Upload
rebecca-parrish
View
212
Download
0
Tags:
Embed Size (px)
Citation preview
A New Look at the Evaluation of Sociological Theories in International Large Scale Educational Assessments
Daniel Caro and Andrés Sandoval-Hernandez CIES 2012 Annual ConferenceApril 25, San Juan, Puerto Rico
Rationale
Two important criticisms of LSA studies areStudies lack a solid theoretical basis
This affects the selection, operationalization and explanation of the relations between variablese.g., extended use of SES indicators
Techniques for developing and testing theories (EFA and CFA) have several limitations, for example:
EFA lacks model fit indicesCFA is too restrictive
This study addresses these criticisms
Purpose
Evaluate sociological theories of economic, cultural, and social capital postulated by Bourdieu, Bernstein, and Coleman
Investigate if theoretical models are reflected by the data of international student assessments
Demonstrate advantages of ESEMExplore factor structureEvaluate model fitTest MI across countries
Theoretical Framework
Economic capital (Bourdieu, 1983)
Economic capital is defined as the command individuals have over economic resourcesThe concept is commonly understood as exchange values, like income and assets that can be easily transformed into cash
Cultural capital (Bernstein, 1971; Bourdieu, 1983)
Cultural capital is accounted by the cultural long-lasting dispositions embedded in the human mind and body, as well as in cultural goods and educational credentials Cultural capital can appear in three states: objectified, institutionalized, and embodied
Theoretical Framework
Social capital (Coleman, 1988)
Social capital are all those aspects of the social structure that can be used by actors as resources to achieve their interestsFamily social capital is represented by family relations that enhance the transmission of other structural resources (e.g. dispositions towards culture)
EFA and CFA
In general, EFA for generating theoryCFA for testing theory
CFA limitations (Marsh, 2009, 2010):The zero cross-loading restriction
Statistically, leads to inflated interfactor correlationsAlso theoretically problematic , e.g. number of books reflects both cultural capital and economic capital
Often well-defined EFA structures are not supported by CFA
In practice researchers perform exploratory CFA
Exploratory Structural Equation Modeling (ESEM)
Marsh et al. (2009, 2010) introduced ESEM, an integration of EFA and CFAEssentially, ESEM is EFA with
Standard errorsModel fit indicesMI tests
ESEM is appropriate if model fit is better than in CFATo our knowledge, ESEM has not been yet applied to sociological theories
Data
PIRLS 200642 educational systems 21,000 students (500*42)36 items from parent questionnaire
PISA 200914 countries 7,000 students (500*14)66 items from student and (optional) parent questionnaire
Analytical Strategy
EFA with all itemsSelected items (indexes) for 3 factors
14 in PIRLS 200612 in PISA 2009
CFA and ESEM with selected itemsModel fitInterfactor correlationsCross-loadings
Measurement invariance (MI) tests by country’s HDI
Child
ren'
s bo
ok
Pare
nts'
read
ing
for
enjo
ymen
t
Pare
ntal
atti
tude
s to
read
ing
Pare
nts'
read
ing
at
hom
e
Book
s at
hom
e
Visi
ts to
libr
ary
/ bo
okss
tore
Aca
dem
ic
com
mun
icati
on
Com
mun
icati
on &
In
tera
ction
Cultu
ral
com
mun
icati
on
Lite
racy
sup
port
ac
tiviti
es
Pare
ntal
oc
upati
onal
sta
tus
Child
rens
’ ow
n ro
om
Fina
ncia
l sta
tus
Pare
ntal
edu
catio
n
Cultural Capital
Social Capital
Economic Capital
PIRLS 2006: CFA results
0.73 0.19
0.30
CFI= 0.77TLI= 0.71RMSEA= 0.07
Child
ren'
s bo
ok
Pare
nts'
read
ing
for
enjo
ymen
t
Pare
ntal
atti
tude
s to
read
ing
Pare
nts'
read
ing
at
hom
e
Book
s at
hom
e
Visi
ts to
libr
ary
/
book
shop
Aca
dem
ic
com
mun
icati
on
Com
mun
icati
on &
In
tera
ction
Cultu
ral
com
mun
icati
on
Lite
racy
sup
port
ac
tiviti
es
Pare
ntal
oc
upati
onal
sta
tus
Child
rens
’ ow
n ro
om
Fina
ncia
l sta
tus
Pare
ntal
edu
catio
n
Cultural Capital
Social Capital
Economic Capital
PIRLS 2006: ESEM results
0.73 0.41
0.190.08
0.300.24
CFI= 0.77 0.96TLI= 0.71 0.93RMSEA= 0.07 0.03
PIRLS 2006: CFA and ESEM factor loadings
F1 F2 F3 F1 F2 F3Economical Capital (F1)Parental education 0.64 0.09 0.02 0.80 0.00 0.00Parental occupational status 0.62 0.06 -0.04 0.74 0.00 0.00Financial status 0.29 0.02 0.08 0.35 0.00 0.00Children's own room 0.24 0.01 0.03 0.24 0.00 0.00
Cultural Capital (F2)Books at home 0.65 0.27 -0.02 0.00 0.83 0.00Children's books 0.70 0.16 -0.01 0.00 0.76 0.00Parental attitudes toward reading 0.07 0.62 0.04 0.00 0.53 0.00Frequency parents read for enjoyment -0.08 0.78 0.07 0.00 0.49 0.00Time parents spend reading at home 0.17 0.56 -0.02 0.00 0.56 0.00
Social Capital (F3)Parents literacy support activities 0.27 0.20 0.47 0.00 0.00 0.61Parent-child cultural communication -0.05 0.03 0.85 0.00 0.00 0.80Parent-child communication and interaction 0.22 0.10 0.47 0.00 0.00 0.56Parent-child academic communication -0.14 0.01 0.71 0.00 0.00 0.66Parent-child visit library/bookstore 0.11 0.16 0.27 0.00 0.00 0.36
ESEM ICM-CFA
Hom
e ed
ucati
onal
re
sour
ces
Cultu
ral
poss
essi
ons
Pare
nts'
read
ing
m
otiva
tion
Book
s at
hom
e
Visi
ts to
libr
ary
/ bo
okst
ore
Aca
dem
ic
com
mun
icati
on
Com
mun
icati
on &
In
tera
ction
Cultu
ral
com
mun
icati
on
Lite
racy
sup
port
ac
tiviti
es
Pare
ntal
oc
upati
onal
sta
tus
Hou
seho
ld
over
crow
ding
Pare
ntal
edu
catio
n
Cultural Capital
Social Capital
Economic Capital
PISA 2009: CFA results
0.68 0.28
0.36 CFI= 0.87TLI= 0.83RMSEA= 0.07
Hom
e ed
ucati
onal
re
sour
ces
Cultu
ral
poss
essi
ons
Pare
nts'
read
ing
m
otiva
tion
Book
s at
hom
e
Visi
ts to
libr
ary
/ bo
okst
ore
Aca
dem
ic
com
mun
icati
on
Com
mun
icati
on &
In
tera
ction
Cultu
ral
com
mun
icati
on
Lite
racy
sup
port
ac
tiviti
es
Pare
ntal
oc
upati
onal
sta
tus
Hou
seho
ld
over
crow
ding
Pare
ntal
edu
catio
n
Cultural Capital
Social Capital
Economic Capital
PISA 2009: ESEM results
0.680.38
0.280.17
0.360.15
CFI= 0.87 0.96TLI= 0.83 0.93RMSEA= 0.07 0.05
PISA 2009: CFA and ESEM factor loadings
F1 F2 F3 F1 F2 F3Economical Capital (F1)Parental education 0.80 0.02 0.00 0.74 0.00 0.00Parental occupational status 0.59 0.19 -0.04 0.72 0.00 0.00Household overcrowding -0.33 -0.05 -0.05 -0.36 0.00 0.00
Cultural Capital (F2)Parents reading motivation 0.21 0.20 0.33 0.00 0.44 0.00Cultural Possessions -0.01 0.70 -0.07 0.00 0.57 0.00Books at home 0.23 0.58 0.00 0.00 0.74 0.00Home educational resources 0.09 0.48 0.05 0.00 0.51 0.00
Social Capital (F3)Parents literacy support activities 0.22 0.04 0.54 0.00 0.00 0.61Parent-child cultural communication 0.03 0.14 0.59 0.00 0.00 0.64Parent-child academic communication 0.02 -0.08 0.77 0.00 0.00 0.70Parent-child communication and interaction -0.03 0.06 0.52 0.00 0.00 0.53Parent-child visit library/bookstore -0.04 0.16 0.41 0.00 0.00 0.43
ESEM ICM-CFA
Measurement Invariance Tests
Two groups of countries
PIRLS 2006: low/medium HDI versus high HDIPISA 2009: high HDI versus very high HDI
CFI TLI RMSEA CFI TLI RMSEA
Configural Invariance 0.97 0.95 0.04 0.96 0.92 0.05
Weak Invariance 0.97 0.96 0.04 0.88 0.85 0.07
Strong Invariance 0.97 0.97 0.03 0.89 0.86 0.07
Strict Invariance 0.97 0.96 0.03 0.82 0.81 0.08
PIRLS 2006 PISA 2009Group Invariance
Discussion
Evaluated theories of economic, cultural, and social capital are reasonably reflected by the dataESEM performs better than CFA to capture these theories
Better model fitGreater discriminant validity (lower interfactor correlations)Cross-loadings are statistically and theoretically significant
Educational/cultural possessions and parental education for cultural and economic capitalParent-child cultural interactions for social and cultural capital
Discussion
But theoretical models seem not to hold between countries
SES measures (economic capital) are not comparable
This raises theoretical and methodological questionsTheoretical
Are other theories more appropriate?Global or contextualized theories?
MethodologicalWhat should be the groups of comparison?Can ESEM manage comparisons among a large number of groups/countries?
Future Steps
TheoreticalConsider competing theories, e..g., rational action theory
Reproduction theories for developing countries and rational action theory for developed countries?
MethodologicalMI
Identify problematic variables for MICountrywise comparison matrixEvaluate MI within regions
What to do if no MI?