Vleta-Theoretical Perspectives of Investigators Publishing Parent

Vleta-TheoreticalPerspectivesof Investigators
PublishingParent-ChildRelationsStudies
in Child Development and Journal of
Marriage and the Family
RICHARD
C. ENDSLEY AND MARILYN
R. BRADBARD.
GREG I-ANG.
Previotts research has docwnented that investigatorsof parent-child (p-c)
relations researchwho publish either in Child Development (CD) or loumal of
Maniage and the Family (IMF) are from essentially two different researclt
communities that function in virrual isolation from each other. The present
study attempted to test the l4tpothesisthat the differencesin methods used and
oge goups of subjects sndied can be accounted for by meta-theoretical
differences that deive from child specialists' closer links to the discipline of
psychologt and family specialists' closer links to the discipline of sociologt.
However, the results indicated tltat the two groups' conceptions of vaiable
ntolait,v and tlrc sources of extental causation were more notable for their
similaities tlun tJrcir differences. Further, the grottps differed on the freEtency
witlt wlich reciprocal interaction tneasurcs were employed, even though botlt
lmve long held similar views about the importance of such assessments.
Tlrcrefore, it wqs concluded thqt methodological dffirences between the two
researclt groups are only loosely diven by meta-theoretical considerations.
Instead, tlrcy inhere in the methodological and/or substantive traditions of their
respective root disciplines.
Our previous research (Endsley & Brody, 1981; Endsley, Bradbard, & Lang, 1988)
documents that child and family research specialists have functioned in virtual isolation
from each other, though some signs of rapprochement are evident. Specificall,v, we
demonstrated that researchers who publish parent-child (p-c) relations research n Child
Development (CD) versus those who publish p-c relations research 1n the loumal of
Manioge and tlrc Familt (IMF) (a) generally reside in different departments; (b) rarely
belong to each other's professional organizsllsas; (c) rarely publish in both journals; and,
(d) employ substantially different methods and subjects.
With regard to the latter finding, in recent years p-c researcherswho publish in
CD are highiv likelv to use observational strategies,as well as questionnaire/interview
strategies. These latter (self-report) methods are tvpically used in combination with
* Richard C. Endsley and Greg Lang are in the department of
Child and Family Developmenr,
University of Georgia. Mari$n R. Bradbard is in the department of Family and Child
Deveiopment, Auburn University. Requests for reprinrs may be sent to Richard C. Endsley,
Department of Child and Family Development, Dawson Hail - UGA, Athens, GA 30602.
[Famity ScienceReview,Vol. 2, No. 2,May, 1989pp. 187-2041
L87
other methods such as experimentai, observational, and/or tests-&-measures procedures.
In contrast, p-c researchers publishing n IMF continue to focus almost exclusively on
questionnaireTinterview stratLgies, and rarely combine these procedures with any others.
Further, CD researchers are much more likely to study infants than any other age group
in their p-c research, while IMF researchers are much more likely to study adolescents
Among the possible reasons given for the
and adults than any other age goup.
distinctively different methods and subjects employed by these two professionally isolated
$oups wai the suggestion that they have fundamentally different meta-theoretical
orientations towardlheir subject matter (e.g.,Endsley & Brody, L981). These purported
differences in meta-theoretical perspectives were examined in the present research.
Metalheoretical
Imptications of Ps,vchologicaland Sociological Percpectives
It is clear that child studies historically have been more closely identified with the
discipline of psychology, while family studies has its intellectual roots more extensively
grounded in socioiogy (Lerner & Spanier, 1978). Among the differences conventionally
fugilighted to distinguish the world views of the two disciplines is that psychologists_are
iniUnea to examine human behavior at a more molecular (less molar) level than
sociologists (Larzelere & Klein, 1987). Moreover, envitonmental causation of behavior
from a psychotogical perspective favors the examination of the individual differences and
interactional patterns of social agents, particularly those in close temporal and physical
proximiry with the targets of their influence. In contrast, a sociological perspective
favors the examination of temporally and physically more removed social structures and
contexts on the targets of influence. Thus, from a psychological perspective, when
parents are the focus of inlluence on children, the parents' individual qualities and
interactional patterns with their children would be regarded as particularly salient
variables 1e e;amine (Maccoby & Martin, 1983). Conversely, a sociological perspective
would emphasize the "background" of parents (e"g., demographic factors such as culture,
race, social class, family structure, employment status) as particularly salient variables
(Peterson& Rsllin5, l!$l)"
These different disciplinary perspectives may be manifested in the different
methodological styles favored by child and family specialists. For example, it is
reasonable to expect that a psychological view that stressesa more molecular perspective
and the analyses of temporally and physically close causes would demand the more
fine-grained, up-close procedures of experimentation, observation, and tests-&-measures
of individual differences. Conversely, the more giobal and temporally and physically
more distant causal inlluences of a sociological perspective might permit, or even favor,
the use of self report techniques such as questionnaires and interviews.
If the different meta-theoretical perspectives fbund in psychology and sociology
have the distinctive methodological inlluences on chiid and family specialists as thev
study p-c reiations, it follows that: (a) the independent and dependent variables found
in p-c studies published n IMF would be relatively more molar than those found in p-c
studies published n CD; and (b) environmental factors presumed to influeuce p-c
relations would be more demographic and less psychological in nature (i.e., less likely
to be individual difference and interactional measures) in the case of p-c studies
published n JMF than iu CD.
188
Familv Science Rwiew
May, 1989
illetlrcci imp
In adc
psvcholoer'
methodolog
deveiopme
on whethcr
theoretical
di-fferendarc
Miller, 1978
families ca.o
informaLion
1978). The
those that r
parents anc
Nesseircac
Based
of p-c relau
more age g
this is not o
the child dr
approached
NIaccobv,l!
relations ctu
As ano
influence re
of multicarr
studies r*'er
in familv stu
nature of fa
1951). Conc
relationship
assumption
these theorc
be a-nincr-*r
interacrion i
or both ;or
metho,ioiog
m et a- t heor
Fr om .
throuqh 19!
empuicei ar
a relation<.\
( Endsier '&
articies.rrrn
one -- seeFr
in a time pc
Mav, 1969
Method Implications of Developmental, Multicausal, and Reciprocal Influence Perspectives
In addition to the meta-theoretical implications of being grounded either in
psychology or sociology on research strategy, there are perhaps even more obvious
methodological implications of adopting a "developmental" perspective. Students of
deveiopment, whether of individuals or families, may disagree on what develops, or even
on whether our understanding of development can be advanced by using the term in a
rhe-oretical sense (that is, by referring to hypothetical processes that emerge,
differentiate, and are reorganizsd over time in systematic ways-- Klein, Jorgeoson,k
Miller, 1.978). However, all accept the notion that systematic changes in individuals and
families can be detected over time, and that these observed changes can provide valuable
information for testing our theories of parent-child relationshlps (Leiner & Spanier,
1978). The most widelv accepted forms of developmental methodology, therefoie, are
those that compare different age groups of subjects (in our case, different ages of
parents and/or children) either cross-sectionallyor longitudinally (Baltes, Reese, &
Nesselroade,1977).
_ Based simply on the names of the journals, it is reasonableto expect that studies
of p-c relations publisheci in CD would more often involve comparisons across two or
more age groups of parents and/or children than would studies published vIMF.
That
this is not necessarily a foregone conclusion, however, is suggested by criticisms within
the child development research community that p-c studies traditionally have been
approachedmore from a socielizationperspectivethan a developmentalperspective(e.g.,
NfaccobY, 1984). These.critics have, correspondingly, called for more studiel of how p-g
relations change over time and over important age periods in the life span.
As another means of assessingthe extent to which our meta-theoretical positions
influence research methods among p-c researchers publishin g n cD and JMF, the use
of multicausal data analytic techniques and measures of reciprocal interactions in p-c
studies were examined. In this instance, both theorists within child development ind
in familystudies have long_acknowledgedthe multicausal and reciprocal (bi-directional)
n1lule of family relationships, including p-c relations (parsons & galei, 1955; Sears,
1951). Concomitantly, there has been concern that our data analytic techniques and
relationship measrues often fail to reflect our commitment to these meta-thioretical
a5sumptions (Larzelere & Klein, 1987; Lerner & Spanier, 1978). To the extent that
these theoretical exhortions are being taken seriously by p-" researchers, there should
be an increased use of multicausal data analytic techniquei and measures of reciprocal
interaction in the p-c studies over time in both journals. However, to the extent that one
or both journals t'ail to reflect such trends, then one could conclude that the
methodoiogical trends are governed by tactors other than the above mentioned
meta-theoretical considerations.
METHOD
From a total of 7,42 p-c studies published in CD and JMF from their inception
through 1986,.308 @r.SVo)were selecredfor anaiysis. A p-c study was defined as "any
empirical article containing at least one measure that explicitlv asiessedsome aspect of
a relationship berween at least one child and one parent within the families studied"
(Endsley & Brody, p. 6). At least 20 articles, but no more than one- third of the total
articles,were selectedrandomly from each time period (ail 5 year periods except the last
one -- see Figures 1-8). The only exception was when less than 20 articles were available
in a time period; in this case all were exemined.
May, 1989
Family Science Reyiew
189
Assessingtlrc Molaity of Independent and Dependent Vaiables
Assessingtt
The analyses required examining the molarity of the independent and depenclent
=
variables rated on 2 3-point scale, in which 7 = molecular,2 = intermediate, and 3
actions
behavioral
represent
variables
dependent
all
molar. In p-c ,ese-.i, virtually
at some levil of molarity. Indepindent variables can also be behavioral, but aiso might
health) conditions.
economic) or internal
represent other external (e.g.,
'ii
-(e.g.,
considering behavioral actions or other conditions,
Nivertheless, whether one
molecular variables are characterized by beng highly discrete events lasting only a few
usually relatively
moments (e.g., seconds, minutes) and by virtue of their simplicity,
-are
essyto sumilortze operationally. Behavioral examples miglt include the brief smiles
by infants while being held by their mothers, or physiological reactions such
"rniUit"a
as momentarry heart rate. Nonblhavioral (independent variable) examples might include
a pin prick of a child's finger to draw blood for a hemocritic screen or the visual
ptiretrie of a non-interactive stranger during the few moments a child's attachment to
Lis mother is being indexed in a laboratory context.
On the other end of the dimension are the truly molar behavioral events and
conditions which are characterized by being topographically diverse, eristing over large
time spans (e.g.,months and yean), and by virue of their conceptual compluitt't are often
dfficiU to iulnmaize operationally. Behavioral examples might include all the actions
that could be summarized in the roles of being married, working, parenting, acquiring
a given level of education, or reflecting one's ethnic or cultural background.. They also
inilude "styles" of functioning such as authoritative parenting or hyperactivity. Other
nonbehavioral molar conditions include factors such as age, gender, health status,
geographic location, and alternative child care arr:rngements.
Finally, by default, intermediate variables are those that fall somewhere between
being highiy discrete, temporally fleeting and descriptivefV obygus and those that are
topJgrap-hiially complex, of longstanding duration, and descriptively vague.- .lehavior
of intermediate level variables include a wide range of actions which, while
"*-u--pt"i
tuki"A various behavioral forms, are intended to serve the same goals within a relatively
circuiscribed time frame (e.g., "5s61hing"behavior of mothers toward their infants'
crying, aggressive exchangesbetween adolescents and their parents, etc.)
Assessing Three Categoies of External Causation
Family Science Review
a combrinonlv othcr
purposes o
index mea
Assessing.l
Finall
lsshnique
on one or I
reciprocei
contempo
judgem en
empioveci
terms of r
measure (\
as both a c
ilfeqsuremt
Reiial
articles. L<
just descri
the asse<s
91Vo, and r
Ivtolarin'ol
p.ati.
The analyses also included three categories of external causation; those that were
demographic, psychological, or physical in nature. A determination was first made that
the causal (i.e., independeut) variables were regarded by the investigator(s) as being
external (i.e., contextual) in nature rather than originating within the subjects themselves.
Thus, studies that inciuded no externai causativevariable, but instead, confined their
analysisof causation to internal factors (i.e., biological and/g1 psychologicalattributes
such as I.Q. or hyperactivify that are regarded as stable individual differences) were
excluded from the analysis. Researchers making demographic contextual comparisons
analyzed molar conditions such as parent education, place of residence, or family
composition. Studiesexamining psychologicalcontextualvariables focusedon behaviors
and/or attitudes exhibited by one or both members of the p-c dyad (or, in_somecases.,
on whole family measures) that directly indexed the interaction and/or relationship(s)
existing between the parents and children. Finaily, studies exa.miningphysicalcontextual
variabfus were dealing with explicit and readily definable elements of the immediate
physical environment such as room size, or number of climbing toys available. lach
;external causation"studywas evaluatedfor the presenceor absenceofeach ofthe three
classesof contextual variables.
190
A rtu
were collc
May, 1989
independer
jucieed lo
documentt
in the perc
;n,..1,
^ - -; . .
ind.icaresli
\Ic.re
oi the i-uci
t\\'o journa
llsures do
consisteot-h
NIav. 19s-9
Assessing tlrc Deveropmentar Methodorogicar perspective
of Each Sndy
A study was judged to have a developmental
methodological
perspective if data
werecollecredon two or moreagegroupstr.nilar*
by usingoneor
a combination
of longitudjnal,
cro-ss-Iecti;nal
andrerrorp*fi;
-i7lr"i-""o,
IJr"-.n strategies.
The
only other provisowas that the multiple time point
assessments
were attemptedfor
of indexingdevelopmentalchaneeandfor Uf" tr-.iti""s
rathe,
P_Tl_:i::
rhan sisr'lr,o
rndex
measurementstability.
AssessingMurticausarand ReciprocarCausarAnarytic
Techniques
Finallx studies were examined to determine
whether they used data analytic
lsshniquesthat evaluatedthe effectsof tw_oor
more independentvariablesoperating
on one or more dependentvariabies,aswell as whether
uatiuur"J*"r" assumedto have
reciprocal(bi-directional)influences.
Multicausalmodels*" u.**"0 to underliemost
conremporarydara anarltis lsghniques(e.g., MANovA-."ltipr"
."gr"r.iooi,-;;;
judgemenrsof murticausalanalyses*"."
.Judty deter;in;jw;;rsuch
procedureswere
empioved. The measur:-:oj of reciprocalirfu;"";;;;
study was def,rnedin
d; -giveo
terms of whether aay statistical proiedure was
applied to"a
p-c interactional
measure(whether behavioralor aititudinal) in
ordJi-to or""uia""uuluate
this measure
- '
as both a causeand an effect on otner i"te'ractioout
-"u.ui"s,
ilIeasurenrentRetiabil ity
Reliabiliw was assessed
by _askingfwo judgesto independenrry
rate a set of 30
articles,15 from CD and,ts froitur;""bilil?,;..og
,h';-8d;;;"ach of rhe measures
just described.percenr.ofagreement
5ue.-tueeigh,pi."r""rry iJ.cribed measuresprus
rrm':&::if;t:::;X?tudvconiainedanvanarvsis;rext;;;;*c;;ffi"!J;
RESULTS
Molaity
of Independent and Dependent Vaiabtes
Preliminarv insoection of the data revealed
that a high percentage of both the
independenr
ani dependeot
u-_iu[r"J-G.;:sG- B; ;'% ;5;.ffiely) exam
insdwere
judgedto fall at levelz
j = molar). Figures1
1t".* tnui )"1'mole"ur8r,_
and 2
docrrmenr
thisfact,andat g" s1-9 ti-", ioai.utethatihere
il;;;"
a modestdecrease
]evel2-molaritui*i"ur1over.rime
in both.journals.especialv
T-l^o:t-:t-*t"g:
?f
for
mdepencienr
variabies.
This decrease percentage
indicatesthat the variabilityi"
in
of level'2 measures,therebv,
h*'io.."u."d slightlyover rrme.
-orariry
More pertinentto the presentstudv,Figure:
1
of the independentand.depi"a."t u-iuui"s-3-ptov"a
ry9 2suggestthat the molarirylevel
in the p-c srudiespublishedin the
fwo journalshas remainei q"it" ,i-il- Lu",
time. tn other words.the data in these
supporr tne'nvpoine;; ;;; JMF uariabres
are operationarized
lx,::^9:,'lf more molar
ar a
consrstenily
leverthan cD variabres.This .o".t*iJoluas
supportedby
May. 1989
Family Science Re'view
L91
:
i
:
!
g
\
!
!f
@
6
@
tr>
*:
cD
<c
t
tr
r.1
.:*
:-
at
F,*
I
o
iq)
x>
\t
1)
::
l)
o
(D
v!
C c|
0)'=o
6E
Qrq
aO
.9a
=()
o
tr'r
oL
?td
o
oE
o>
EE
Eo
(J-
€E
?tt r
EX
-9
o
(o
-i)
--
'J
=a
o
v
::
I
o
n
<J
tl
tII l
at
Q't
?t
)=
o
\r
:>
.=
rN33U3d
-:
Q*.
-:
9?ol
!a
.j
-j
o
:
J-
L9Z
^-=
:t
tr)tr
c)
=7
;3
=e
,-\
l!
=CJ
qQ
>:
5
0
(DE
s(!
.!;
:e
Family Science Review
May, 1989
l\lrr'. l9S9
,i-
:
a.'
dt
,:>
:\
I
!5
:U
0
l!
-tr
o
f;
:x
o
(o o
o
o
(o E.
I
;
c
o
i\e
q)
co
it r
-:
o
9'r
a)
FL
*=
T
or
o
n
o
I
UJ
L
td
u.) F
q)
OE
2-
!t!
-i J
,€
EO
(J-
OI
ll
II
tl
tl
tl
i>
<i )
':
tE
Jn
\r
o
ro
I
ol
o
?F
)t
:. >\
o
o
9
o
(o
o
v
o
(\l
1N33U3d
r:
9o
:i
':
lvIay, 1989
Familv Science Re\.iew
r93
inspection of the mean molarity level of the independent and dep.endentvariables in the
rwo journals from 1950-86 and r-test results that were nonsiguifrcanl (Ols>.0S;-. _fn_e
mean molarity of the independent variables were 1.93 and 2.01 fot CD and IMF,
molarity of the dependent variables was 2.02 and 2.11,
respectively; [h"
-euo
respectively.
Type of Ertemal Causal Vaiables Used
a4
Figures 3 to 5 summarize the percentage of p-c studies over time in each journal
that employed demographic, psychological, and physical variables, respective"ly,- as
external iauses of p-c ielations. While Figure 3 appears to support the hypothesis that
demographic variables are more frequently employed nIMF p-c studies than in CD p-c
studiei, the differences evaluated via chi square test are small and nonsignificant for both
the 1975-86period (57% vs.70Vo). In general.
the 1950-74period (54Vo vs.69%) aJold
while there appear to be modest, sychronized fluctuations from one decade to the next
in both journals, there is no clearcut long-term trend other than demographic factors
have been and continue to be examined as external influences on p-c relations by a
majoiry of the investigators publishing in both journals.
In contrast, the most notable feature in Figure 4 is the dramatic increase over time
in bothjouraals in the use of psychological variables as sources of external influence on
p-c relaiions. For example, during the 1950-74 period, only 1!% andl6Vo of CD and
p-c studies, respectively, examined psychological variabies as external causes.
lmf
However, by the 1975-86 period, 70% and 49Vo of CD andIMF p-c studies, respectively,
did so. While for the latter time period the chi square tests revealed the difference to
be significant in the predicted direction (p<.02), the trends for the two journals in
Figure 4 are more notable for their similarities than for their differences.
Finally, Figure 5 clearly documents the fact that physical variables are rarely used
as external calrses of p-c relations in either journal. However, when used, they have so
far only appeared in CD studies.
c'
F
=>
::i
t?
\':':
S=
.-=
9.!
,3
=a
* ;i
.j
'J
--
Percentageof Developmental Parent'Clrild Srudies
Uniike the measures of variable molarity and rype of external causeswhich yielded
few dilferences between the two journals, Figure 6 clearly supports what was confrmed
by chi squiue test (p<.001); namely, that CD publishes a higher percentage of
developmental methodology p-c studies (i.e., p-c studies comparing two or more age
groups) than JMF. llowever, it is also evident from Figure 6 that the percentage of
developmental methodology p-c studies in both journals has increased over time, so that
in the 80s "r,&IF c;.n now be regarded as a developmental journal in its own right.
Correspondingly, the 80s is the lrst decade in which the majority of p-c studies
published rn CD can be said to i-ucorporatea developmental design.
Use of Multicausal and Reciprocal Analytic Strategies
Family Science Review
xx
==
--:
-1)
Figure 7 summarizesthe percentageof p-c studies over time in both journals that
employed multicausal data analytic procedures, whils Figure 8 summarizes the
percentage of p-c studies that used measures of reciprocal influences to evaluate the
p-c relationship. As indicated in Figure 7 and confirmed by chi square analysis, during
the first 20 years of IMF as a research journal (1950-69), p-c researchers were
significantly less likely to employ multicausal data analytic techniques than were those
L94
T3.
May, 1989
I
May, 19bT
sr
co
I
5
@
F
A
or
i
c?
lfl
*?
>
sr
=
o
o
9o
tJ-
o
vii
rE
=^
;
q
-{,
.5
.9
co
o ':o
().E
.gs
6- >
E
l6
v t Eq)
j
oE
=r =
Eo
CJ-
<)l
?tl r
!...
$ )=
o6
x:
rx
^oY
o
g
(o E
trJ
oo9ut
o=
l r 'F
o,
.Y
-X
0
o
o
clr
I
o)
\r
I
o
tf,
tl
tl
tl
al
o
(o
o
sf
1N33U3d
q*
o9.)
May, 1989
Family Science Rer.iew
195
)
.,
:
a
't
\
!
!1.
@
5
i-
c
@
Y'A
:-, >\
tr)
gt
I
\r
o
.-:
>.::
o
^-=
I
:*=
o
t!
o
-:9
j :o
ig
\c)
:E
;
c
0
o
gr
-!?
=>
qY
qa
, --=
= co
E-
o
sr
).
-€
*€
-x
I
o
fc)
tt
-:
=3
-i !
ll
tl
tl
tl
tl
:h
'-X
v'J
->
o
uo
!S
Eo
(J1
i=
)e
h.E
":ra cC.
?ur
o>
u,) i-
i?
at
o
o
o
(o
;=
o
\r
-:
!-
|Ji )
lrj
61 L
co
o) ':Xo
o>
-!
o><
9o
oo
(oE
IN33U3d
=!
9:)
."
C)
196
Familv Science Review
May, 1989
Mav, i9S9
:
-:
'L
v
s
:
!
q
@
6
@
gr
I
r.-)
\r
!;
I
o
o)
(o
:a
-:€
:)
I
(o
-:e
u9
co
2-
I
o
F
;
b
()-
c)
uo
;X
3=
UJ
|n =
:o
q>
E,
td
o oro
q)
crr
9';
HG l
o
EO
L) -
??
tl
( li
p5
!g
()€
tl
tl
tt
al
o
o
@
t]r
o
g
O"
g;
IN=3U3d
9o
)
!
:'
May, 1989
Famiiv Science Reyiew
197
I
:p
t
q
@
\)
6
@
ol
r.)
(r
5
E
o
o
o
(o
;
o
(o I(E
I
o
u
*
I
e
c,
o
a
=
UJ
o L
lo
-Co
cn
( D' -
I
o
n
qC.
t'=
E-
ii
.2=
:
LrJ
=
F
-tr> ,
>o
c)
H( t
oj
€o
qEl
:,) ;
=€
rl)
QI I T
Cq2
= .J
I
L,,-
cii
q--
tl
tl
tl
c-
al
:Jh
YX^
o
\r
o
!- c
3a
o
o
o
(o
o
q
P=
rN=3U=d
:>
'n).
t=
?l
Pn
-=
-!
1r
aI
198
Family Science Review
May, 1989
N{av. 1959
:\
g-
)
.)1r
\
c
u
tJq)
*
o
+
;\
tr>.
1,
CO
(l)
o
(r
o)
crl
lrJ
oUJ
a
q)
6
?5
Eq
o
€.g
uo
EE
9?(g
EO
(J
ECI
etl t
a;
tl
ll
':
o
\r
o
I
ro
tl
at
O::
Utr
aX
AA
qQ
i) i)
o
(o
o
s
rN33U3d
=E
9 :.)
rq)
ii
May, 1989
Family Science Review
L99
publishing in CD (p<.05). However, since the early 70s virtually all p-c researchers i-n
both journals have employed multicausal techniques in analyzing their data.
In contrast to the almost universal use of multicausal data analytic techniques, the
use of reciprocal measures has become common only in recent years, and only in p-c
studies published n CD. Specifically, from 1950-69, only one study appeared in each
journal that employed reciprocal measures. F{owever, since that time the number has
risen sharply r\ CD, while remaining virtually absent n IMF. While Figure 8 clearly
documents, and chi squ:ue analysis confirms, the rise in CD since 1970, it also reveals
a sharp and significant decline in the use of these measures among CD studies published
in the last two years assessed(both p's<.001).
F
DISCUSSION
Meta-tlrcoretical Perspectit'eversus Substantiver/Methodological Trariitions itt Psycltologt
and Sociologt
While some of the trends support the hypothesesunder investigation,the general
impression of the results more clearly supports the view that the fwo groups of p-c
researchers do not have fundamentally different conceptions about the nature of their
subject matter. Specifically, both (a) conceptualize their variables at a similar level of
molariry (i.e., almost always at level Z); (b) continue to examine demographic factors in
a majoriry of their studies; (c) have become increasingly preoccupied with psychological
factors as causal agents; and (d) ignore the role of immediate physical circumstances as
causal agents io p-c relations. Further, in matters where there is meta-theoretical
consensus, i.e., the virtues of multicausal analytic strategies and analysis of reciprocal
influences, there are differences between the two sets of studies in the readinbss to
implement these viewpoints into research practice.
If we can conclude that the two research communities actually assume similar
meta-theoretical positions regarding their subject matter, what are some of the
alternative po-ssibfitiesavailable for erplaining the method and the research subject
differences of the two groups? one possibiliry, already raised in a previous study
(Endsley, et al., 1988), is that the fwo groups don't really study the samt phenomena,
even though they are all studying p-c relations. Specifically, p-c researcheri publishing
in CD have becom€ enamolsd with infant-parent relationships, and a substantiJ
percentage of these
.studies are concerned with parenting behavior that promotes
cognitive and linguistic development.l Conversely, one simply doesn't find sludies of
cognttiYe or linguistic development in IMF, regardless of the age of the child subjects.
A similar contrast can be found in the more frequent studies in JMF than in Cp ot
adoiescent-parentinteraction qualiry (e.g.,supportiveness,control, power). Thus, it mav
be that these and other substantive differences account for the different .esearcL
methods. Obviousiy one cannot interview hfants as one can adolescents. Further,
cognitive and linguistic assessmentsmay require the more standardized and controiled
methodologies of experimentation, observation. and tests-&-measures than do
assessmentsof the degree of supportiveness,control, or power exerted bv parents over
their teenage children.
However, to some extent, the existence of different substantive topics in the study
gf p-9 relations still begs the question, because to some unknown degree, an
investigator's choice of,a research topic is prompted by his/her ready familiatiry wittt
particular research methods. Thus, as we have argued elsewhere (Endsley et al., iggg),
methodological "spillover" occurs from one generation of investigators to the next, re--
200
Family ScienceRwiew
May 1989
(
.-=
{,) =')
oC
I
==
=l
-a
i
J
lv{av, i9E9
r&
q)
F
$
q
@
5
ao
o)
-
tr)
3
E
Y
0
t!
o
o
:l
p
o)
(o
;
AO
E
0
-:
n
co
U9E.
o
or
' uJ
ao-
9'=
.-7
I
9
u,
=ct
U=
6->
E-
ull
l-
(D
oE
HC)
ES
Uk
o
sr
Ec,
( Ja
={2
!()
?ll ?
aQ
I
o
|a
tl
tl
tl
ol
E'-
=3
T. F
!r
:)
c9
o
o
oo
(osr
IN33U3d
' v5
i) -
?i)
cc
3:
May, 1989
Family Science Revierv
201
flecting to some degree the disciplinary roots of the earlier generations, even as new
substantive domains emerge. As has been previously documented (Endsley & Brody,
1981-;Endsley et al., in press), the disciplinary roots of psychology and sociology are
strongly represented in p-c researchers who publish in CD and IMF, respectively. The
"field" of social psychology probably better represents the distinctive methodological
impact of psychology and sociology than any other. Indeed, there is not one field of
social psychology but two (Stryker, !977), and the distinction between the two social
psychologies is heavily methodological. That is, one field is inhabited by "psychological
social psychologists" who essentially do experimental laboratory research, while the other
field is inhabited by "sociological sociai psychologists" who essentially do field
questionnaire research. That there is such a distinction between two groups of
investigators supposedly interested in the same phenomenon is a testimony to how
methods, originally characteristic of distinctively different and earlier estabtshed fields
of inquiry (e.g., experimental psvchology versus demography) spill over and continue to
define and distinguish the disciplines of psychology and sociology even at the juncture
where the nwo now overlap substantively.
Integrating Child and Family Sadies Tltrough Developmental/Interaction
Research Design
That there has been an increase over time in the study of developmentol p-c
relations in both CD and.JMF indicates a growing responsivenessin this area of research
to an intellectual tradition that emerged primarily outside of mainstream psychology and
sociology in the 20s and 30s (Cairns, 1983; Sears, 1975). The developmental perspective
with its attendant research methodologies (e.g., cross-sectional, longitudinal, &
retrospective self-report methods) was, therefore, well established before either child
study or family study specialists took up the investigation of p-c relations in the late 40s
and 50s (Endsley & Brody, 1981). We believe this new developmental thrust is a further
sign (in addition to those mentioned in an earlier study -- Endsley, et al., 1988) that a
rapprochement is emerging between child and family specialists in their respective
research enterprises (Kelly, 1986; Lerner & Spanier, 1978).
Finally, the present findings regardirg the trends in the use of measures indexing
,'€g^ip',5\f,| itiltclac.li3n'd9-s9,'I€ connffe^lJ. Lhe emerge,re,ars,rrc,h rreffurff
rn Ct.bc,t'
not'nIMF, despite a comparably long history of concern about the importance of such
processes,suggeststhat child developmentalist55tudying p-c relations have been more
responsive to critiques about the mismatch of research to theory than have family
speiialists. (For other evidence of differential responsiveoess [o criticisms aboul
mismatches of research to theory, see Endsley et al., 1988). This responsiveness,in turn,
appears to be highly correlated with the different methodological preferences of the two
research groups. Investigatorswith predispositionsto employ observational lsqhniques
would seem to be in a better (more responsive)position to assessreciprocal interactions
than those using questionnairer/interview strategies, for observational technologies
necessariiybring researchersface-to-face with actual reciprocal encounters. This "up
ciose and personal" experiencenot only highiights the prevalenceof reciprocifv in hrrman
relationships,but also offers insight into the operational procedures whereby reciprocal
influences can be assessed(Cairns, 1979,Klein et al., 1978). To further emphasizethe
impact of method on the studv of reciprociry, inspection of the present data revealed that
the drrmatic drop in studiesof reciprocal interactions in CD during 85-86 was associared
with the equally sharp rise in questionnaire/interview strategies in that period.
Specifically, none of the studies employing only questionnahe/interview methods in that
period examined reciprocal effects, while 4O percent of the studies employing onlv
observational methods did so.
202
Family ScienceReview
Ma.v.i989
This i
of recipro
require a&
gnrmporttl
sp n n q 3 !d r
Balter P B
Inodt
Cairnr R. B
Hir.isd:
Cairns. R. I
tianab
Encislo.'.R
sDec:3:na.r:<
Encisln'.R.
1e1'e3
:'ne::ic
Kellv. H r1
Duct (
Klein, D \l
rec'tpr
anci .rd
Larzelere. R
i Ecs. r .
Lerner. R. \
,4 life-s
Maccobr'. E,
Macco['. E.
interac
PeBOrr
Parsonl T. ,
IL: Frc
Peterson.G,
IC Srer
Sears. R. R,
Psrrlra
Strv-ker.
S tl
rel*'an
1.
Bas:ri
r'iis,co
ar'.rc]e
re s33r
3s r3!
anicle
corDEl
P€cr s
d.. iq
each r
retlect
Ma-v",i9E9
This is not to say that self-report methods are inherently unsuitable ir the study
However, it would appear that employing such methodology will either
reciprociry.
of
require adopting a synchronous view of reciprociry where causal analyses are deemed
unimportant; or, alternatively, conducting causally oriented p-c studies using longer time
spans and variants of longitudinal research designs (Klein et al., 1978).
REFERENCES
Baltes, P. 8., Reese, H. W., & Nesselroade, J. R. (1977). Ltfe-span developmental psychologt:
Intrcduction to rcsearch methods. Montery, CA: Brooks/Cole.
Cairns, R. B. (Ed.), (1979). The analysis of social intemuions: Methods, issues,and illustmtions.
Hillsdale, NJ: Lawrence Edbaum.
Cairns, R. B. (1983). The emergence of dwelopmental psycholory. In P.H. Mussen (Ed.),
Handbook of Child Psychologt: VoL 1. History, theory, and merhods. New York: John Wiley.
Endslev,R.C.,Bradbard,M.R., &Lang, G. (1988). Protessionalisolationofchildandfamily
specialists studving parent-child relations: Update and further
anaivses.Famiiv Science Review, 1, 225-238.
Endsle_v,R. C. & Brod.v, G. H. (1981). Prot'essional isolation of child and family specialists as
rwealed in a time series analysis of parent-child relations research
methods. Famiiy Relations, i0, 5-15.
Kelly, H. (1986). Personal relationstrips: Their nature and significance. In R. Gilmour and S.
Duck (Eds.), The emeryingfreld of perconal relationships. Hillsdale, NJ: Lawrence Eribaum.
Klein, D. M., Jorgensen, S. R., & Miller, B. C. (1978). Research methods and dorelopmental
reciprocity in families. In R. M. Lerner and G. B. Spanier (Eds.), Child influences on marital
and family intemction: A life-span penpective. New York Academic Press.
Larzelere, R. E. & Klein, D. M. (1987). Methodology. In M. B. Sussman and S.K. Steinmetz
(Eds.), Handbook of marriage and the fatnily. New York: Plenum.
Lerner, R. M. & Spanier, G. B. (Eds.), (1978). Child influences on marital and family interaction:
A life-span pecpective. New York: Academic Press.
Maccoby, E. E. (1984). Socializationand developmentalchange. Child Development,55,3L7-3?3.
Maccoby, E. E. & Martin, J. A. (1983). Socialization in the context of the family: Parent-child
interaction. In P. H. Mussen (Ed.), Handbook of child psycholog: Vol. 4 Socialization,
perconality, and social development. New York: John Wiley.
Parsons, T. & Bales, R. F. (1955). Family, socialization and the intemction ptocess. Glencoe,
IL: Free Press.
Peterson, G. W. & Roilins, B. C. (1987). Parent-child socialization. In M. B. Sussman and S.
K. Steinmetz (Eds.), Handbook of mariage and the fanily. New York: Plenum.
(1951). A theoretical
personalitv and social behavior. Ameican
theoretical framework for
Sears,R.
R. R.
R. (1951).
or personality
Prychologist,
6,47U83.
Stryker,S- (1974. Dwelopments in "two socialpsychologies":Toward an appreciationof mutual
relevance. Sociometry,40, 145-160.
FOOTNOTE
l.
Based on reviewer reactions, it is probably important to clarifu that we a-renot
discounting the influence of editorial leadership in shaping the nature of the p-c
articles that do get published in the two journals when we refer to the "p-c
researcherspublishing n CD" or IMF. Rather, we view those published researchers
as representatives of the fwo research communities that collectivelv contribute
articles and support the organizations who run the journais in question. These
communities, of course, include both the researcherswho submit articles and their
peers who review them. As stated elsewhere(Endsley & Brodv, 198L;Endsley, et
al., 1988), it is our view that journal editors and their associates are drawn from
each researcfi g6mmrniry precisely because they are judged to be best able to
reflect the research interests and standards of their respective groups at the time
May, 1989
Familv Science Review
203
of their selection. But clearly, the editorial process,like so many other social
processes,
is both a causeand an effectof the type and qualiryof p-c researchthat
uitimatelygetspubiished.
Inl
Famiiy ScienceReview
Vsfrrms 2 of the FamityScienceRaiew will have a SpecialIssue on:
Career & ProfessionalDevelopment in Family Science
Some of the artides will be:
"career Areas for Family ScienGts: Entry Positions"By Robert E. I(eim and Debora-r
J. Cassidv.
"?amily Frnancialptenning" By Gerald Mason
"l:amily Life Extensioa: A National Profile of t-heProfession' By Mick coleman and
Chrystal Barranti
"Careersfor Family scieoce PhD's outside of Academe: Trends and Implications' By
Gary L. Bowen
"The Professionof Marriage and Family Therap/ By Robert p. ggahmann,and Joe
Edgar Glena
"Non-AcademicCarcer Promotionsin Family Sciencc: Advaacementand Potential' By
Carolta Love
A SpecialSectionAlso will be published on:
'Religious Precept and Family Pracrice: The Impact o6 11sl;Eonon Family Life'
COMPLETE TIIE FORM BELOW TO BEGIN YOTJRST]BSCRIPTION
FAMILY SCIENCE REVIEW
Please begin a subscription of the Family Science Rqiew
c l gl gl mi ni r
Ciqn
ap:
Individuals
NCFR and AAMFT members
of the :-re: <
or *hct hc
gredusla s
-
$20.00_
s15.00_
s3o.0o
_
Include a check payableto FamiryscienceReviewand send to:, Fanity science Review,
Departoeat_o! Family Relatioos & Hurnan Development,The ohio state universiry,
Q6lrrm[us.OH 43210.
244
Infor:
Iountal Ci
informatio
each journ
Social Sci
articles ha
and \IcDc
1976: \tac
Rushtou .
relative su
Porter ( 19
journal se
This
l.iterature i
most iulu
Address:
1
\losr
Kail anci I
of these o
publisirine
of articles
professioo
trainine pr
and Lr m a
and \Iarst
for-.
N e m e:
z.
hbt
hmtn
tlrc .\
booh
consi
tlte .i,
Family Science Review
May, 1989
Robct
'
Unises:n*. l
and the rrq
Carmel '*lr
procedurcs
Instirure.&