ArticlePDF Available

Does Religion Foster Generosity?

April 2014
The Social Science Journal 51(4)

DOI:10.1016/j.soscij.2014.03.012

Authors:

This paper reviews recent studies that claim to provide support, through statistical analysis of survey data, for the traditional proposition that being religious makes people more generous. The studies have serious shortcomings. First, the data consist exclusively of self-reports. Second, the dependent and independent variables are conceptually problematic and ill-defined. Third, even if there is a positive correlation between religious involvement and personal generosity, it may be due to selection bias. Thus, these studies do not provide serious evidence for the traditional hypothesis. Moreover, it has been directly controverted by experimental studies of economic and other behaviors.

Content uploaded by Roy Sablosky

Content may be subject to copyright.

The

Social

Science

Journal

51

(2014)

545–555

Contents

lists

available

at

ScienceDirect

The

Social

Science

Journal

journa

l

h

om

epa

ge:

www.elsevier.com/locate/soscij

Does

religion

foster

generosity?

Roy

Sablosky∗

1119

37th

Street,

Sacramento,

CA

95816,

USA

a

r

t

i

c

l

e

i

n

f

o

Article

history:

Received

3

July

2013

Received

in

revised

form

20

March

2014

Accepted

20

March

2014

Available

online

26

April

2014

Keywords:

Generosity

Methodology

Religion

Religious

social

networks

Selection

bias

Social

capital

Social

desirability

Surveys

a

b

s

t

r

a

c

t

This

paper

reviews

recent

studies

that

claim

to

provide

support,

through

statistical

analy-

sis

of

survey

data,

for

the

traditional

proposition

that

being

religious

makes

people

more

generous.

The

studies

have

serious

shortcomings.

First,

the

data

consist

exclusively

of

self-

reports.

Second,

the

dependent

and

independent

variables

are

conceptually

problematic

and

ill-deﬁned.

Third,

even

if

there

is

a

positive

correlation

between

religious

involvement

and

personal

generosity,

it

may

be

due

to

selection

bias.

Thus,

these

studies

do

not

provide

serious

evidence

for

the

traditional

hypothesis.

Moreover,

it

has

been

directly

controverted

by

experimental

studies

of

economic

and

other

behaviors.

©

2014

Western

Social

Science

Association.

Published

by

Elsevier

Inc.

All

rights

reserved.

1.

Introduction

In

the

last

decade

or

so,

several

studies

have

claimed

to

offer

strong

empirical

support

for

the

hypothesis

that

religion

has

beneﬁcial

consequences

for

individuals

and

for

society

as

a

whole.

According

to

these

authors,

care-

ful

survey

data

analysis

reveals

that

religious

participation

generates

social

capital

(Putnam,

2000);

that

it

fosters

civic

responsibility

(Monsma,

2007;

Smidt,

den

Dulk,

Penning,

Monsma,

&

Koopman,

2008);

that

it

boosts

volunteering

(Campbell

&

Yonish,

2003)

and

charitable

giving

(Bekkers

&

Schuyt,

2008;

Brooks,

2004;

Bryant,

Jeon-Slaughter,

Kang,

&

Tax,

2003;

Havens,

O’Herlihy,

&

Schervish,

2002;

Hodgkinson,

Weitzman,

&

Kirsch,

1990;

Independent

Sector,

2002;

Nemeth

&

Luidens,

2003;

Regnerus,

Smith,

&

Sikkink,

1998;

Wang

&

Graddy,

2008);

that

it

makes

peo-

ple

more

generous

(Brooks,

2003,

2005,

2006;

Putnam

&

∗Tel.:

+1

301

919

0229.

E-mail

address:

roy.sablosky@sablosky.com

Campbell,

2010;

Weipking

&

Maas,

2009);

and

that

it

pro-

motes

empathy

and

altruism

(Smith,

2006).

Of

course,

this

hypothesis

is

not

new;

it

is

traditional,

so

all

the

authors

cited

above

are

identiﬁed

as

tradition-

alists

in

this

study.

What

is

new

is

their

claim

that

the

traditional

hypothesis

is

strongly

supported

by

quantita-

tive

evidence.

The

data

are

from

widely

respected

sources

such

as

the

General

Social

Survey

(National

Data

Program

for

the

Sciences,

1972–2008),

the

Social

Capital

Community

Benchmark

Survey

(Saguaro

Seminar,

2001),

and

surveys

from

the

Pew

Forum

on

Religion

and

Public

Life

(see

for

example

Pew

Research

Center,

2010).1

1Putnam

and

Campbell

(2010)

also

draw

on

the

2006

and

2007

Faith

Matters

surveys

(International

Communications

Research),

which

were

commissioned

speciﬁcally

for

them

and

funded

by

the

John

Templeton

Foundation

(Putnam

&

Campbell,

2010,

p.

557).

It

should

be

noted

that

the

works

mentioned

in

this

paragraph

are

based

almost

entirely

on

data

taken

in

the

United

States.

The

exceptions

are

the

studies

of

Bekkers

and

Schuyt

(2008)

and

Weipking

and

Maas

(2009),

which

were

conducted

in

the

Netherlands.

http://dx.doi.org/10.1016/j.soscij.2014.03.012

0362-3319/©

2014

Western

Social

Science

Association.

Published

by

Elsevier

Inc.

All

rights

reserved.

546

R.

Sablosky

/

The

Social

Science

Journal

51

(2014)

545–555

Two

books

stand

out

from

this

body

of

work

for

the

scope

and

forcefulness

of

their

claims:

Who

Really

Cares?

by

Arthur

Brooks

(2006)

and

American

Grace

by

Robert

Putnam

and

David

Campbell

(2010).

“When

it

comes

to

charity,”

writes

Brooks

(2006,

p.

2),

“America

is

two

nations—one

charitable,

the

other

uncharitable.”

Compared

to

the

non-

religious,

he

says,

“religious

people

are,

inarguably,

more

charitable

in

every

measurable

way”

(p.

40,

emphasis

in

the

original).

Putnam

and

Campbell

(2010,

pp.

453–454)

vigorously

agree:

Some

Americans

are

more

generous

than

others.

.

.

In

particular,

religiously

observant

Americans

are

more

generous

with

time

and

treasure

than

demographically

similar

secular

Americans.

.

.

The

pattern

is

so

robust

that

evidence

of

it

can

be

found

in

virtually

every

major

national

survey

of

American

religious

and

social

behav-

ior.

Any

way

you

slice

it,

religious

people

are

simply

more

generous.2

This

paper

argues

that

these

ﬁndings

are

invalidated

by

an

array

of

methodological

faults.

First,

the

data

consist

exclusively

of

self-reports,

which

are

known

to

be

unre-

liable.

Second,

both

the

dependent

and

the

independent

variables

are

conceptually

problematic.

The

word

“religios-

ity”

can

refer

to

many

different

phenomena;

there

are

no

objective

criteria

with

which

to

identify

the

most

signiﬁ-

cant

ones

or

to

judge

whether

the

presence

of

any

of

them

qualiﬁes

the

bearer

as

truly

religious.

Nor

can

generos-

ity

be

measured

directly,

at

least

if

it

is

understood

in

its

traditional

sense,

as

a

virtue.

In

the

studies

in

question,

generosity

is

operationalized

as

relatively

higher

levels

of

tax-deductible

charitable

contributions.

This

is

misleading

in

two

ways.

First,

tax-deductible

giving

is

not

the

same

thing

as

being

generous.

Second,

the

tax-deductible

dona-

tions

reliably

associated

with

religious

participation

mostly

go

to

churches;

and

most

church

spending

goes

to

other

than

humanitarian

purposes.

Furthermore,

even

if

there

is

a

positive

correlation

between

religious

involvement

and

generosity,

it

could

be

due

to

selection

bias—that

is,

the

direction

of

causal-

ity

may

be

the

opposite

of

what

is

typically

assumed.

Putnam

and

Campbell

identify

what

they

call

“religious

social

networks”

as

a

cause

of

generosity;

this

conclusion

results

from

assuming,

rather

than

demonstrating,

that

the

network

effects

they

identify

in

their

data

are

due

to

specif-

ically

religious

factors.

Finally,

I

argue

that

the

traditional

hypothesis

has

been

directly

controverted

by

experimental

studies

of

economic

and

other

behaviors.

2.

Methodological

challenges

in

studies

of

survey

data

Religion

and

its

effects

are

difﬁcult

to

study.

Reli-

gion

is

surrounded

by

misconceptions,

many

of

which

are

2Putnam

and

Campbell

(2010)

do

not,

of

course

agree,

with

everything

in

Brooks

(2006).

In

particular,

they

differ

strongly

with

Brooks’s

asser-

tion

that

political

conservatism

is

associated

with

generosity,

calling

it

“an

elementary

statistical

mistake”

(pp.

457–458;

see

also

note

28,

pp.

632–633).

entertained

even

by

experienced

sociologists.

Mark

Chaves

(2010,

p.

6)

writes

that

sociology

of

religion

is

afﬂicted

by

the

“religious

congruence

fallacy,”

whose

“telltale

sign

.

.

.

is

a

regression

model

in

which

the

coefﬁcients

attached

to

religious

service

attendance,

religious

belief,

or

religious

afﬁliation

are

interpreted

causally.”

In

other

words,

“almost

every

claim

of

the

form,

‘People

act

in

a

certain

way

because

they

are

in

a

particular

religion

or

because

they

attend

reli-

gious

services

or

because

they

hold

this

or

that

religious

belief”’

is

based

on

a

misunderstanding.

2.1.

Self-reporting

and

social

desirability

bias

The

traditionalists’

data

come

entirely

from

surveys.

It

is

well

known

(Cahalan,

1969;

Parry

&

Crossley,

1950;

Phillips

&

Clancy,

1972)

that

survey

responses

are

subject

to

social

desirability

(SD)

bias,

the

tendency

of

respondents

to

adjust

their

responses

toward

conformity

with

social

norms.

As

Goffman

(1959,

p.

35)

points

out,

every

response

is

given

in

the

context

of

a

social

performance.

“When

the

individ-

ual

presents

himself

before

others,

his

performances

will

tend

to

incorporate

and

exemplify

the

ofﬁcially

accredited

values

of

the

society,

more

so,

in

fact,

than

does

his

behavior

as

a

whole.”

People

tend

to

over-report

socially

encouraged

behaviors,

such

as

philanthropy,

and

under-report

socially

deprecated

behaviors,

such

as

drug

abuse.

Holtzman

and

Kagan

(1995,

ch.

1,

p.

5)

ﬁnd

that

“there

is

often

a

mini-

mal

correlation,

or

none

at

all”

between

self-reports

and

related

external

measures

(such

as

of

behavior).

Doris

(2002,

p.

179,

n.

42)

notes

that

“failures

of

behavior

to

conform

with

avowed

values

and

self-conceptions

are

well

documented

in

psychology”

and

cites

several

examples.

Studies

of

generosity

are

especially

vulnerable

to

SD

bias.

The

classic

papers

on

SD

bias

examined

responses

on

topics

such

as

voting,

sexuality,

and

drug

use.

But

generosity

is

not

merely

subject

to

social

desirability,

as

those

topics

are;

it

is

the

epitome.

Nothing

evokes

social

approbation

more

consistently

than

generosity—with

the

possible

exception,

in

some

communities,

of

church

atten-

dance.

This

means

that

we

should

be

maximally

skeptical

regarding

people’s

reports

of

their

own

generosity.

This

applies

to

volunteering

as

much

as

it

does

to

monetary

donations.

SD

bias

can

skew

results

in

another,

less

obvious

way.

If

groups

of

respondents

differ

in

the

degree

or

direction

of

their

SD

bias,

this

can

create

spurious

correlations

or

mask

real

correlations

between

variables

(Bell

&

Buchanan,

1966;

Ganster,

Hennessey,

&

Luthans,

1983;

Presser

&

Traugott,

1992;

Stocké

&

Hunkler,

2007).

For

example,

in

a

study

of

the

relation

between

age

and

various

psychological

meas-

ures

such

as

agreeableness

and

conscientiousness,

Soubelet

and

Salthouse

(2011,

p.

758)

ﬁnd

that

social

desirability

accounts

for

50%

of

the

age-related

variance

in

eight

self-

report

variables.

When

we

come

to

religion,

measurement

problems

pro-

liferate.

Self-report

measures

.

.

.

fail

to

make

a

distinction

among

(1)

what

people

say

they

believe,

value,

and

do;

(2)

what

they

honestly

believe

they

believe,

value,

and

do;

and

(3)

what

they

actually

believe,

value,

and

do..

.

.

R.

Sablosky

/

The

Social

Science

Journal

51

(2014)

545–555

547

Problems

may

be

especially

severe

when

correlating

these

self-reports

with

religion.

This

is

because

all

major

religions

make

rather

clear

prescriptions

about

the

right

answers

to

such

questions.

(Batson,

Schoenrade,

&

Ventis,

1993,

pp.

382–383,

emphasis

in

the

original)

These

observations

call

into

question

all

responses

to

any

survey

question

of

the

form

“How

important

is

X

to

you?”

where

X

is

God,

prayer,

church

attendance,

or

any-

thing

else

people

approve

or

disapprove

in

each

other.

2.2.

Measuring

generosity

In

standard

usage,

generosity

is

a

virtue—that

is,

a

dispo-

sition

rather

than

a

behavior.

The

Oxford

English

Dictionary

(2009)

deﬁnes

a

generous

person

as

one

who

“shows

a

readiness

to

give

more

of

something,

esp.

money,

than

is

strictly

necessary

or

expected.”

The

Science

of

Generosity

Initiative

at

Notre

Dame

University

(n.d.)

deﬁnes

generos-

ity

as

“the

virtue

of

giving

good

things

to

others

freely

and

abundantly.”

The

initiative’s

Collett

and

Morrissey

(2007)

distinguish

religiosity

from

“mere

prosocial

behav-

ior”

by

its

“connotations

of

noble

and

magnanimous

motivations.”

Such

a

property

would

be

very

difﬁcult

to

operationalize.

In

practice,

of

course,

sociologists

must

focus

on

behavior

rather

than

internal

dispositions.

The

tradi-

tionalist

studies

use

charitable

spending

as

a

proxy

for

generosity—or,

rather,

they

use

self-reports

of

charitable

spending.

This

puts

us

several

steps

away

from

the

mea-

surement

of

anyone’s

virtue,

and

the

link

will

get

even

more

tenuous

as

we

examine

the

methodology

in

more

detail.

In

1950,

Parry

and

Crossley

compared

answers

to

the

question

“Did

you,

yourself,

give

to

the

Denver

Community

Chest

drive

this

fall?”

with

the

records

of

the

Community

Chest

organization.

Forty

percent

of

the

self-reports

were

false.

The

authors’

conclusion

(p.

76)

is

bleak:

“It

can

safely

be

said

that

this

sort

of

question,

whether

it

concerns

the

Community

Chest

or

some

other

charitable

organization,

is

not

very

helpful

for

survey

use.”

More

recently,

Wilhelm

(2007)

reports

wide

variations

in

reported

charitable

dona-

tions

between

different

large-scale

surveys

such

as

Giving

and

Volunteering

and

the

General

Social

Survey.

Putnam

and

Campbell

(2010,

p.

444)

note

casually

that

“generosity

can

be

measured

most

simply

by

measur-

ing

gifts

of

time

and

money”;

yet

the

surveys

on

which

their

work

is

based

do

not

measure

such

gifts:

they

mea-

sure

anonymous,

unveriﬁed

self-reports

of

such

gifts.

None

of

the

traditionalist

studies

directly

measures

charitable

donations,

to

say

nothing

of

charitable

ideas

or

noble

inten-

tions.

Using

data

from

the

2004

and

2006

GSS,

Putnam

and

Campbell

ﬁnd

(p.

451)

that

“frequent

churchgoers”

are

more

likely

than

other

people

to

“give

money

to

charity,”

“donate

blood,”

“allow

a

stranger

to

cut

in

front

of

them,”

and

seven

other

friendly

things.3But

none

of

the

acts

of

3Do

volunteer

work

for

a

charity;

give

money

to

a

homeless

person;

give

excess

change

back

to

a

shop

clerk;

help

someone

outside

their

own

household

with

housework;

spend

time

with

someone

who

is

a

bit

down;

offer

a

seat

to

a

stranger;

and

help

someone

ﬁnd

a

job.

generosity

reported

to

the

GSS

interviewers

were

corrob-

orated

by

other

means.

Strictly

speaking,

the

correlations

that

Putnam

and

Campbell

ﬁnd

here

imply

only

that

peo-

ple

who

report

frequent

church

attendance

are

also

more

likely

to

report

frequent

acts

of

generosity.

If

people

deemed

“more

religious”

and

“less

religious”

by

the

traditionalists

were

equally

likely

to

exaggerate

their

gift-giving,

the

methodological

problem

created

by

these

distortions

would

perhaps

be

manageable.

But

evidence

that

SD

bias

increases

with

religiosity

is

reported

by

Batson,

Naifeh,

and

Pate

(1978),

Francis,

Pearson,

and

Kay

(1983),

Powell

and

Stewart

(1978),

and

Trimble

(1997).

Another

conceptual

problem

complicates

the

mea-

surement

of

generosity.

Throughout

the

traditionalist

literature,

the

words

“generosity,”

“charity,”

and

“giving”

are

used

as

synonyms,

but

charity

has

an

additional,

spe-

cialized

sense.

In

the

US

tax

code

and

in

colloquial

English,

a

charity

is

a

nonproﬁt

corporation;

donations

to

such

orga-

nizations

are

also

called

charity.

The

problem

is

that

the

concept

of

generosity

is

distinct

from

that

of

tax-deductible

contributions.

The

ﬁrst

means

giving

more

than

expected,

out

of

the

goodness

of

one’s

heart;

the

second

means

giv-

ing

money

to

an

organization

that

qualiﬁes

as

not-for-proﬁt

under

the

tax

code.

It

is

possible

to

make

charitable

dona-

tions

without

being

generous

(by

giving

much

less

than

one

could);

and

it

is

possible

to

be

generous

without

making

charitable

donations

(by

giving

people

cash,

goods,

assis-

tance,

or

encouragement—especially

if

one

gives

more

than

was

expected).

The

two

are

categorically

different,

yet

both

are

routinely

called

charity

in

the

traditionalist

literature.

Brooks

(2006,

p.

3)

writes,

for

example:

“It

is

in

all

our

interests

to

ﬁgure

out

what

makes

people

charitable,

and

what

makes

them

uncharitable.”

Is

he

speaking

of

gen-

erosity

here,

or

of

tax-deductible

donations?

It

sounds

like

the

former,

but

the

surveys

he

cites

measured

only

the

latter—or

rather,

self-reports

of

the

latter.

Another

problem

is

that

the

traditionalist

studies

cited

here

lump

churches

in

with

all

other

tax-exempt

organiza-

tions.

But

churches

are

a

special

case.

Churches

are

classed

as

not-for-proﬁt

for

a

traditional

reason:

because

religion

has

traditionally

been

assumed

to

be

beneﬁcial

to

the

com-

munity

at

large.

But

to

accept

this

is

to

beg

the

question.

One

does

not

count

one’s

membership

dues

for

the

local

health

club

as

a

charitable

contribution.

This

is

because

they

are

not

a

gift

to

the

community

at

large;

they

beneﬁt

only

the

members

and

proprietors

of

the

club.

But

this

is

true

of

churches

as

well,

at

least

to

a

ﬁrst

approximation.

Chaves

(2012,

ch.

9,

p.

382)

reports:

Most

congregations

focus

primarily

on

their

religious

activities..

.

.

The

median

dollar

amount

spent

by

con-

gregations

directly

in

support

of

social

service

programs

in

2006–07

was

about

$1,300

.

.

.

[or]

about

2

percent

of

the

average

congregation’s

total

annual

budget..

.

.

The

vast

majority

of

congregations’

resources

support

wor-

ship

services,

religious

education,

and

pastoral

care

for

their

own

members.

Donating

to

one’s

church

is

a

different

kind

of

ges-

ture

from

donating

to

a

humanitarian

organization,

such

as

the

Red

Cross.

It

is

a

mistake

to

add

these

two

kinds

of

activities

and

treat

the

sum

as

a

single

variable

called

548

R.

Sablosky

/

The

Social

Science

Journal

51

(2014)

545–555

“charity”—especially

given

that

the

term

sounds

like

a

synonym

of

“generosity.”

This

confusion

of

terms

is

foun-

dational

to

the

traditionalist

literature.

2.3.

Measuring

religiosity

We

have

seen

that

properly

conceptualizing

and

measuring

generosity

is

very

difﬁcult,

and

that

the

tradi-

tionalists

have

not

succeeded.

Religiosity

is

even

harder.

In

studies

whose

methodology

presupposes

the

quanti-

tative

measurement

of

subjects’

religiosity,

four

different

types

of

assessment

are

found.

The

most

common

is

to

use

self-reported

frequency

of

church

attendance.

Some

studies

use

a

single

measure

constructed

from

weighted

multiple

self-reports

such

as

frequency

of

church

atten-

dance,

frequency

of

prayer,

personal

importance

of

religion,

and

so

on—such

as

Putnam

and

Campbell’s

(2010)

“index

of

religiosity.”

Many

use

a

single,

self-reported

psycho-

logical

measure.

Ahmed

and

Salas

(2009),

for

example,

simply

asked

participants

whether

they

were

religious.

Finally,

some

studies

wrestle

with

multiple

values,

derived

from

a

questionnaire

designed

to

elicit

different

types

of

religiosity—for

example,

the

Means,

Ends,

and

Quest

dimensions

introduced

by

Batson

(1976)

or

the

Dimin-

ished,

Privatized,

Public,

and

Integrated

modes

of

Smidt

et

al.

(2008).

As

a

practical

matter,

these

can

be

regarded

as

just

two,

not

four,

different

approaches

to

the

measurement

of

reli-

giosity.

One

method

asks

people

for

introspective

reports

on

their

thoughts,

feelings,

and

beliefs;

the

other

asks

peo-

ple

how

often

they

go

to

church.

Let

us

consider

each

of

these

approaches

more

carefully.

2.3.1.

Thoughts,

feelings,

and

beliefs

Religion’s

physical

and

behavioral

trappings—symbols,

words,

music,

costumes,

buildings,

books,

and

ritualistic

actions—are

designed

to

be

easy

to

identify.

It

is

assumed

that

underlying

these

public

displays

there

is

a

private

experience;

and

this

inner

experience,

not

the

public

dis-

plays,

is

what

religion

is

supposed

to

be

about.

We

call

this

inner

experience

religious

belief.

It

is

regarded

as

intensely

private,

and

at

the

same

time

as

being

shared

and

under-

stood

by

the

vast

majority

of

Americans.

However,

what

people

refer

to

as

religious

belief

is

a

very

idiosyncratic

matter.

The

set

of

possible

meanings

of

the

statement

“I

believe

in

God”

is

unbounded.

Although

each

denomina-

tion

teaches

a

more

or

less

ﬁxed

set

of

ideas

and

practices,

the

way

those

teachings

are

incorporated

into

each

per-

son’s

understanding

of

how

the

world

works

is

extremely

individualistic—and

opaque

to

investigation.

Researchers

assume

that

among

all

these

varieties

of

belief

there

is

a

common

factor

that

can

be

identiﬁed

and

quantiﬁed.

It

is

almost

a

tautology

that

the

validity

of

introspec-

tive

measures

cannot

be

checked.

This

problem

is

of

course

not

limited

to

religion;

it

affects

any

setting

where

we

wish

to

collect

data

on

people’s

private

thoughts

or

feel-

ings.

Survey

questions

about

personal

beliefs

assume

that

respondents

can,

and

will,

introspectively

ascertain

what

their

own

beliefs

are.

But

gaining

reliable

access

to

sub-

jects’

sincere

beliefs

is

difﬁcult,

to

say

the

least.

The

subjects

themselves

do

not

necessarily

have

such

access

(Bargh

&

Chartrand,

1999;

Nisbett

&

Wilson,

1977).4Furthermore,

if

the

beliefs

in

question

are

religious

ones,

the

task

will

be

even

harder.

As

we

saw

earlier,

the

public

expression

of

religious

belief

is

subject

to

strong

social

norms.

In

view

of

all

these

considerations,

it

is

safe

to

predict

that

surveys

will

tell

us

more

about

what

respondents

think

they

are

expected

to

say

than

about

what

they

really

believe.

2.3.2.

Church

service

attendance

Because

it

is

impossible

to

measure

belief

itself,

some

scholars

adopt

church

attendance

as

a

proxy

for

religiosity.

But

this

method

has

its

own

problems.

Year

after

year,

in

all

kinds

of

polls

and

surveys,

about

40%

of

Americans

report

that

they

attend

church

services

every

week,

or

near

enough.

However,

empirical

studies

suggest

that

about

half

of

these

people

are

not

telling

the

truth.

The

correct

ﬁgure

for

weekly

church

attendance

is

closer

to

20%.

This

has

been

ascertained

by

direct,

phys-

ical

methods

such

as

going

to

every

church

in

a

selected

county

and

counting

the

people

in

the

pews

(Brenner,

2011;

Chaves

&

Cavendish,

1994;

Chaves

&

Stevens,

2003;

Hadaway

&

Marler,

2005;

Hadaway,

Marler,

&

Chaves,

1993,

1998;

Marcum,

1999;

Marler

&

Hadaway,

1999;

Presser

&

Stinson,

1998).

Presumably,

this

over-reporting

is

a

result

of

social

desirability

bias.

Hadaway

et

al.

(1998,

p.

127)

offer

a

strik-

ing

anecdote

in

support

of

this

hypothesis:

At

the

1994

Society

for

the

Scientiﬁc

Study

of

Religion

session

devoted

to

our

[1993]

article

.

.

.

one

of

the

com-

mentators,

a

retired

ofﬁcial

of

the

National

Council

of

Churches

of

Christ,

acknowledged

that

she

would

say

she

attended

services

last

week

even

if

it

was

not

true.

She

went

on

to

say

that

she

would

not

consider

her

answer

to

be

a

lie,

but

instead

an

afﬁrmation

of

her

involvement

in

and

commitment

to

the

church.

It

is

important

to

note

that

church

attendance

is

held

out

as

a

noble

activity,

indeed

a

generous

one.

One

attends

church

services

to

worship,

which

is

conceptualized

as

a

selﬂess

act

of

devotion,

a

giving

of

oneself

to

the

deity.

According

to

church

teachings,

then,

those

who

attend

“service”

are

by

deﬁnition

more

generous

than

those

who

do

not.

Going

to

services

is

good;

therefore,

all

else

being

equal,

people

who

go

more

often

are

better

people.

This

is

a

core

tenet

throughout

all

varieties

of

monotheistic

reli-

gion.

Those

who

attend

are

reminded

of

it

every

time

they

go.

Because

of

this

steady,

long-term

reinforcement,

vast

numbers

of

people

are

likely

to

describe

the

nobility

of

church

attendance

as

a

truth

they

know

in

their

gut.

This

is

precisely

the

kind

of

widespread

assumption

that

leads

to

biased

reports.

It

should

also

be

mentioned

that

in

some

places,

being

seen

as

a

devoted

practitioner

of

the

local

faith

is

not

optional.

Those

who

refuse

risk

extreme

sanctions

from

their

more

zealous

neighbors.

To

the

extent

that

religious

participation

is

subject

to

such

pressures,

one’s

frequency

4This

is

perhaps

a

contentious

claim.

Nisbett

and

Wilson’s

(1977)

paper

has

been

cited

over

6,000

times.

The

other

points

in

this

section

do

not

depend

on

this

one.

R.

Sablosky

/

The

Social

Science

Journal

51

(2014)

545–555

549

of

participation

is

a

measure

of

conformity

rather

than

of

belief.

This

is

true

even

if

we

measure

actual

rather

than

reported

attendance.

2.3.3.

Which

proxy

is

more

accurate?

How

can

we

tell?

We

have

seen

that

there

are

problems

with

using

either

personal

belief

or

church

attendance

as

a

proxy

for

reli-

giosity.

A

deeper

issue

is

revealed

if

we

ask:

Which

proxy

is

closer

to

the

true

nature

of

what

we

are

trying

to

mea-

sure?

This

question

cannot

be

answered

by

science.

It

is

a

matter

of

opinion—of

persuasion.

Consider

that

for

any

behavior,

attitude,

or

quality

that

one

party

holds

up

as

the

epitome

of

religiosity,

another

party

can

say,

“But

that’s

not

really

being

religious.

It’s

really

this—”

and

name

something

else.

There

is

no

independent

standard

against

which

such

claims

can

be

checked.

Consider

the

following

questions,

all

commonly

used

to

measure

religiosity.

1.

Do

you

consider

yourself

Protestant,

Catholic,

Jewish,

Muslim,

another

religion,

or

no

religion?

2.

How

often

do

you

attend

devotional

services?

3.

How

often

do

you

participate

in

church

activities

other

than

devotional

services?

4.

How

strongly

do

you

agree

with

the

statement,

“People

need

to

obey

the

Ten

Command-

ments,

because

they

are

the

exact

word

of

God”?

The

ﬁrst

question

is

about

group

identiﬁcation;

the

second,

ritual

behavior;

the

third,

voluntary

association;

and

the

fourth,

social

policy.

These

are

not

just

differences

in

emphasis.

The

questions

refer

to

different

realms

of

investigation.

It

is

impossible

to

measure

a

property

across

ontological

categories.

It

might

be

charged

that

if

we

press

hard

enough,

no

sociological

term

has

a

precise

deﬁnition.

But

extreme

pre-

cision

is

not

necessary.

Consider

the

concept

of

altruism.

In

sociology,

altruism

is

behavior

that

beneﬁts

other

people

at

a

cost

to

oneself.

Though

not

very

precise,

this

deﬁnition

is

precise

enough

to

be

useful.

Room

remains

for

disagree-

ment

over

details,

but

the

broad

outlines

of

the

quality

to

be

measured

are

not

in

doubt.

For

the

concept

of

religiosity,

on

the

other

hand,

even

the

broadest

outlines

are

unformu-

lated.

Even

among

traditionalists,

there

is

no

consensus

on

what

it

refers

to.

Without

such

consensus,

it

is

meaningless

to

speak

of

measurements

of

this

property

or

the

statistical

analysis

of

such

measurements.

2.3.4.

The

recruitment

effect—an

alternative

hypothesis

Let

us

imagine

that,

despite

the

many

methodological

and

conceptual

hurdles,

we

determined

with

some

conﬁ-

dence

that

frequent

churchgoers

really

are

more

generous

than

people

who

go

less

often.

It

is

important

to

recognize

that

at

this

point

we

still

would

not

have

shown

that

their

church

participation

is

the

reason

for

their

generosity.

There

could

be

any

number

of

other

explanations

for

the

corre-

lation.

In

particular,

the

possibility

would

not

have

been

ruled

out

that

the

generous

people

we

ﬁnd

in

churches

are

in

church

because

they

are

generous,

rather

than

the

other

way

around.

Inferring

from

an

observed

correla-

tion

between

religiosity

and

generosity

that

religion

makes

people

more

generous

requires

(among

other

things)

the

premise

that

religious

inﬂuences

have

a

shaping

effect

on

individual

personality.

If

religiosity

is

related

to

prosocial

behavior

under

some

contexts,

it

is

possible

that

having

a

prosocial

disposition

causes

one

to

be

religious

or

that

a

third

variable

(such

as

dispositional

empathy

or

being

prone

to

guilt)

causes

both

prosocial

and

religious

tendencies.

(Norenzayan

&

Shariff,

2008,

p.

60)

Now

consider

the

following

hypothesis.

Because

of

the

reputation

that

religion

has

established

for

itself

in

our

culture,

people

who

are

naturally

generous

are

likely

to

believe,

whether

or

not

it

is

true,

that

joining

a

church

will

help

them

accomplish

the

magnanimous

tasks

they

enjoy

performing.

On

average,

this

makes

such

people

more

motivated

to

seek

out

congregational

membership.

That

is,

the

average

Good

Samaritan

is

more

likely

to

join

a

church

than

is

the

average

skinﬂint.

This

“recruitment

effect”

is

a

form

of

selection

bias

that

would

neatly

explain

the

cor-

relation

between

religiosity

and

charity—again,

assuming

that

such

a

correlation

were

reliably

detected—but

with

the

direction

of

causation

opposite

to

what

is

traditionally

claimed.

Saroglou

(2010)

compares

personality

characteristics

from

the

Big

Five

model

(McCrae

&

Costa,

2008)

to

indi-

vidual

variations

in

personality.

In

a

meta-analysis

of

71

studies

from

19

countries

(N

=

21,715),

religiosity

is

observed

to

be

closely

correlated

with

agreeableness

and

conscientiousness.

According

to

Saroglou

(p.

108),

these

two

personality

characteristics

“seem

to

predict

religious-

ness

rather

than

be

inﬂuenced

by

it”;

that

is,

they

“can

be

considered

predispositions

of

religiousness.”

This

is

consistent

with

the

recruitment

effect

in

that

it

suggests

that

personality

leads

to

religious

participation

rather

than

the

reverse.

In

longitudinal

studies

of

adolescents

by

McCullough,

Tsang,

and

Brion

(2003)

and

by

Heaven

and

Ciarrochi

(2007),

personality

attributes

at

an

earlier

time

are

predictive

of

religious

values

at

a

later

time.

These

ﬁnd-

ings,

too,

are

consistent

with

the

recruitment

effect.

2.3.5.

Putnam

and

Campbell’s

religious

social

networks

The

recruitment

effect

described

in

the

preceding

sec-

tion

throws

an

interesting

light

on

a

core

proposal

of

Putnam’s

Bowling

Alone

(2000)

and

Putnam

and

Camp-

bell’s

American

Grace

(2010).

In

the

latter

book,

Chapter

13

(“Religion

and

Good

Neighborliness”)

begins

with

a

series

of

statistical

arguments

in

support

of

the

proposition

that

religiosity

is

strongly

correlated

with

helping

behavior.

However,

after

18

pages

of

such

persuasion,

the

presen-

tation

takes

a

left

turn.

Under

the

subheading

“Correlation

or

Causation?”

we

read:

Nonexperimental

research

cannot

exclude

the

potential

effects

of

self-selection,

that

is,

the

possibility

that

some

unexpected

factor

(perhaps

even

something

genetic)

induces

both

religiosity

and

neighborliness,

producing

a

spurious

correlation,

so

that

simply

forcing

people

to

attend

church

more

often

would

not

make

them

more

neighborly.

(Putnam

&

Campbell,

2010,

p.

461)

Putnam

and

Campbell

then

test,

and

discard,

the

hypothesis

that

“values

and

beliefs”

are

the

“secret

ingre-

dient”

that

makes

some

Americans

more

generous

than

others

(p.

463

ff.).

Helping

behavior,

they

ﬁnd,

is

not

550

R.

Sablosky

/

The

Social

Science

Journal

51

(2014)

545–555

causally

related

to

any

of

the

25

measures

of

religiosity

used

in

their

Faith

Matters

survey

(p.

466),

nor

to

their

own

“index

of

religiosity”

(which

they

deﬁne

on

p.

19).

Having

begun

the

chapter

seemingly

with

a

case

for

religiosity

as

the

“secret

ingredient,”

they

turn

around

and

strongly

deny

this.

Only

then

do

they

identify

the

new

construct

that

they

believe

is

doing

the

causal

work

in

fostering

generosity:

“religious

social

networks”

(p.

471).

The

proposal

is

paradoxical.

If

none

of

the

meas-

ures

associated

with

being

religious

is

essential

to

these

networks,

why

call

them

religious?

That

such

activities

take

place

in

a

nominally

religious

setting

is

not

enough

to

support

the

argument.

People

who

have

had

a

broken

leg

successfully

set

at

a

Catholic

hospital

will

credit

their

subsequent

recovery

to

the

competence

of

the

staff,

not

to

the

hospital’s

religious

afﬁliation.

According

to

Becker

and

Dhingra

(2001,

p.

329),

“Social

networks,

rather

than

beliefs,

dominate

as

the

mecha-

nism

leading

to

volunteering,

and

it

is

the

social

networks

formed

within

congregations

that

make

congregation

members

more

likely

to

volunteer.”

Cnaan,

Kasternakis,

and

Wineburg

(1993,

p.

44)

ﬁnd

that

“religious

belief

is

not

a

key

factor

in

social

service

volunteering.”

Putnam

and

Campbell

(2010,

pp.

472–473)

themselves

deny

that

religious

belief

fosters

generous

behavior:

The

statistics

suggest

that

even

an

atheist

who

hap-

pened

to

become

involved

in

the

social

life

of

a

congregation

(perhaps

through

a

spouse)

is

much

more

likely

to

volunteer

in

a

soup

kitchen

than

the

most

fervent

believer

who

prays

alone.

The

evidence

.

.

.

is

unequivocal—when

it

comes

to

the

religious

edge

in

good

neighborliness,

it

is

belonging

that

matters,

not

believing.

It

seems

odd

to

stipulate

that

an

atheist

could

be

an

exemplar

of

this

“religious

edge.”

As

we

saw

above,

one

could

hypothesize

that

generous

people

are

under

the

impression

that

religion

is

a

way

to

accomplish

good

things,

and

that

this

makes

them,

on

aver-

age,

more

likely

to

join

a

church

than

selﬁsh

people

are.

If

we

were

looking

for

data

in

support

of

this

recruitment

effect,

it

might

look

a

lot

like

Putnam

and

Campbell’s

data

on

“religious

social

networks.”

In

his

earlier

book,

Putnam

(2000,

p.

66)

writes

approvingly

of

churches

as

a

friendly

training

ground

for

civic-minded

people:

Churches

provide

an

important

incubator

for

civic

skills,

civic

norms,

community

interests,

and

civic

recruit-

ment.

Religiously

active

men

and

women

learn

to

give

speeches,

run

meetings,

manage

disagreements,

and

bear

administrative

responsibility.

But

such

beneﬁts

are

not

unique

to

churches.

Nor

are

churches

necessarily

the

most

efﬁcient

way

to

provide

them.

Religious

beliefs

are

not

the

only

reason,

and

probably

not

even

the

principal

reason,

that

people

join

religious

organizations.

Roberts

and

Davidson

(1984,

p.

347)

ﬁnd

that

of

all

the

sources

of

religious

involvement

they

measure,

“beliefs”

are

the

least

important.

Religious

orga-

nizations

promise

spiritual

beneﬁts,

but

they

also

provide

material

beneﬁts

such

as

child

care,

professional

network-

ing,

access

to

mating

markets,

and

the

opportunity

to

sing

in

front

of

a

sympathetic

crowd.

People

may

be

attracted

to

the

organization

for

many

reasons

that

are

independent

of

the

religious

messages

that

are

supposedly

its

raison

d’être

(Ellison,

1995,

p.

92;

Stark

&

Bainbridge,

1980,

p.

1394).

And

beliefs

may

be

adopted

after

membership

is

obtained—sometimes,

as

Stark

and

Bainbridge

(p.

1377)

put

it,

“faith

constitutes

conformity

to

the

religious

outlook

of

one’s

intimates.”

Note

that

in

such

cases,

future

surveys

will

classify

these

new

members

as

having

the

values

and

beliefs

associated

with

the

organization.

But

their

beliefs

did

not

cause

them

to

join;

it

was

the

other

way

around.

Putnam

himself

(2010,

p.

119)

ﬁnds

non-religious

orga-

nizations

more

effective

than

religious

organizations

in

promoting

all

types

of

volunteerism—as

do

Campbell

and

Yonish

(2003,

p.

105)

and

Smidt

et

al.

(2008,

p.

168).

There

are

similar

ﬁndings

for

monetary

donations.

Participation

in

a

voluntary

association

of

any

kind

boosts

secular

char-

itable

donations;

but

where

the

organization

is

religious,

the

effect

is

not

stronger;

it

is

the

same

or

weaker

(Brooks,

2005,

p.

10;

Monsma,

2007;

Nemeth

&

Luidens,

2003,

p.

118;

Putnam,

2000,

p.

120;

Smidt

et

al.,

2008,

p.

120).5

Brown

and

Ferris

(2007,

p.

94)

ﬁnd

that

“once

stocks

of

social

capital

are

controlled

for,

religiosity

decreases

giving

to

secular

causes.”

Jackson,

Bachmeier,

Wood,

and

Craft

(1995)

ﬁnd

that

belonging

to

a

range

of

voluntary

organizations

signiﬁ-

cantly

increases

both

volunteering

and

donations,

whether

or

not

any

of

the

organizations

are

religious.

Belonging

to

more

than

one

association

increases

the

likelihood

of

hear-

ing

about

humanitarian

activities.

Simply

being

asked

is

one

of

the

most

important

factors

in

how

often

people

vol-

unteer

(Brown,

1999,

p.

29).

Therefore,

people

who

belong

to

more

organizations

tend

to

volunteer

more

often.

This

“mobilization

effect”

(Bekkers,

2002,

p.

8)

is

not

unique

to

religion.

Therefore,

even

if

the

network

effects

that

Put-

nam

and

Campbell

observe

are

real,

there

does

not

seem

to

be

any

reason

to

hypothesize

that

religion

is

funda-

mental

to

their

operation.

After

all,