest-retestata.
Moral
udgment
nterviewslso showed
high
degree
f
nternal
onsistency
in
stage
scores
ssigned
with
the
great
majority
f
the
interviews
eceiving
all
their
scores at
two
adjacent
stages.
Factor
analyses
by
dilemma
and
moral
ssue
showed
single
general
moral
tage
factor.
Moral
judgment
was
found o
be
positively
orrelated
with
ge,
socio-
economic
tatus,
Q,
and
education.
Stage
scores
n
childhood
were
signifi-
cantly
orrelatedwith
dulthood
cores.
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
5/129
COLBY ET AL.
The results
f this
study
were
interpreted
s
being
consistent ith
a
cognitive-developmental
tage
model.
Subjects
seemed
to
use
a
coherent
structuralrientationn thinkingbout a variety fmoraldilemmas. heir
thinking
eveloped
n a
regular
equence
of
tages,
neither
kipping
stage
nor
reverting
o use of
prior
tage.
The
Standard ssue
Scoring
ystem
was
found
o
be
reliable,
nd it was
concluded hat t
provides
valid
measure
of
Kohlberg's
moral
udgment
tages.
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
6/129
I.
INTRODUCTION
This
Monograph resents
the results of a
20-year
longitudinal
study
of
moral
development.
The
study
represents
n
attempt
to document the
basic
assumptions
of
Kohlberg's cognitive-developmental
account
of
moral
judg-
ment.
According
to this
account,
moral
judgment
is said
to
develop
through
a
sequence
of
six
stages.
Kohlberg
(1969, 1976)
followed
Piaget
(1965)
in
defining tages
according
to the
following
criteria:
1.
Stages
imply
distinct or
qualitative
differences n
children's
modes ofthinkingor ofsolving the same problem at differentges.
2.
These
different
modes of
thought
form an
invariant
sequence,
order,
or succession
in
individual
development.
3. Each of
these different nd
sequential
modes of
thought
forms
a "structured whole."
A
given stage
response
...
represents
n
under-
lying thought
organization
which
determines
responses
to tasks which
are
not
manifestly
imilar.
4.
Cognitive stages
are
hierarchical
integrations. Stages
form an
order of
increasingly
differentiated nd
integrated
structuresto fulfill
a
common function.
[Kohlberg
1969,
pp.
352-53]
In
line with this
notion of
stages,
the
cognitive-developmentalapproach
to moral
judgment
focuses
on the
qualitative
form of the
child's moral
reasoning
and on
developmental
changes
in
that
reasoning. Kohlberg
has
attempted
to
describe
general
organizational
or
structural
features
of
moral
judgment
that
can
be shown to
develop
in
a
regular
sequence
of
stages.
The
concept
of
structure
mplies
that
a
consistent
ogic
or formof
reasoning
can
be
abstracted from the
content
of
an
individual's
responses
to a
variety
of
situations.
It
implies
that
moral
development
may
be defined
in
terms
of
the
qualitive reorganization of the individual's pattern of thoughtrather than
the
learning
of new
content. Each
new
reorganization
integrates
within a
broader
perspective
the
insights
that
were achieved
at lower
stages.
The
developing
child
becomes
better
able to
understand and
integrate
diverse
points
of
view on a
moral
conflict
ituation and
to
take more of
the
relevant
1
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
7/129
MONOGRAPHS
situationalfactors
nto account.
In
this
sense,
each
stage
presupposes
he
understanding ained
at
previous
tages.
As a
result,
ach
stage
provides
moreadequate wayofmaking nd justifyingmoral udgments.The order
in
which
the
stages
develop
is said
to
be
the
same
in
each
individual,
not
because the
stages
are
innate,
but because of the
underlying
ogic
of
the
sequence.
(See
table
1
for
summary
f
the
stages.)
Kohlberg
has
hypothesized
hat
the
developmental
evels
that he
has
described are
stages
in
a strict
Piagetian
sense. To test this
hypothesis,
longitudinal
ata are
required
ollowing
ubjects
ver a
relativelyong
span
of time.
First,
he
stage
concept mplies
hat under normal
environmental
conditions hedirection f moralchangewillalwaysbe upward.Second,it
implies
that there will
be no
stage skipping.
The individual
must
pass
through
ach
stage
n order o reach
the next
tage
n the
sequence.
Third,
the
stage concept
mplies
hat an
individual's
hinking
will be
at
a
single
dominant
tage
cross
varying
ontent,
hough
se ofthe
adjacent
tage
may
also be
expected.
Previous
esearch
has
supported
he
general
notion
of an
age
developmental
rder of
qualitative
responses
nd a
hierarchy
f
pref-
erence
and
comprehension
n
these evels.Such an
orderhas
been shown
by
cross-sectional
ge
studies
n
a
variety
f
cultures
see
Edwards
1981).
It
has
also been shownbya number ftrainingtudieswhich upport heassump-
tionthat
change
after
xposure
o
moral
udgments
t
other
tages
s
always
to
the
next
stage
up.
For
example,
Blatt and
Kohlberg
(1975),
Colby,
Kohlberg,
et al.
(1977),
and
Lockwood
(1977)
found
that
exposure
to
group
moraldiscussionsn which
range
of
tages
was
presented
n
general
led
to movement
o the next
stage
up.
Rest
(1973)
found
hat there
was a
Guttman cale
hierarchy
n
comprehension
f
the
stages,
hat
s,
ndividuals
comprehended
ll
stages
ower than theirown
dominant
tage,
they
com-
prehended he nextstageup ifthey xhibited n thepretest ome (15%)
usage
of
that
stage,
but
they
did
not
comprehend
tages
more
than one
above their wn
dominant
tage.
In
addition to
data
supporting
he
general
dea
of a
developmental
hierarchy
n
Kohlberg's
evels,
large
number f
studies
eviewed
by
Blasi
(1980)
have
generally
ound
significant
ssociations
between moral
udg-
ment
evel
and moral
onduct.
Reviewsofthe
extant
published
iterature n
research
n
the
development
f
moral
udgment
based
on
the
Kohlberg
levels
generallyupport
he
cognitive-developmentalssumptions
bout
the
antecedents nd
correlates
f
moral
udgment
evelopmentsee
Rest
1983).
The
data
we have
just
cited as
supporting
he
developmental
evel
hypothesis
ave
usually ompared
group
means.
The
results
f
these
tudies
have
not
directly
upported
he
strong
tage
claim,
as
critics
ike
Kurtines
and
Greif
1974)
have
pointed
out. In
part,
the
ambiguity
n
some of
these
2
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
8/129
COLBY ET AL.
S.
H
r
.-4>
? -w
l
l)
4)
-
eel
Ic:~
0~
r
Otnt n
ne c
4J
?4
J
,
'
.o-
-
0;
:
.o>t
t
4
4)
0
0
4J
-4
-
P-400r
c
,Kr
>- n
0
to0-o
,
4
4
o
-4oQ
>
d
0"0
L)
I
n
cd
0
0
0
~ 0C
O
4J
tn
4-)
>
4
cd
d
C
~ ~
b
cnccO
~
c
-7.o9o4-4
s.-
VoL-
oZ
-) "
4
0-0
C%
0
4J
4u
0
4-1
?
ac
~hc
oe
Cl o C:'-
4J
"g
-o
.-
dCd
4J
.(
t
n
x
,E
4-)
0
W
>
nn
Cn
cd
"o-
+00
C
4J4Q
"
I
cd
+J)+
ob
0
o4n
4J>0f
4
Olt
0
0
0,0
tn
U
0
tn
k
4-1
rn , h
Qrn
4
tn
0rd
t
?
r
i
-
)
c-,-
0
ya :
0l
u
C)
0(n
?o o
' o 0
4-
o4
7-
r
,,w-
Slo
0
4
-
1-.4
d
13 r
4,4
.
0
1.4
4-) 0
?~
0.4rc
ce
.4
-)
ot
Cc AQ o,
~C
a1-4
4-c,
)
0
C
4-)E
0
d
-)
0
tn
Q
1.4
-)
4)
0
-)
4-)
0"0
E
.,
U
>
-4
E
0
n
4-.4
4-)
e
I
r= C1 f ~ 0
1.4
1.4M
rf
.4
tn
w
r
c r c 7 0 bC;c
-,)
>
0 4-)
4-)
dC:
c
tn
rn
04,1 -4
0n
0
d0 01..
I=Cd
db oCd
c~
"0
Cd
CdC
e
C~
(rgcd
0
0
$cd~j~
, ra
0f
oc
cd 0 d
p
"'C
cdw -
cd
Cd
l
1.
-4
cl."
w
-t
0-
Eo
+,
in
Ics
41
0
Cd
0
0
C
dr
O e
X
d
w
Cd
b
1.4
0c
,i
E
17
,c
--,
(1
1-
1.4
0-
cd
C
14
c
4-)
cn
D
-
0
yo
d
boo
d
0
bo
d
Ic$
d
0
Q
Cd
cd
4
d
tw
0
d
Cl.
C's
'ci
I.,
ci
dIc$
0
0
Cd
0
M4.-
.4
>
0
1-4
1.4
-)
d
0lc
d
)
4
0
4
W4
O
1.oo rfdM4 'ofetn-) 0
d 0 laoM4-44 0Cl, o
M00- 0>-
l.0 0-4,4blo 's
d
01
?:m
w
4-)R
-.
C%
0
%-44
'n
4-)
4
or.
4)
1
Cr Cd-
lo(
Cdl?
"C/)
blco
d
>1E
0-
cd
cn
tn
U
)
E
?
>C) Cd Q Q
0
a4
~
C
d C
r
Cd
3
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
9/129
MONOGRAPHS
C,
_
Cl)
_ ,
..
Q
-#.-4
-
00
~
04-4
0
C/)
>
0C
w
Cd
c~Q
-4
Q 4-
U0)
-d-
0
,o$
r-
4-)I) 4
0
0,
H
0
0 0-4Uw0,)
?
Cd
$4
Co
>r?.4
r
---bO
-
,_
rn
,.c:
oo
C)
c
d
0
-A
V,
+3
H
g
~o
cn
..4
U044
0
nc
rn
0
Eo,
W
,
4
C
%400
4-)--
4-
C
4-)
4-)
-
.-4
r
t
0
w
~~P-4
~ ~
0
,
-.4
4-)
4-
bo
C
I:
-
o-
n
-
-4
.-4u
d
-)0
4
J
0
0
-
0
C
O
4
-
0
d
A
-
.
44
0
^0
n00,
-
0
"4
bo
r
n
'*-
:0
-
_
.4=
C4r
k4
*,4
J
0
-)
C
4-)
(U
C
4,
Cd d 0
4
Cbo
cd
d
>C's(.
Cd
rn
_I
$,
0
rn
- = 4) M >17 C
0l)
4
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
10/129
COLBY ET
AL.
findings
as been
due to the imited
eliability
f
Kohlberg's
1958
method
of
interviewing
nd
scoring
moral
stage.
The
longitudinal
ata
published
byHolstein 1976),Kuhn (1976), and White tal. (1978) have shown ome
anomalies
n
stage sequence.
t
has not
been
clear whether
hese
nomalies
represent
failure
f
fit f he
strict
tage
modelto moral
udgment
evelop-
ment
or whether
hey
have
represented
onfusions
n
the
conceptual
defini-
tionsof the
stages
or
problems
n
the
reliability
r
validity
f
the measure.
In an
early
report
f
his own
longitudinal
ata
over
10
years,
Kohlberg
(Kohlberg
& Kramer
1969),
while
reporting
ome
measurement
roblems
leading
to
anomalies,
tressed
genuine
failure f
the
stage sequence
hy-
pothesisn thecollegeyears eadingtoso-called ophomore etrogressionn
development.
n
1973,
Kohlberg
reinterpreted
he
anomalies
as
resulting
not from
retrogression
ut from
ncorrect
onceptualization
f
stages
as
they
appear
in
development
fter
high
school.
He
reported
ome
clinical
analyses
fcases
suggesting
hat
ollegeretrogressors
ere
xhibiting
stage
"41"
(having
movedout
of onventional
morality
ut not
yet
nto
principled
moral
udgment)
and
proposed
revised
onceptual
definitionsf the
fourth
and
fifth
tages.
n
fact,
uch
revision
ccountedfor
nly
few
fthe anom-
alies
reported
by Kohlberg
and
Kramer.
This
implied
that there
were
generalproblems n the reliabilitynd validityofthe stage criteria nd
scoring
method.
The
present
Monographttempts
o
address
the
validity
of the
stage
model as
applied
to
longitudinal
ata and
the
associated
problems
f
stage
definition
nd
measurement
hich
his
ask
has
required.
It
reports
reanalysis
f
1956-68
ongitudinal
ata
along
with
nalysis
of
the
subsequent
ata collectedfrom
he same
subjects
rom 968
through
1976. The
current
nalysis
nvolved he
pplication
f
new
scoring
method,
Standard ssue Scoring Colby, Kolhberg,Gibbs, t al., in press),based on
a
substantially
evised
ccount of
the
stages.
One
particularly oteworthy
change
n
moral
udgment
tage
scoring
s
the
omission f
stage
6 from
he
current
manual.
Stage
6 was
omitted
artly
ecause none
of the
nterviews
in the
ongitudinal
ample
seemed
ntuitively
o be
stage
6,
partly
because
the standard
ilemmas
re not
deal for
ifferentiating
etween
tages
and
6.
The
question
f
whether
tage
6
should
be
included s
a
natural
psycho-
logical
stage
in
the
moral
development
equence
will
remain
unresolved
until
research
using
more
appropriatemoral dilemmas nd interviewing
techniques)
s
conducted
with a
special
sample
of
people
likely
to
have
developed
beyond
tage
5.
Because the
scoring
method s
critical o
the
validity
f
our
study,
we
will
include a
fairly
xtensive
escription
f
Standard
ssue
Scoring
before
proceeding
o the
Methods
chapter.
5
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
11/129
MONOGRAPHS
DEVELOPMENT
F
THE
STANDARD
SSUE MORAL
JUDGMENT
INTERVIEW
ND SCORINGSYSTEM'
In
the
early
1970s
Kohlberg
and his
colleagues
began
to redefine
he
central
features
f
the
moral
udgment
stages
and
to construct
more
adequate
assessment nstrument. even of
Kohlberg's
longitudinal
ases
were
used as
the data
base
for this
endeavor.
It
was
expected
that
this
process
would
yield
a
more
precise
nd accurate
picture
f
moral
udgment
development.
or
example,
t
was
expected
hatwhen
Kohlberg's
emaining
longitudinal
ases
were
analyzed using
the
revised
tage
criteria,
he
data
would fit
the
core
assumptions
f
cognitive-developmental
heory
more
closelythan theydid in Kramer's originalanalysis.Because the longi-
tudinal
interviews sed
to validate
the reformulations
ere
kept
entirely
separate
from
hose
used to
generate
them,
this
revision
process
avoids
circularity y allowing
test
of
prediction
o
data other
han those
used
to
generate
he
scoring
ystem.2
The
early (Kohlberg
1958)
scoring
systems,
entence
Rating
and
Global
Story
Rating,
were
based
on
what
was
essentially
content
nalysis.
That
is,
both
systems
ocused
n
what
concerns
subject
brought
o
bear
in
resolving
dilemma
(e.g.,
a
concern for ove
betweena
husband and
wife,
he
mportance
f
obeying
he
aw,
a fearof
punishment)
nd
treated
those
concerns s
indicators f
developmental
tage.
These
scoring
ystems
yieldedenough
equence
anomalies n
the
1968
analysis
f
the
ongitudinal
data to
warrant
ubstantial
evision f
the
stage
definitions.
his
revision
process
resulted n a
clearer
differentiation
f
moral
udgment
structure
from
ontent.
That
is,
moreformal
r
abstract
eatures
f
moral
udgment
were
dentified
nd
formed he
core of
the
new
stage
conceptions.
The
basic
developmental
oncept
underlying
he
revised
tage
equence
is levelof ociomoral erspective,hecharacteristicoint fviewfromwhich
the ndividual
ormulates oral
udgments.
n
discussing
evel
of
ociomoral
perspective,
et us
begin
by
saying
that
we
believe the
perspective
aking
underlying
he moral
stages
s
intrinsically
oral
n
nature
rather
han
a
logical
or
social-cognitive
tructure
pplied
to
the moral
domain. In
this
interpretation
e
follow
Damon
(1983)
and
Turiel
(1979)
in
their
onten-
tion
that
here
re
many
ypes
f
perspective
aking,
ach of
which
develops
separately
s
a
resultof
experience
n
a
particular
domain.
In
this
view,
spatial, social,
and
moral
perspective
aking
are
fundamentally
ifferent
processes ather hanapplications fa
singlegeneral
tructureo different
content
reas.
That
is,
the
form f
spatial
perspective
aking
s
intrinsically
'
The
research
reported
here
was
supported
by
National
Institute
of
Child
Health
and
Human
Development
grants
HD02469-01
and
5R01HDO4128-09.
2
We will
briefly
ddress the
ssue of
circularity
ater n
this
paper.
For a
fuller
iscus-
sion
of
that
issue
and of
the
theoretical and
methodological
changes
referred o
here,
see
Colby
(1978).
6
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
12/129
COLBY
ET
AL.
spatial,
not
moral or
social,
and
the
form f moral
perspective
aking
s
intrinsically
oral,
not
spatial
or
social.
According o Kohlberg(1976), stagesof social perspectiveaking, s
portrayed
y
Selman
(1980)
both
generally
nd
specifically
n
descriptive
concepts
of
friendship,
eer
group,
nd
parent-child
elations,
re
parallel
to
stages
of moral
perspective
aking.
As distinct
from
descriptive
nd
strategic
ocial
perspective
aking,
Kohlberg
1976,
1981)
defines
moral
perspective
aking
s
being
a deontic
or
prescriptive
udgment
f
obligations
nd
correspondingights.
e defines
the
domain
ofmoral
udgment
tages
s the
domain
of
usticereasoning
nd
describes hefollowing ourmoral orientations sed byrespondentso his
dilemmas
of
conflicting ights:
1)
general
or
impartial
following
f
rules
and
normative
oles;
(2)
utilitarian
maximizing
of the welfare
of each
person;
3)
perfectionistic
eeking
of
harmony
r
integrity
f
the self
and
the
social
group;
and
(4)
fairness,
alancing
of
perspectives,
maintaining
equity,
nd social contract.
While
an
emphasis
n
ustice
s
most
bvious
n
the
fairness
rientation,
"just,"
impartial,
r consistent
nd
general
rule
maintenance
s
also central
to
the normative rder
orientation. he
utilitarian
rientation onsiders
justiceas the operation fquantitativelymaximizingocialwelfare onse-
quences.
n
the
perfectionist
rientation,
he
central
lement
s
treating
he
self,
he
other,
nd
the self'srelations
o
others
s
ends,
not as means.
m-
plicit
n
this orientation
s
fairness
r avoidance
of
exploitation
f
others
and the need to
benefit
hem.
Kohlberg
(1981)
claims that
ustice
is
the
most structural"
r
"operational"
domain of
moral or evaluative
hought.
For
Kohlberg,
s
for
Piaget, ustice
structuresre
operations
f
ocial inter-
action
parallel
to the
operations
f
logico-mathematicalhought.
Justice
"operations"ofreciprocitynd equalityparallel operations
f
reciprocity
and
equality
n
the
ogico-mathematical
omain.
These
operations
re basic
to the idea
of
fairness s
a
balancing
or
weighing
of
conflicting
laims
through
perations
f
reciprocity,
quity,
esert,
nd
prescriptive
ole
taking
(putting
neself
n
the
place
of
the
other,
r
the Golden
Rule).
Each
stage
uses these
operations
t its own
level of
moral
perspective.
or
example,
at
stage
2
the Golden
Rule
is
integrated
s
concrete
eciprocity,
return avor
for
favor nd
blow
for
blow." At
stage
3,
the Golden Rule is
interpreted
s
imagining
ow
the selfwould feel
n
the other's
place
before
eciding
how
to act.
The
stage
3
moral
perspective,
hen,
differs
rom
he
stage
2
moral
perspective
n
engaging
n
ideal role
taking
n
dyadic
or
group
relations.
Comparable
differentiations
etween
any
two of the
five
stages
can
be
made
using
the
five
evels
of
moral
perspective.
These levels of
moral
perspective,
briefly
described
in
table
1, provide
a
general
organization
of
moral
judgment
and
serve to inform
and
unite
7
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
13/129
MONOGRAPHS
other
more
pecific
moral
oncepts
moral
norms
nd
elements
o be defined
later)
such as
the natureof the
morally ight
r
good,
the
natureof moral
reciprocityr moralrules,ofrights, fobligationor duty,offairness, f
welfare
onsequences,
nd
of moral
values
such as obedience
to
authority,
preservation
f human
life,
and maintenance
f contracts nd affectional
relations.
Within
ach
of
hese
pecific
moral
oncepts norms
nd
elements),
the
form f
developmental
hange
is
to some
extent
pecific
o
the nature
of
the
particular
oncept
n
question.
However,
he
general
moral
perspec-
tive
can be
seen to
underlie
ts more
pecific
manifestations.
The
change
n
stage
conceptions
meantthatthe
moralconcerns
which
had been consideredndicative f tage ntheearly ystemsecamecontent
in
relation
to the
newly
dentified tructures. his reformulation
f
core
structure
mplied
need
to
redefine
he
unitof
analysis
n
scoring
uch
that
the
concerns
hatwere confused
with
tage
n
the 1958
systems
ecame the
units
f ontent or
tage nalysis
nthe1971
Structuralssue
Scoring ystem
(Kohlberg
1971).
In
the
atter
ystem,
tage
was
assigned
o material
within
each
content nit
on the basis
of
evel
of
perspective.
Structuralssue
Scoring
proved
o be a
substantial
dvance over arlier
systems.
he
new
stage
definitions
ocusing
n
evel
of ociomoral
erspective
not onlyyieldedmoreorderly ata, theyalso made possible more con-
vincing
ationale or he
nternal
ogic
of
hemoral
udgment tage equence.
There
were
a number
of
problems
with
this
approach,
however.
The
need
to
determine
evel
of
perspective
n
stagescoring
n
interview
mplied
that the
scoring
nit must
be
large
and
that
scoring
riteriamustbe
very
general
nd abstract. his
meantthat
coring
ecisionswere
ubjective
nd
oftenunreliable.
Moreover,
the reliance
of
this
scoring ystem
n
very
general
features
f
stage
structuremeantthat
findings
f
nvariant
ongitu-
dinal
sequence
and
stage consistency
cross ssues could
be
attributed o
consistency
n
or
a universal
equence
of
the
general
features ather han
providing
vidence
or
onsistency
r
sequentiality
f he
detailed
onceptual
differentiationsncluded
n
more
pecific
moral
tage
definitions.
The aim ofStandard
ssue
Scoring
was to
overcome hese imitations
f
Structural ssue
Scoring-to
achieve
greater
bjectivity
nd
reliability
n
scoring
by
specifying
lear and
concrete
tage
criteria
and
to define
he
developmental equences
of
the
specific
moral
concepts"
within ach
stage
as well
as
the
sequence
of
the
global
or
general tage
tructures.he
redefi-
nitionofthe scoringunit was thekeyto achieving heseaims. An under-
standing
of
the
general
framework f
the
Standard
Issue
Scoring System
and of
the
procedures
sed to
constructt
s
essential o
an
explanation
f
ts
validity.
et us
turn,
herefore,
o a
description
f Standard
ssue
Scoring.
The
procedure
or
onstruction f the Standard
ssue
Scoring
System
was
designed
o
avoid the
problem
f
circularitytheoretical
erification
s
a
self-fulfilling
rophecy).
With this
in
mind,
seven
cases
were
selected
from
8
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
14/129
COLBY
ET AL.
Kohlberg's
longitudinal
ample.3
These
"construction
ases,"
chosen
at
random
from
mong
those ested
t
all
six
times,
were
assignedglobal stage
scoresbased on intensive iscussionnd analysis sing oncepts rom truc-
tural
ssue
Scoring.
The
responses
o
each
dilemma
were
then
lassified
nto
clearly
defined
scoring
units,
or "interview
udgments."
Each of
these
interview
udgments
formed he
basis for
a
"criterion
udgment"
to be
entered
n the
scoring
manual. The
stage
scoreof each criterion
udgment
was
assigned
on
the
basis of the
global
score
of the nterview
romwhich
t
was derived
nd
a
conceptual
nalysis
fthe
dea
it
embodied.
The criterion
judgments
generated
by
these seven
construction
ases were
ater used to
score theremainingnterviewsn the ongitudinal tudy hrough process
of
matching
nterview
material
to criterion
udgments
n
the
manual.
Those
cases not
n
the construction
ample comprised
blind
ample
which
was
not
used at
all until he
scoring
manual
had been
completed.
MORAL
JUDGMENTNTERVIEW
Three
forms f moral
udgment
nterview
ereconstructed. ach
form
consists
f hree
hypothetical
oral
dilemmas,
nd
each dilemma
s
followed
by 9-12 standardizedprobe questions designed to elicit justifications,
elaborations,
nd clarificationsf the
subject's
moral
udgments.
For
each
dilemma
these
questions
ocus
n
the two
moral ssues hat were chosen
to
represent
he central
value
conflict
n
that dilemma.
For
example,
the
familiar Heinz dilemma"
(Should
Heinz steal
a
drug
to save his
dying
wife
f
the
only
druggist
ble
to
provide
he
drug
nsists n
a
high
price
that
Heinz cannot afford
o
pay?)
is
represented
n
Standard
Scoring
s
a
con-
flict
etween
he value of
preserving
ife
nd
the
value
of
upholding
he aw.
Life and law are the two standard ssues
for
his
dilemma,
nd
the
probing
questions
re designed o elicit nformationn thesubjects' onceptions f
these two ssues.Of
course,
he
dilemma
can also be seen
to
involve
other
value
conflicts,
or
xample,
between
he husband's ove forhis wife
affilia-
tion)
and the
druggist's
roperty
ights property).
n
order
o
standardize
the
set of
ssues cored
or
ach
interview,
hecentral ssues
or
ach
dilemma
were
predefined.
his
preidentification
f
standard ssues not
only
allows
the
sampling
f moral
udgments
bout the same six issues
per
interview)
for ach
subject,
t also
allowed the constructionf
three
parallel
forms
f
the moral udgment nterview. hus, the first ilemma n interview orms
A
and B
focuses
n
the
same
two
ssues,
ife nd
law. The
second
dilemmas
ofthe
two
forms oncern
he conflict etween
morality/conscience
whether
to be lenient
oward
omeonewho
has
broken he
aw
out
of
onscience)
nd
punishment whether
o
punish
someone who
has brokenthe
law).
The
3
Each
case included
four
to
six interviewswith the
same
subject
collected at
4-year
intervals.
9
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
15/129
MONOGRAPHS
third dilemmas
nvolve a conflict etween
authority
e.g.,
obeying
one's
parent)
and contract
abiding by
or
holding
someone
to an
agreement).
FormC involves hesame
six
issuesbut nsomewhat ifferentairsthan n
forms and
B.
(See
App.
A for
nterview
orms.)
The
first
tep
in Standard
Issue
Scoring
nvolves he classificationf
the
subject's
responses
o
a
dilemma
nto the two standard ssue
categories.
This
is a
fairly
imple
procedure.
n
the Heinz
dilemma,
for
xample,
all
responses rguing
or stealing
he
drug
are
classified s
upholding
he
life
issue;
all those
arguing
gainst
tealing
he
drug
are
classified s
upholding
the law
issue.
In
the
follow-up
ilemma
Should
Heinz be
punished
f
he
does steal the
drug?)
all
responses
rguing
for
eniency
re
classified
s
morality/conscience;
ll those
arguing
forpunishment re classified s
punishment.
The
first
tep
in
scoring
esponses
o a
dilemma,
then,
s
to
separate
material
nto the two issue
categories.
ince the issue units
re
large
and
often
ontaina
great
deal of
material,
tandard ssue
Scoring
nvolves wo
further
ubdivisions
by
norm
and
by
element,
ee
table
2)
before
tage
TABLE
2
CATEGORIES OF MORAL CONTENT
A.
THE
ELEMENTS
Upholding
normative rder:
1.
Obeying/consulting
ersons
or
deity.
Should
obey,
get
consent
should
consult,per
suade).
2.
Blaming/approving.
hould
be
blamed
for,
isapproved
should
be
approved).
3.
Retributing/exonerating.
hould
retribute
gainst
(should
exonerate).
4.
Having
a
right/having
o
right.
5.
Having
a
duty/having
o
duty.
Egoistic
consequences:
6. Good
reputation/bad eputation.
7. Seekingreward/avoidingunishment.Utilitarian
onsequences:
8. Good
individual
onsequences/bad
ndividual
onsequences.
9. Good
group
consequences/bad
roupconsequences.
Ideal or
harmony-servingonsequences:
10.
Upholding
haracter.
11.
Upholding elf-respect.
12.
Serving
ocial
ideal or
harmony.
13.
Serving
human
dignity
nd
autonomy.
Fairness:
14.
Balancing
perspectives
r
role
taking.
15.
Reciprocity
r
positive
desert.
16.
Maintaining
quity
and
procedural
airness.
17. Maintaining ocial contract rfreelygreeing.
B.
THE
NORMS
1. Life
4. Affiliation
(
9.
Civil
rights)
a)
Preservation
(5.
Erotic
ove
and
sex)
(10.
Religion)
b)
Quality/quantity
6.
Authority
11.
Conscience
2.
Property
7. Law
12.
Punishment
3.
Truth
8.
Contract
10
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
16/129
COLBYET
AL.
scoring begins.
This results
n
a
fairly
mall
unit
of
analysis.
The
Standard
Scoring
unit,
the
criterion
judgment,
is
defined
by
the intersection
of
dilemma X issue X norm X element. Classificationofresponses
by
issue in-
volves
determining
which
choice
in
the
dilemma
is
being
supported
or
which
of
the
two
conflicting
ssues
is
being
upheld.
Classification
by
norm
is
a
further ubdivision
of
the interview material
by
its value content.
The
norm
represents
the moral value
or
object
of
concern
that is used
by
the
individual to
justify
his or her choice
in
the dilemma. For
example,
one
might
argue
that
Heinz
should
steal the
drug
to save
his
wife's life
(life
issue)
because
of the
importance
of their
loving relationship
(affiliation
norm). The elements representthe differentways in which the significance
of
a
norm
may
be construed.
They
are the reasons
for
endowing
the
norms
with
value. To
continue the above
example,
Heinz's love forhis
wife
(affili-
ation
norm)
might
be considered an
important
reason to
save her
(life
issue)
because that
is
a husband's
proper
role
(duty
element)
because of
his
gratitude
toward her
(reciprocity
lement),
or for
number
of
other reasons.
Each of these
is
treated
by
Standard Issue
Scoring
as
a
discrete moral
idea,
and
each
represents
a
separate
unit of material. The
procedural
compli-
cations of
subdivision
by
norm and
element were
found
to be
necessary
in
order to define a unit that was narrow enough to be homogeneous, to cap-
ture what seems to be
a
single,
discrete
moral
concept
or
idea,
yet
broad
enough
to
represent
the idea's full
conceptual
or
structural
significance
for
the
subject.
That
is,
the
system
provides
a
way
for
the scorer
to
categorize
interview material
in
a
nonarbitrary
way
into
manageable,
conceptually
coherent units
(interview
judgments)
which can
then be
stage
scored
by
matching
them
to
very
specific
and
concrete criteria
in
the
scoring
manual
(criterion
udgments).
(See
App.
B for
illustration
of norms and
elements.)
This systemofcontentclassificationprovidesfor meaningfuldefinition
of
the
scoring
unit
which,
in
addition
to
being
essential
to
those
preparing
the
scoring
manual,
is
also
necessary
for
each
scorer
in
the
process
of
analyz-
ing
an
interview.
In
effect,
material
in
the interview
transcript
s
classified
according
to
three
types
of
content
category
before
it is
classified
by stage
or
structure.
In
addition to
resolving
the
unit
problem,
this
approach
is
useful in
preventing
some
of
the
content-structure
onfusions
which
have been
prob-
lems
for
earlier
moral
judgment scoring
systems.
For
example,
stage
3
reasoning oftenfocuseson love or the affiliative elationshipas a reason for
Heinz
to
steal
the
drug
in
Dilemma
III.
That
is,
the
content
of
affiliation s
likely
to
occur in
the
context of a
stage
3
(interpersonal
concordance)
struc-
ture. Earlier
scoring
systems,
which
failed
to
clearly
differentiate
ontent
and
structure,
tended to
misscore
as
stage
3
reasoning
that
was
in
fact
structurally
more
advanced but
that
focused
on
affiliative content.
(For
example,
"Heinz
should
steal
the
drug
for
his
wife
because
of
his
deep
com-
11
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
17/129
MONOGRAPHS
mitment o her
and
to
the
marriage
nd
the
responsibility
hat
results rom
that
commitment.")
By
first
ategorizing
ccording
to
content
nd
then
addressinghequestions fstructurerstage,StandardScoringprocedures
involve
explicit
differentiation
f
form
nd
content
nd,
in
effect,
emind
the rater that identification
f a
particular
ontenthas not answered
he
stage scoring
uestion.
THE EXPLICATED
RITERION UDGMENT
A
structural
r
theoretical
xplication
has
been
provided
for
each
criterion
udgment
n
the manual.
Each
explication
ncludes statement f
the underlying tage structure eflectedn the criterionudgment Stage
Structure),
etailed criteria
or
defining
match o
the
criterion
udgment
(Critical
ndicators),
xplanations
f
distinctions
mong
criterion
udgments
a
scorer
s
likely
to
confuse,
nd
several
examples
of
interview
material
whichcan
be considered o match he criterion
udgment.
See
App.
C for
illustration
f
explicated
riterion
udgment.)
This criterion
udgment
ormatwas
an
important
dvance over
earlier
scoring
ystems.
he
subjectivity
f
scoring
decisions
s minimized
y
the
concrete nd explicit pecificationfexactlywhatconstitutes matchbe-
tween nterviewmaterial nd
a
criterion
udgment
n
the
scoring
manual.
STANDARD
SSUE
SCORING
RULES
As
we have
said,
Standard Issue
Scoring
nvolves
irst
lassifying
he
responses
o each dilemma nto two broad
categories-the
standard
moral
issues
for
hat
dilemma.Within
ach
issue a
stage
score
s
entered
or
ach
match
between criterion
udgment
n
the manual and a
moral
udgment
in the interview. sually,somewhere etweenone and five uchmatches
are
assigned
or
he ssue.
Although
n
practice
hese
matches end
to cluster
at
a
single
tage
or at two
djacent tages,
here
s
no
restrictionn
the
coring
rules
hat
requires
uch
consistency.
he
rules
llow
scores
o
be
assigned
t
all
five
tages
f
matches
re
found
t those
tages.
In
calculating
n
overall
core
for
three-dilemma
nterview
orm,
ne
assigns
summary
core for
ach of
the
six
issues.
As
many
as
three
tages
may
be
represented
n
the
ssue
ummary
core.The six
ssue
cores
re
then
combined
o
yield
a
global
interview core and
a
continuous moral
matu-
rity"score (weighted verage) forthe interview.See App. D forfurther
discussion
f
scoring
ules.)
SUMMARY
In
summary,
ohlberg
has
postulated
ix
stages
n
the
development
f
moral
judgment
and has
described
these
stages
as
characterized
by
holistic
12
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
18/129
COLBY
ET AL.
internal
organization
r internal
onsistency
f
structure cross
differing
content nd
by
the invariant
equence through
which
ndividuals
roceed
through he stages. In Kohlberg's earlyworkthesestageswere assessed
using
n
interview ased
on
hypothetical
moral dilemmas nd
two
systems
for
coring esponses
o
those nterviews:
entence
Rating
and Global
Story
Rating.
Some
12
years
ater,
new
scoring ystem,
tructural
ssue
Scoring,
was
introduced.
From a
psychometric
oint
of
view,
there were
serious
problems
with
these
scoring ystems.
he
early
systems
acked
validity
s
measures f
structural-developmental
tages
since
they
nvolvedwhat was
essentially
content
analysis
of
interview
esponses.
The
later
system
achieved ncreasedvaliditywhen used byvery xperienced atersbut was
subjective,
ifficulto
use,
and
unreliable,
articularly
hen
used
by
those
who
had
not
had
extensive
raining
nd
who had
not tudiedwith
Kohlberg.
Data will
be
presented
ereto
show
that
he
current
ystem,
tandard
ssue
Scoring, rovides
or
eliable
nd valid
assessment
f
moral
udgment tage.
This
was
achieved
through edefining
he
unit
of
analysis
nd
the
relation
between moral
udgment
content
and
structure
nd
through
pecifying
more
precisely
he
process
of
inference
rom
nterviewmaterial to
stage
scores. We will also
argue
that this
improved
method
of
analysis,
when
appliedto Kohlberg's ongitudinal ata, provides moreadequate testof
the
stage
hypothesis
n
moral
udgment
evelopment
han has
been
possible
to
date.
13
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
19/129
II.
METHOD
DESIGN
The
design
f
he
original
ross-sectional
tudy,
whichwas ater
followed
up
longitudinally,
as determined
y
a
numberof
theoretical oncerns.
Three variables
were ncluded
n
the
design: ge,
socioeconomic
tatus,
nd
sociometric tatus.
ocioeconomic
tatus
was
expected
o
be
positively
sso-
ciated withmoral
udgment
development
n
part
because it
was
assumed o
be
an
indicator
f
sense
of
participation
n
the
ociety
s a
whole.
Kohlberg,
drawing
n
G. H.
Mead
(1934),
considered
his
enseof
participation
o be
an important eterminantfmoraldevelopment. ociometric tatuswas
intended
o
be
an
indicator f
peer
groupparticipation,
hich
Piaget 1965)
argued
was
crucial
to
moral
development.
he
age
variablewas
intended o
establish
he
age developmental
haracteristics
f
types
of
response
o
the
moral
dilemmas.
These
originally
ross-sectional
ubjects,
tratified
y
three
levels
of
age
and
two
levels
of
social
class
and
sociometric
tatus,
were
followed
ongitudinally
t
regular
-4-year
ntervals or
0
years.
At each
testing
ime
ubjects
were
nterviewed
n
the
nine
hypothetical
moral
dilemmas
making
up
the
threeforms
f
Kohlberg's
moral
udgment
interview: ormsA, B, and C. At some
testings
ubjects
lso
responded
o
additional
nstruments
ot
reported
n
this
Monograph,
ncluding
nterviews
concerning
ttitudes oward
social
and
occupational
roles
and
attitudes
toward
ex,
the
Thematic
Apperception
est,
Loevinger's
entence Com-
pletionTest,
Piagetian
cognitivemeasures,
nd
Selman's
role-taking
nter-
view.4
SUBJECTS
The basic sampleconsisted riginally f84 boysfilling hefollowing
2
X
2
X
3
factorial
esign
see
table
3).5
4All
of the
data from
ohlberg's
ongitudinal
tudy
re
archived t
the
Henry
A.
Murray
Research
Center
f
Radcliffe
ollege
where
hey
re
available
to
interested
researchers
or
urther
nalysis.
6
Girls
werenot
ncluded
n
Kohlberg's
riginal ample
because
dding
ender
s
a
fourth
ariable
would
have
required
oubling
he
ample.
Given
he
aboriousness
f
the
14
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
20/129
COLBY
ET
AL.
TABLE
3
DESIGN OF
STUDY
LOWER
ES
HIGHER
ES
Integrates
Isolates
Integrates
Isolates TOTALS
10
years....
6
(6)
4
(6)
6
6)
5
6)
21
13
years....
4
(6)
3
(6)
5
(6)
5
(6)
17
16
years....
6
(6)
2
6)
3
(6)
4
(6)
14
Totals.... 16
9
14 14
25
28
53
NOTE.-Figuresnot nparenthesesndicate umber fcross-sectionalubjectswho were ollowedp longi-
tudinally.
igures
n
parentheses
ndicate umberf
subjects
n the
original
ross-sectional
ample.
Asnoted n
text,
ive dditional
working-classubjects,
ge
19 at
time
,
were
dded n
1964,
t time
.
The
population
had
the
following
characteristics:
(a)
Age.-Subjects
were
10, 13,
and
16
years
old
at time
1.
(b)
Class.-To
facilitate
filling
he
design,
two suburban
Chicago
school
systems
were
selected,
one
predominantly
upper
middle
class,
the
other
pre-
dominantly
lower middle
and
working
class.
Fourth-,
seventh-,
and tenth-
grade
classes
formed the basis
for
selection.
A
dichotomous
judgment
of
a
boy's socioeconomic status was based on his parents' occupation and edu-
cation,
as
reported
in
the
school folder.
In
spite
of efforts o
obtain
discrete
groups,
it was
necessary
to
take children
along
a
fairly
broad
continuum
with a
rather
arbitrary, though
conventional,
dividing point.
The
fathers
of
boys
in
the
lower- and
lower-middle-class
group
included
unskilled,
semiskilled,
and
skilled laborers
and
white-collar
workers without
a
college
education.
The
fathers
of
boys
in
the
upper-middle-class
group
included
small
businessmen, accountants,
and
salesmen with
a
college
education,
semiprofessionals,
xecutives,
and
professionals.
(c)
Sociometric
tatus.-When
entering
a
given
classroom the
investigator
described
the
procedures
to
be
followed,
including
a
"revealed
differences"
discussion
among
three
boys.
The
boys
were
then asked to
write the
names
of
three
other
boys
with
whom
they
would like
to have the
discussion.
The
sociometric test was
informally
discussed
with the teacher
and
compared
with
notes
in
the
school folder
before
a
final
selection was
made,
in
order
to
somewhat
reduce
determinants
of
school
and
athletic
achievement
and
temporary
fluctuations
of
popularity.
The
teachers were
asked to
comment
on the boys' social connectedness, not on their moral characters or repu-
tations.
As
in
the
case
of
socioeconomic
status,
there
were not
enough
subjects
available
to obtain
only
extreme
groups,
so
that the
dichotomy
tends to
divide a
continuum.
Boys
who
were
never
chosen
or
who
were
chosen
once
interviewing
nd
scoring
procedures,
uch a
large sample
was not
feasible.
n
retrospect,
however,
the
omission
of
girls
s
regrettable.
15
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
21/129
MONOGRAPHS
or twicebut never
by
someone
hey
had
themselves
hosenwere
designated
sociometricsolates.
Boys
with
at least two
reciprocal
hoices or who
were
chosenat least threetimes, t least once reciprocally, eredesignated s
integrates.
(d)
Intelligence.-IQ
coreswere
takenfrom choolrecords t time
1
and
were based
on various
group
tests
routinely
dministered
n
the
various
schools
e.g.,
the Otis
and
the Thurstone
MA).
An
attempt
was
made to
equalize
intelligence
or
the
social class
and
sociometric
roups.
Those
whose
IQs
were above
120
or
below 100 were excluded
from
he
study.
Although
omplete
qualization
was not
achieved,
he
IQ
differences
ere
small
and
nonsignificant.
ean
IQ
for middle-class
oys
was
109.7;
for
lower-class
oys
t was 105.9.
Mean
IQ
for
ntegrates
as
111.2,
for
solates
104.4.
(e)
Religion
nd
ethnic.-The ethnic
and
religious
omposition
f our
sample
s
presented
n
table
4.
LONGITUDINALOLLOW-UP
ND SAMPLE
ATTRITION
The
study
ncluded
six
testing
imes-the
original
nterview
nd
five
follow-upnterviews. ecause notevery ubjectcould be reachedat each
time,
the number
of
interviews
er
subjectranged
from
ne to
six.
Only
those
subjects
with
at least two interviewswere
included
n our
current
analysis.
This conditionwas
met
by
58 of
the
subjects.
As
shown
n
table
5,
the most
typical
number
of
interviews
ompleted
fora
single ubject
was
four.
All
but
three
subjects
were interviewed t
least three times.
Age,
sociometric
tatus,
nd
socioeconomic tatus
f
origin
f
hese
58
ongitudinal
subjects
re
presented
n
table
3.
There
was no
attempt
o locate for ater
ongitudinal
ollow-up
hose
boyswhowere unavailablefor ollow-upt time2. As shown ntable3,the
initial
dropout
from
he
cross-sectional
ample
was not
evenly
divided
by
age
and
social class.
Whereas
12
working-class
oys
dropped
out,
only
8
upper-middle-class
oys
did
so.
Ten
of
the time
1
6-year-olds
ropped
out,
while
only
3
of
the
10-year-olds
id
so. To
compensate
or he
attrition f
TABLE 4
RELIGIOUS COMPOSITION OF
CROSS-SECTIONAL
SAMPLE
Group N
Working-class
roup:
C
atholic....
.........................................................
14
Protestant...
........................................................
22
Upper-middle-class
roup:
C atholic.......
...................................................... 3
Jew
sh
..............................................................
2
Protestant...
........................................................ 31
16
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
22/129
COLBY ET AL.
working-class
ubjects,
n
1964
(testing
ime
3)
we added 5
working-class
19-year-olds
o the
sample.
Because
IQ
and sociometric
tatus
data
are
missing orthesesubjects subjects91-96), theyare not includedin our
analyses
of
social
class,
education,
Q,
and sociometric
tatus
n
relation
to
moral
udgment.
However,
because
they
were each
interviewedhree
r
four
imes,
hey
were
ncluded
n
our
analyses
f
ongitudinal
equence
and
internal
onsistency
f
moral
udgment.
Thus the
final
ongitudinal
ample
was
skewed
somewhat
toward the
younger
ohort but was balanced in
terms
f
socioeconomic
tatus.The cohortbias should not affect he
basic
findings
f the
study,
owever,
ecause orderliness
f
ongitudinal equence
and internal onsistencyfresponses o the interviewwerenot relatedto
age
cohort.
To
determine
whether he
subset of
subjects
who
dropped
out after
time 3
were
higher
or
lower
n theirmoral
udgment
han
the
sample
as
a
whole,
we
compared
time
1
moral
maturity
coresof
the whole
sample,
subdivided
by
age
group,
with
time
1
moral
maturity
coresof the
drop-
outs,
again
subdivided
by
age group.
As
shown
n
table
6,
the
dropout
means
were
almost
dentical o the
total
sample
means,
and
there
was
no
tendency
or
ubjects
who
dropped
out
earlier
n
the
study
o show ower
TABLE
5
FREQUENCY
OF
LONGITUDINAL
FOLLOW-UP
Number
of
nterviews
Completed
nd
Scored
N
2
...................................................................
3
3...................
.......
..........................................
.
8
4
..................... .
..........
..................
..................
25
5.......................
...............................................
12
6.
...............................................................
........
10
Total............................................................ 58
NOTE.-Because
interviewsrom
even of
these
ubjects
were
used to
constructhe
scoring
manual,
he
blind
ample
usedfor
ata
analysis
ncluded
nly
1
subjects.
TABLE
6
TIME
1
MORAL
MATURITY
SCORES
FOR
DROPOUTS
FROM
SAMPLE
AGE AT
TIME
1
(Years)
10 13 16
Mean
MMS
at
time
1
of
subjects:
Dropping
out after
ime
2............... 203
(2)
Dropping
out after ime
3...............
Dropping
out after
ime
4...............
176
4)
Dropping
out
after
ime
5...............
191
5)
Mean
MMS
at time 1
of
dropouts..........
190(11)
Mean
MMS
at
time
1
of
total
sample.......
189
21)
240
2)
227
(2)
234
4)
236(17)
260
5)
249
5)
255
10)
262
14)
17
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
23/129
MONOGRAPHS
scores t
time 1
than thosewho
dropped
out
later.
Thus,
mean increase n
moral
maturity
coresover time cannot
be
attributed o lower
stage
sub-
jectsdropping ut ofthe ongitudinalample.
MORAL
JUDGMENT
NTERVIEWS
Subjects
were interviewed irst
n
1955-56
and at
3-4-year
ntervals
thereafter.
he
last
set
of
nterviews
including
5
subjects)
was
completed
in
1976-77.
Although
he
probe
questions
iffered
lightly
rom
ne
testing
time to
another,
the
same
nine
dilemmas
were
used each time. These
dilemmas were
later
used
in
constructing
he
Standard
Issue Interviews
and
represent
n effectheuse of all threeStandard Forms
A,
B, and
C)
at
each
testing
ime
for ach
subject.
All
interviews ere
conducted
ndi-
vidually
nd
were
tape
recorded
nd transcribed.
MORAL
JUDGMENT CORING
Interviewswere
scored
according
to
the Standard
Issue
Scoring
Manual
(in
press)
forms,A, B,
and C.
(See
App.
D
for a
discussion f
scoring
rules.)
Because the
number
of
interviews o be scored
was
very
large,
differentater cored ach ofthe three orms
, B,
and C. Allthree
raters
were
highly
xperienced,
nd
reliability
mong
themfell
within
he
limits
iscussed
n
the
reliability
ection
Chap.
III
below)
ofthis
Monograph.
None of the
interviewers
articipated
n
coding
the
interviews.
coring
of
all
interviews as
done blind.
That
is,
raters
oded
the
responses
o
each
dilemma t
each time
without
nowing
he
subject's ge,
identity,
esponses
to
(or
scores
on)
other
dilemmas t
the
same
testing
ime,
or
responses
o
(or
scores
n)
any
of the
dilemmas
t
other
esting
imes.
18
This content downloaded from 202.38.182.7 on Thu, 21 Jan 2016 10:48:37 UTCAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
8/19/2019 Monogrph longitudnal study.pdf
24/129
III.
RELIABILITY ND
VALIDITY
OF
STANDARD
ISSUE
SCORING
RELIABILITY
Reliability
ata
of several
types
were
compiled
for he
Standard
ssue
instrument.
s
the
following
esults
ndicate,
he
instrument
as
proven
o
be
highly
eliable.
Test-retest
eliability.-Test-retest
oral
udgment
nterviews ere con-
ducted
with 43
subjects
using
form
A,
31
subjects
using
form
B,
and
10
subjectsusingbothA and B. No test-retestata have been collectedfor
form
C
as
yet.
The same forms
were
used
at
times
1
and
2
with
onditions
of
testing
eld constant
nd
intervals
etween
ime
1
and
time
2
ranging
from to 6
weeks.
Subjects
were
chosen
from
mong
volunteers
n
several
Boston
area
elementary
nd
high
schools,
olleges,
nd
graduate
schools.
The
college
and
graduate
chool
tudents
ere
paid
for
heir
ime.
Subjects
ranged
in
age
from
to 28
years
and
approximately
alf
were
male and
half
female.
Interviews
were
scored blind
by
two
raters
using
the
Standard
Issue
ScoringManual (in press).
Test-retest
eliability igures
re
summarized n
table
7.
As
shown n
that
table,
correlations
etween
ime
1
and
time
2
for
forms
A
and
B
are
both n
the
high
nineties.
ince
the
correlations
ould be
veryhigh
without
much
absolute
agreement
etween
coresat
time
1
and
time
2,
we
have
also
presented
ercent
greement
igures.
or
almost
ll
subjects,
he
scores
on
times
1
and
2
were
within
/3
stage
of
each
other
one
step-from
1
to
1[2]
or