2. Theoretical debate
• The idea that the developed world is in a process of
secularization still generates debate, but it is commonly
accepted.
• There are many empirical evidences that support it (see
Norris&Inglehart 2004 among many others).
• But how do we measure religiosity to be able to say it is
eroding?
• We argue that the theory of secularization should take into
account a multidimensional concept of religiosity.
• Secularization could be affecting the different dimensions of
religiosity in different ways.
• Conventional / institutional religiosity = eroding
Personal / individual religiosity = ?
3. Theoretical debate
• Different authors seem to point in this direction:
(but systematic measurement efforts are hard to find)
– Inglehart (1990)
• Talks about a renewed emphasis on spiritual values in
postindustrial societies.
• Reflects on the perils of using just indicators of practices to
measure the concept of religiosity.
4. Theoretical debate
– Inglehart & Norris (2004), Sacred and Secular
• In postindustrial societies people are becoming:
– indifferent to traditional religious values,
– but they are not abandoning private or individualized spirituality.
• People are increasingly interested in the meaning and
purpose of life.
– When survival is uncertain need for security in religion
– When survival is guaranteed need for meaning
• At the same time in this countries, there is less support for:
– Traditional religious authorities
– Established religious practices
5. Theoretical debate
– Pollack & Pickel (2007)
• Link between individualization theory and secularization.
• Modernization will produce a change in the forms of religion
more than a decline in its social significance.
• Traditional forms of religiosity will be replaced by more
subjective ones:
– detached from church,
– individually chosen,
– and syncretistic.
• With the indicators they used, they concluded that the rise of
individual religiosity cannot compensate for the loss of
institutionalized religiosity.
6. Data and methodological strategies
• Departing from here, we started with the idea of a cross-country
over time comparison of religiosity indicators (individual level).
• We wanted to question whether the secularization-thesis was really
valid for the different dimensions of religiosity:
• Religion as an institution – Religious impulse, or
• Conventional religiosity – Individualized religiosity
• But soon we realized we were facing a measurement problem:
complicated concepts-by-postulation not clearly defined in the
literature (religiosity, secularization).
• Then we turned our attention to the measure of religiosity.
• In the literature we found basically theoretically-driven measures
and a lack of empirically-driven ones.
• Main problem: finding good measures for our theoretical concepts in
the available surveys.
7. Data and methodological strategies
• We wanted to use an appropriate methodology for our
measurement model Structural Equation Modeling.
• Difficulties to fit our theory and data into a measurement
model (SEM):
– Inappropriate variables
– Incomplete data (waves, gaps)
• But as we believed in the existence of a two dimensional
religiosity model, went on with the analysis.
• And we did not want to proceed as in part of the
literature: only theoretical definitions from which to
construct more or less arbitrary indexes.
• We use the 1st round of the ESS, Spanish subset.
8. Data and methodological strategies
• Importance of religion in life (11 point scale)
• Church attendance (7 point scale)
• Frequency of praying (7 point scale)
• Self-assessed religiosity (11 point scale)
9. Data and methodological strategies
• Importance of religion in life (11 point scale)
– Wording: “How important is religion in your life?”
From 0 extremely unimportant to 10 extremely important.
– It sits on a battery of items representing spheres of life (family,
work, friends, etc.)
– To Norris & Inglehart it is a religious value (ultimate goal).
– Supposedly previous to religious beliefs and participation.
10. Data and methodological strategies
• Church attendance (7 point scale)
– Wording: “How often do you attend religious services apart from
special occasions?”
1 Every day; 2 More than once a week; 3 Once a week; 4 At least once a month;
5 Only on special holy days; 6 less often; 7 never.
– An institutional / communitarian form of religious participation.
– Established by family bonds, habit and social environment, and
not only by personal beliefs
– It does not necessarily reflect the personal views of individuals.
11. Data and methodological strategies
• Frequency of praying (7 point scale)
– Wording: “How often do you pray apart from religious services?”
1 Every day; 2 More than once a week; 3 Once a week; 4 At least once a month;
5 Only on special holy days; 6 less often; 7 never
– A personal / individual form of religious participation.
– It is possible for people to pray even though he does not attend
religious services.
12. Data and methodological strategies
• Self-assessed religiosity (11 point scale)
– Wording: “How religious are you?”
From 0 not at all religious to 10 very religious
– Subjective self-considerations which is not necessarily link to
conventional religiosity (however conventional religious people
would also consider themselves religious).
13. Data and methodological strategies
• We wanted to test:
– if these 4 indicators converge in a single latent
variable of religiosity,
– or to what extent we can talk of more than one
dimension of religiosity?
14. Measurement model for religiosity
(ESS 1rd, Spain)
X1 X2 X3 X4
X1 1.00
X2 0.64 1.00
Ɛ1
Importance of
religion in life X3 0.59 0.61 1.00
y1=ŋ1 X4 0.63 0.67 0.72 1.00
β11
Ɛ2
Frequency of Ɛ5
participation
y2=ŋ2
β21
Religiosity
ŋ5
Ɛ3
Frequency of
praying β31
y3=ŋ3
Ɛ4 β41
Degree of
religiosity
y4=ŋ4
Latent Y model
15. Religiosity 1b
dat ni=4 no=1800 ma=km Matrix of the
disturbance terms:
km No covariance
1.00 between disturbances
.64 1.00
.59 .61 1.00 No. eta Variance/covariance
(endogenous of latent variables
.63 .67 .72 1.00 variables)
labels
y1 y2 y3 y4
model ne=5 ny=4 be=fu,fi ly=fu,fi te=di,fi ps=di,fr
va 1 ly 1 1 ly 2 2 ly 3 3 ly 4 4 Lambdas fixed to one
fr be 1 5 be 2 5 be 3 5 be 4 5
There are lambdas and betas
fi ps 5 5
va 1 ps 5 5
start .5 all
output mr tv mi ss
Fixed variance of
latent Y
16. Complete measurement model for religiosity
(ESS 1rd, Spain)
0.44
δ1
Importance of
religion in life
x1
0.75
δ2 0.38
Frequency of
participation
0.79
x2
Religiosity
0.35 0.81 ξ1
δ3
Frequency of
praying
x3
0.86
δ4 0.25
Degree of
religiosity
x4
Df. = 2
Chi-Square’s p-value = 0.00
17. Measurement model with correlated errors
X1 X2 X3 X4
X1 1.00
X2 0.64 1.00
δ1
Importance of
religion in life X3 0.59 0.61 1.00
δ21 x1 X4 0.63 0.67 0.72 1.00
λ11
δ2
Frequency of
participation
x2
λ21
Religiosity
ξ1
δ3
Frequency of
praying λ31
x3
δ4 λ41
Degree of
religiosity
x4
18. Religiosity 2
dat ni=4 no=1800 ma=km
km
1.00
.64 1.00
.59 .61 1.00
.63 .67 .72 1.00
labels
y1 y2 y3 y4
model ny=4 ne=1 ly=fu,fr te=sy,fi ps=fu,fi
free te 1 1 te 2 2 te 3 3 te 4 4 Variance of
disturbances set
free te 2 1 free
va 1 ps 1 1
out mi Only the covariance
between e1 and e2 is
set free
Set the disturbance
of latent variable to
one
19. Complete measurement with correlated errors
0.49
δ1
Importance of
religion in life
0.10 x1
0.72
δ2 0.43
Frequency of
participation
0.75
x2
Religiosity
0.34 0.81 ξ1
δ3
Frequency of
praying
x3
0.88
δ4 0.22
Degree of
religiosity
x4
Df. = 1
Chi-Square’s p-value = 0.24
20. • Correlated error terms: some of the covariation is due to
sources different to the common factor (and not included in
the model).
• So we tried another model departing from the idea of the two
dimensions of religiosity
21. ζ2
Ɛ4
Subjective λ42 Self-assessed
religiosity religiosity
ŋ2 y4
Ɛ1
Importance of
religion in life psi Ψ21 Ho = 1
Ha ≠ 1
y1
λ11
ζ1
Ɛ2
Frequency of Conventional
participation λ21
religiosity
y2 ŋ1
λ31
Ɛ3
Frequency of
praying
y3
22. • If self-assessed religiosity and our latent construct for
religiosity not happen to be completely related, this would
mean that there is something else that conventional indicators
of religiosity are not covering.
• We had the hypothesis that this “something else” is the
religious impulse or the personal / individual side of religiosity.
23. Estimated via SQP
ζ2
Ɛ4
Subjective λ42 Self-assessed
religiosity religiosity
ŋ2 y4
Ɛ1
Importance of
religion in life psi Ψ21 Ho = 1
Ha ≠ 1
y1
λ11
ζ1
Ɛ2
Frequency of Conventional
participation λ21
Religiosity
y2 ŋ1
λ31
Ɛ3
Frequency of This model was not
praying fitting due to
y3 correlated errors
24. ζ2
Ɛ4
Subjective λ42 Self-assessed
religiosity religiosity
ŋ2 y4
Ɛ1
Importance of
religion in life psi Ψ21 Ho = 1
Ha ≠ 1
y1
λ11
ζ1
Ɛ2
Frequency of Conventional
participation λ21
religiosity
y2 ŋ1
λ31
Ɛ3
Frequency of
praying
y3 The pattern of
correlations among
errors made us think
of an alternative
model specification
25. Theoretical model Ɛ3
Frequency of
praying
ζ2 λ32
y3
Subjective
religiosity
Ɛ4
ŋ2
λ42 Self-assessed
religiosity
y4
Ho = 1
psi Ψ21
Ɛ1 Ha ≠ 1
Importance of
religion in life
λ11 ζ1
y1
Conventional
religiosity
λ21 ŋ1
Ɛ2
Frequency of
participation
y2
26. Religiosity 4b
dat ni=4 no=1800 ma=km
km
1.00
.64 1.00
.59 .61 1.00
.63 .67 .72 1.00
labels
y1 y2 y3 y4
model ny=4 ne=2 ly=fu,fi te=sy,fi ps=sy,fi
free ly 1 1 ly 2 1 ly 3 2 ly 4 2
fre te 1 1 te 2 2 te 3 3 te 4 4
va 1 ps 1 1 ps 2 2
free ps 2 1
out rs mi
Set free the
covariance between
latent constructs
27. Results (Spain, ESS 1st round) Ɛ3
Frequency of
praying 0.34
ζ2 λ32 =0.81
y3
Subjective
religiosity
Ɛ4
ŋ2
Self-assessed
λ42=0.88 religiosity 0.32
y4
(psi) Ψ21= 0.92
0.39
Ɛ1
Importance of
religion in life
λ 11=0.78 ζ1
y1
Conventional
religiosity
0.33 ŋ1
Ɛ2
Frequency of λ21=0.82
participation
y2
Df. = 1
Chi-Square’s p-value = 0.24
28. Conclusions
• From the Spanish data, two dimensions of religiosity
emerge:
– One related to traditional religious values and practices
– The other is linked to the subjective and personal sphere
• Although these two dimensions are highly correlated,
they are not the same, as shown by our measurement
model.
• A conventional religious person would score high on the
subjective religiosity dimension, but it is also possible for
someone to have a personal sense of religiosity and be
relatively detached from conventional religiosity.
29. Conclusions
• The idea of two dimensions of religiosity is similar to
Pollack and Pickel (2007) thesis of individualization.
• However, as shown by their empirical analysis, the rise
of individually religiosity cannot compensate for the
losses of institutionalized religiosity.
• Modernization produces:
– Secularization (big part)
– Individualization of religion (small part)
• We still have to test it.