Cambridge University Press et al v. Patton et al

Filing 324

DEPOSITION of Daphne Greenberg, Ph.D. taken on April 21, 2011 by Cambridge University Press, Oxford University Press, Inc., Sage Publications, Inc.. (Attachments: # 1 Exhibit D-1, # 2 Exhibit D-2, # 3 Exhibit D-3 - part1, # 4 Exhibit D-3 - part 2, # 5 Exhibit D-4, # 6 Exhibit D-5, # 7 Exhibit P-1)(Rains, John)

E
X
H
I
B
I
T
3

Chapter

Research Design and
Issues

Methodological

for

Adult

Development and Learning
Grace

Both

adult

areas

of research

search

that

in their

practices

opment and learning
to

opportunity

but

also

about

Decisions

and

related

the

made by
not

the

aim

here

sues
types

of

those

to

is

involved

working

provide

points

research

some

topic

the

explanation

blending

together

priateness

in

this

that

combined

of

which

of

just

one

are

greater
are

may be
of

entire

topic

first

key

of

The

findings
issues

section

of

statistical

cance

with

ones

research

the

and

to

generalizability

mea

discusses

appro

measurement

scale

clinical

versus
the

evaluating

the

development

briefly addresses

chapter

significance

respect

of

how

reliability

validity
scales

change

of

the

third section

measure

The

and

issues

sampling

including

of

designs

sequential

influence

cross-

the

signifi

importance

of

findings

is

for

STUDY

TYPES

devel

at various

on

texts

The

these

final

issue

be

Rather

useful

adult

referred
detail

to

can

including

descriptive

various

evaluating

areas

Readers
with

the

of

The

asked

as

chapter

overview
and

the

such

Issues
of

be

need

pursued

research

on

focuses

sample

measurement

disciplines

should

that

be

designing

publications

lar

an

of

of

type

surement

endeavors

in

opment and learning
to

scope

the

and

types

study

quasi-experimental

section

research

researchers

in the

devel

several

longitudinal

second

of

separate

questions

individual

included

are

sectional

unique

previously

should

of

the

in

challenges

that

youth

experimental

re

of adult

describes

chapter

tradition

new

the

theories

of

Willis

Sheriy

have

investigators

yet

the

own

its

type

relative

the

intersection

out

new

face

two

these

carve

has

the

to

respect

Given

the

at

Each

right

and

established

are

learning

own

with

conducted

is

conducted

research

and

development

and

traditions

Caskie

Experimental

Studies

particit

devoted

to

section

of

In

an

experimental

assigned

to

study

membership

in

participants

one

of

Defendants
52

Exhibit 3 - 1

two

are
or

randomly

more

Exhibit

6Eeie6-

IESQ

groups

RESEARCH

that

will

be

clinical

trial

compared For example
on

et

al 2002

to

one

of

group

the

than

behavior

Thus

for

word

should
in

artificiality

mental

the

are

the

results

too

of

generalizability
leads

trols

in the

lab

that

desigus

are

discussed

number

useful

for

interventions
of

the

comparison

relation

to

the

experimental group

pre-

and

erwise

is

no

important

intervention
that

may

result

than

once

i.e

the

of experimental
design

group
factors

to

all

The placebo

design

on

the

the

behavioral

placebo
outcome

of

being

to

must not
the

treatment
measures

it

that

arid

oth

but

any

in

gain

tested

the

more

has

type

group
control

include
For

may be

that

type
of

intervention

research

de

same

This

control

treatment

to

the

The second

nonspecific

and

vention
cient

confounding
of

trainers

having

conducting

at

all

site

ito

will

issues

practical
the

including

few

in

intervention

ex

necessity

at

the

same

with

the

type

of

inter

who

are

equally

interventions

multiple

avoid

to

com

salient
result

profi

intertentions

in

evidence

initial

estimate

treatment

critical

administering

of

Be

of factors

most

across

design

con

comparison

may

the

and

number
this

of

Although
design

number

on

differ

to

Studies

the

group

battery

contact

effect

placebo

considered

and

an

practice

design

provides

chological

and

simply from

use

of

Quasi-Experimental

control

for providing

effect

score

or

of

the

in

problem

useful

type

constant

Additionally

with

in

different

effective

very

Willis

design

vary

used

receives

receives

assessment

intervention

be

designs

group

that

that

group

postinterventinn

receives

behav

ist

be

change

held

being

more

optimal

Hofland

same

determine

to

producing

the

distincfly

the

this

may

difficult

for

groups

experimental

These

be

con

of

to

using

experimental

comparative

more

sessions

training

needed

be

all

determine

the

or

of

several

for

to

hrst to

interventions

factors

conducting

or control

no-treatment

comparison

of design
an

the

First

uses

cause

can

con

number

may

with

exist

ponent

level

two

at

cerns

the

experimental

of

types

adults

might seem

this

compo

else

while holding

Finally

which

can

groups

intervention

para

approaches

determine

in

an

the

sessions

comparing

of

different

varying

compared

1981

of

imple

the

by mie

training

and

identify

components

the

in

therefore

Fourth

the

increased

tervention

it

Having

decrease

volves

site

in terms

intervention

of

difficult

generalize

havioral

sign

effect

selling

particularly
with

rc

Lon

women
also

to

limit

Human

for
the

in the

five

may be

and

separately

components

factor

example

Baltes

if

may be

it

to

can

difficult

For

experi
situa

various

is

systematically

experimental

stant

of

National

for

greater

world

real

2001

Willis

more

it

to

study

findings

artificiality

one

number

may

Center

to control

controls

design

to

effective

of

condi

design

components

requirement

of

most

involves

of

real

the

combination

or

being

example

e.g

Men

study

greater

making

setting
ior

to

For

of the

studys

many experimental

con

would

Although

study

1992

the

generalize

assign

the

various

ability

Comparison

identify

nent

to

the

is

design

group

implemented

essential

independently

help

of

are

An

tervention

metric

concern

variables

Older

of

the

approximate

men

this

control

that

many

included

only

Research
on

gender

too

intervention

which

in

53

ISSUES

their treatment

to

control

component
design

combination

groups

studies
to

of the

equally

researcher

about

said

designed

Surveys

Resource

be

for

be

to

the

of participants
the

hetween-group

ment

used

is

and

factors

Third

variable

the

validity

validity of the

study

gitudinal

to

also

controlling

search

allow

blindness

tion

using

random

of

the

variables

should

process

designs

experimental

studies

external

the

ex

an

influence

may

experimental

internal

studys

tions

studied

confounding

possible

decrease

grouped
that

through

ment experimental
trol

i.e

being

among

group

or that

variable

the

control
to

variable

are

or construct

or

groups

Ball

assigned

randomly

control

independent

outcome

distributed

groups

training

or

adults

randomly

participants

participants

the

predict

cognitive

condition

which

were

ACTIVE

in the

of older

training

participants

assigning

perimental

on

the

three

By

other

cognitive

AND METHODOLOGICAL

DESIGN

any
psy

at

maintain

studies
is

groups

exist for

entirely

groups

the

such

existing

These

as

studies

1979

ies

studies

shortly

lii

pendent

are

studies

practice

are

educational

age

studies

be

level
studies

Campbell
and

differences

described

in greater

assigned

include

stud
to

an

multiple

variables

or

grouping

variables

which

mixture

of

randomly

assigned

variables

Exhibit 3 - 2

as

pre

of quasi-experimental

cannot

most

compare

randomly

Cook
of

changes

examples

individuals

to

quasi-experimental

1953

of age

he

of interest

or

age

called

Cross-sectional

because

include

gender

Stanley

longitudinal
detail

characteristics

are

Campbell

wants

to

low-

appropriate

cannot

individuals

because

signed

and

researcher

which

to

random assignment

which

possible

sometimes

ever

level

difficult

effect

Many

age

inde

may
and

FOUNDATIONS

54

i.e

classificatory
that

define

the

1990

the

compared

with

three

verbal

sificatory

perimental

which

to

signed

When

is

unknown

may

different

For example

cohort experiences
influence

internal

the

factors

differences

behavior

were
than

other

pose

we

levels

were

of

the

randomly

e.g

in

as

training

fac

the

variable

sis

different

for

belier

of

the

memory

can

that

are

examine

decrease

variable

memory
number of

other
be

with

related

variables

conducted

condition

on

struction

and

If

the

signment

of

all

the

experimental

and

would

it

e.g
due

to

who

in

improvement
the

receiving

received

the

gender

in

the

assignment

had

the

sample

would

have

been

the

experimental group

and

founded

may

variables
not

insidious

tion

been

such

be
as

as

if

as

such

as

in

older

and

ban

campuses

all

real

group

adults

younger

with

Alternative

are

conducting
and

lion

than

Two

methods

are

basic

likely

to

be

used

quantitative

all

random

into

Con

group

exam-

maybe
or

loca

from subur

explanations

for

researcher

or she

under
in

into

When
he

in

center
ords

has

study

viduals

qualitative

may

they

be

also

oriented

The

For

between

supported

1988

study

Exhibit 3 - 3

in

simply

and

et al

in

behaviors

and

1980
and

in the

men

rec

demon

interactions

staff

of

indi

citizen

then

are

members

whether

they

resident

Am

naturalistic
of

the

senior

that

terms

behaviors

conducted
the

store

residents

individuals

observes

behaviors

coded

observation

the

e.g

grocery

in Baltes

home

dependence

which

or

about

example

also

with

environment

park

nursing

observed

were

naturalistic

contact

researcher

natural

library

quantitatively

uses

direct

no

information

strated

ato

more

in descriptive

considered

although

for

observa

naturalistic

The methods

not

may

study

are

women

site

more

incorporated

fact that

yet

all
study

groups

study

randomly

and

at

descriptive

case

whether

If

to

assigned

rather

come from an urban
adults

be

not

are

conditions

any experimental

are

hypothetical
studies

assigned

studies

distributed

control

this

in

confounding
all

equally
the

to

measure

the

men and

the

obvious

environment
college

used

rela

Studies

individuals

studies

descriptive

partici

the

women

were

In

been

interactions

or to

instruction

have

of

as

fathers

outcome

parentchild

instruction

of

in

to

mothers

untangle

to

impossible

observed

differences

any

be

with

child

of

were

would

condition

their

explain

are

could

sup

assigned

entirely

there
that

status

tionship

parentchild

techniques

completely confounded
pant

effect
of

quality

consisted

group

the

participants

child-rearing

control
to

on

tlse

because

that

by

causal

rathcr

For example

on

study

instruction

corre

reported

differences

group

variables

caused

we

performance

made

at

ba

have

if

courses

socioeconomic

these

to

be

nut

e.g

variables

However

memory

of

responses

sessions

training

education

adult

could

conclusions

more

ses

strong

number

between

one would

perfonnance

participant

con

to explain

trying

lated

i.e

to

ability

that

concluding

and

of correct

training

to

participants

of training

design

number

variable

assigned

If

number

the

found

were

and

between

dependent

variable

to

sta

simply

is

randomly

parametric

sessions

end

were

to

used

measure

relationship

the

independent
assigned

correlation

it

and

individuals

or not

each

experimental

extraneous

interactions

as

the

linear

variable

independent

whether

the

is

random assignment

statistical

relationship

describes

mislead

is

conclusions

Correlation

Descriptive

rearing

pie

that

of

type

the

which

dependent

we

the

the

causal

whether

on

the

positive

to

other

to

represent

studythe

that
to

the

by

or

variables

are

in the

we want

caused

due

than

describe

sions

young.and

education

the

or construct

which

or for

as

development

validity of

that

clude

those

compared

independent

age groups

and

learning

as

truly

gender

the

be

randomly

are

are

including

Confounding
the

females

or

with

covary

not

such

differences

i.e age

difference

that

tors

whether

ex

to

used

was

tistic

cannot
regarding

were

groups

clas
the

contrast

design

males and

is

make

to

correla

that

imply causation

ability

dependent

actually

study

and

conclusions

individuals

or

conditions

This

in that

ing

admonition

necessarily

the

used

been

not

when

considered

be

always

well-known

the

does not

lion

experirnen

six

cueing

In

group

preexisting

individuals

group

and

quasi-experimental

variables

adults

has

assignment

Thus

included experimental

it

causal

design

made in

old

age

of

random

Dudley

old

integration

integration

variable

one

conditions

because

quasi-experimental
conditions

verbal

cueing

and

young

to

assigned

and

should

differences

group

compared For ex

Puglisi

of

variables

assigned
be

will

recall

two

conditions

crossed

that

groups

who were randomly
tal

randomly

Smith Morrell

Park

ample

not

caring

observation
for

young

DESIGN AND METHODOLOGICAL

RESEARCH

children

en

in

public

naturalistic

la
rd-

is

iit

to

in adults

learning

may be more

ample one

might

turn lead

en

efficacy

le

acy

to

number

its

to

their

might be

nal

In contrast

tact

to the

observation

at

study

the

tense

detail-gathering

study

only

way

studies

of

case

be found
of

Examples
more

2001

studied

were

jects

about

questions

and

of

lyzed

and

rated

process

data

collected

from

about

their

over

their

late

lives

forties

the

case

studies

Paul

and

John

tail

incorporation
chological
study

of

that

specifically

identity

in

open-ended

were

sessions

Schaie

late

determine

or

men

calculator

adolescence
in great

to

de

administration
with

of
the

men and

the

psy

to

be

to

In

other

may

be

that

period

their

time

Exhibit 3 - 4

the

data

between

the

less

than

of

may

the

age

individ

experience

held

of learning

cal

or without
addition

the

the
in

on

it

by hand

experience

In

collected

that

older

with

increases

with

on

if

adults

problems

rather

tisne

of

it

cross-sectional

only

one

should

measure

participants

at

co

frilly

i.e niatura

younger

may have

influence

are

birth

For example

problems

effects

cross-

cannot

age

is

de
Baby

born

in

differences

explanation

mance cannot be examined
because

different

study

cohort-specific
the

the

was

from

ability

the

typically

used

outperfonn

words

is

e.g

years

increased

by hand

that

participants

observed
to

cohorts

problems

ment i.e

any

alternate

younger

noted

case

study

in cross-

groups

age

mathematic

do simple math

culator

the

developmen

however

simple mathematics

an

in

of

differences

found

that

However

the

due

are

cohort

of

test

working

40

about

confounded

of

must be drawn

whether

to

mind
are

cross-stctional

were

book

of

researcher

longitudinal

range

group

then

age groups

uals

e.g
cross-

and

membership

or the

which

Thus

appear

on

con

Cross-sectional

particular

in

keep

Cohort

year

in

study

included

demonstrates

the

the

sectional

the

to

age

definition

horts

study

time

inform

age changes

and cohort

studies

by

timed

James William

may occur

of

l3oomers
By

in

in

is

some

abortion

planning

important

sectional

Mea
ana

when

interest

1996b

is

fined

can
for

to

necessary

process

males

presented

both

bends

various

Lachman

point

also

self-

or

in

on

appropriate

very

possible

adults

also

is

cognitive

al 1986

differ

about

point

exam

eta1 1990

1984

particular

differences

1978

for

work

study

range

tal

at

other

researchers

attitudes

infor

grouped

have

performance

groups

age

of age

tion

collected

were

interviews

men

sectional

study

case

psychotherapy

of

of

and

Honos-Webb
on

reported

and

and

development

individuals

Levinsons

interviews

measures

approach

of four

Cohn
of

be

to

influences

identity

also

behavior

It

The two sub

of

Levinsons

Levinson

tend

studies

identity

period

or behavior

vnting

influences

also

composed

multiple

men

one

at

Park

et

Costa
for

When

various

struct

and

1998

Jelalian

1986
how

are

studies

memory

McDowd

Lachman

knowing

age

numerous

attributions

McArthur

conduct

to

semi-structured

development

the

con

be

e.g Cervone

self-esteem

Finally

of adult

also

how

of

2003

The

the
the

gather

assessed

are

grotsps

in

to

is

Participants

characteristics

and

domains

to case

can

all

example

Craik

studies

individuals

use

given

study

differences

differences

age

about

case

of therapeutic

transcripts

For

training

case

Gillem

current

and

depression

In

1995

woman who had completed

sures

the

the

their

and Greenberg

Stiles

the

as biracial

experiences

then

time

medical

research

primarily

on

in

age

and

age

ined

cross-sectional

about

by

of

about

in Stake

biracial

their

members

bmily

task
the

of

focus

mation

In addition

quantitative

with

schedule

terview

their

two

in

development

in

observation

situation

described

case

feasible

studies

study

than

qualitative

Throne

of

case

of
the

examination

or

under

make

to

more

information

can

The

personality

con

direct

method

the

direct

case

process

Detailed

inter

individuals

measures

individuals

study

the

using

include

or psychological

2004

the

an

uf naturalistic

individuals

and

records

of

liter

observing

by

extensive

and

process

to interviews

ducted

self-

to assess accurately

involve

few

involves

method

may
and

Computer

method

descriptive

research

study

addition

in

efficacy

researcher

Typically

computer

which

self-perception

difficult

studies

case

between

in

alone

observation

of

but

may be more

process

through

ig

made

as
to

Studies

Cross-Sectional

skills

measured

indirectly

of errors

such

di

skills

computer

tests

also

ex

lack

in their self-perceptions

changes

about

of

study For

adults

new

learn

the

under

that

expect

Test

development

given

into

and
were

interviewed

matic Apperception

observe

to

were

wives

Although

used

of adult

individuals

would

course

ta

the

be

types

to tap

difficult

with

literacy

could

certain

rect contact

observed

were

places

observation

55

ISSUES

perfor
study
point

in

FOUNDATIONS

56

When
tial

vant

between

be

that of equating

is

that

demographics

the

may

the

with

equivalent

the

of

bias

For

individuals

in

survivorship

for inclusion

the

is

the

into

exam
may

it

reexamined
Steen

possi

selection

of

study

at

have

again

changes

The
nal

data

same

twice

studies

Alwin

sometimes
ent
that

this

be

found

because
not

attributed
of

type

show

research

on

shown

and

the

the
adult

wide

were

that

but

ent

nal

data

whereas

data

gitudinal

sectional

difference

age

most
to

With

design

this

age

time

For

age

are

the

their

senior

with

this

and

Nordie
of

year

in

complex
Research

75-year-old

Denmark

in

set

membership

group

at

of

two

more

their

and

and

in their

single-cohort

to

environment

ment

of sim

occasions

1987 exam

to

from

Finlandwho

twice

twenties

slightly

study

included
three

during

up

late

longitudinal

individuals

in

the

of

later

during

which

of

age changes

one

case

two

sort of

change

the

influence
implication

Elder

possible
50

or

to

that the
the

that

change

in

had

for the

to

years

In

as
this

change

in

occurrence

of

11 2001

September

in

likeli

specific

perceptions

due

States

between

years

assume

likely

the

States

from

age

three

had

cohort

this

anxious

peoples

on

we

perceived

are

people

reasonably

occurred

than

to

during

period

influence

also

the

United

the

more

maturational

three-year

that

the

about

United

the

on

explanations

during

2000

in

reassessed

on

as

attacks

Exhibit 3 - 5

that

in age

measure

of

their opinions

we

If

times

re

behav

about

changed

different

found

become

might

terrorist

Caspi

has

attack

attack

the

The

increase

the

and

perceptions

that

individuals

the

was more

any

observed

that

and

the

the

certain
to

that

means

effects

year

occurred
to

vari

dependent
period

increase

sessments

de

longitudinal

the

in

50-year-olds

53

they

re

the

construct

time

terrorist

the

countries

were

at

years

events

the

study

single-cohort

something

terrorist

observed

in the

calendar

be

least

accompanies
inform

The confound

next

of

study

that

effects

the

of

single-cohort

i.e

due

likelihood

53

in

differences

example suppose

group

can

cannot

changes

between

For

lon

be found

ii

age

that

hood

followed

the

is

than

the

women

of

any

cannot

change

rather

with
of

cohort membership

same

of study

time-of-measurement

at age

over

the

cohort

assessed

is

2003

mid-fortics

project

and

specif

period

overview

extensive

experiences

type

over

change

adulthood

time-of-measurement

2004 Schaie

more

mid-

on Aging

Sweden

continued

studies

development

Also

within

changes

Moane

college

early

adult

into

with

and

individuals

similar

studied

being

cross-

profound

comparable

of individuals
or

sample

once

de

childhood

2001

share

this

about

searcher

asked

lon

variables

longitudinal

of

personalities

finally

more

single

observed

cohort

of

ioral

subject

obtained

Schaie

single-cohort

is

example Helson

ined

and

the

An

the

of

study

searcher

abilities

in longitudi

for assessing

design

all

longitudinal

creased

basic

time due

ilar

do

some depend

dependent

pattems

Because

occurring

example

same

replicated

in the

caution

1968

Strother

The

was

stud

types

continued

the

adult

Hofer

with

cross-sectional

For

reflected

Sehaie and

confounds

studies

differences

shown

information

three

of

adulthood

assess

of

studies

gitudinal

able

For

the

not

changes

than

on

age

were

for other

age

decrement

between

age range

substantial

variables

been

or later

samples

sign

mental

of

to

designed

variable

with

trends

collected

wide

over

cross-sectional

age

development

data

which

locations

understand

young

midlife

per

that longitudinal

have

assessments

of

into

three

longitudinal

to

we

those

individuals

stated

begun

with

studies

However

age

treated

differ

concluded

longitudinal

discrepancies

longitudinal

population

same

ovcr

at

dependent

must be

as

have

Investigators

individuals

the

and

as well

change

chronological

conclusion

2001

development

representative

in

longitudi

study and

on

to

cross-sectional

always

has

samples

cross-sectional

differences

could

such

of

time

over

time

over

2001

age

understanding

to

of

predictors

compared

i.e

ages

essential

Campbell

exaririiie

through

or behavior

and

to obtain

Hofer

arid

in adult

velopment

individuals

gained

are

construct

variability

time

of

to

is

study

group

collections
in

change
the

longitudinal

assessments

multiple

from

75 years

ically

the

albeit

aged

example

Studies

Longitudinal

in

of

assessed

assessments

of

cohort

Schroll

Berg

second

this

single

was

ies

Heikkinen
With

1997

sons

hood

The purpose

80

age

Viidik

Sehaie

educational

there

Furthermore

rele

relationship

variable

groups

to find

experiences

individuals

age

influence

dependent

poten

groups on

old

of

older

study

and

or

bility

the

young

comparing

difficult

levels

and

age

in

ple

cross-sectional

conducting

problem

of

the

for

and

the

meaning

Bem

time
of

finding

1987

of

measurement

particular
in

longitudinal

found

that

can

construct

or

study

sample

of

RESEARCH

II

ill-tempered

ye

tain

which

adults

reached

adulthood

leagues

on

if

pcriods

to

perform

ability

had

less

tudinal

better

times
in the

1920s

worked

outside

the

poverty

55

different
lescents

ft

status

would

for

suc

can

particularly

of

ipants

men might

asked

in

longi

implications

at

male

ado

cluded

who

contend

indicator

of

whereas

longitudinal

single-cohort
Several

on

design

of these

discussed

design

wherein

one

with

intent

the

tion

on

also

be

to

designs

this

2001

and

religion

existing information
the

collection

throughout
of

ple

from

prospectively
data

collection

tained
these
the
ies

1986

problems

the

use

First

religious

participants

willing

to

curred

over

complete
period

all

of

on

65

and

four

age

In

sign

participation

of

as accurate

that

of

utility

the

are

of the

as

the

as

decreased

to

accelerated

by

Dun

growth

for

true

lon

Hamagami

these

growth

amount of
difficulty

insight

data

could

sim

Large-scale

longitudinal

pa

longitudinal

and

This

further

de

Duncan
that

the

estimates

provide

patterns

data

corresponding

errors

the

for

both

and

MeArdle

increased

parameter

needed

Farring

modeling

growth

accelerated

standard

cohort

unstable

the

However

estimates

the

demonstrated

from

Duncan

distinct
into

stud

cohorts

studied

is

the

differences

Ohlin

investigated

1992

both

longitudinal

Duncan

cohort

latent

shorten

longitudinal

link

Tonry

incorporated

context

per

ulations

to

being

to

reduce

independent

one

1996

found

ob

no

are

design

also

research

developmen

cohort

no

ranges

Hamagami

were

also

for

of

several

age

estimates

gitudinal

must

accelerated

collecting

designs

and

collected

data

costs

Hops

the

two

of

assumption

from

the

and

lead

to

range

parameter

1996

aim

data

McArdle

also

designs

Because

missing

collected

demonstrates

into

the

design

stud

differential

Sequential Designs
Alwin

Campbell

of

from

obtained

healthy

waves

decade

years

colleagues

were

histories

who were

of

design

religiosity

sam

of retrospective

recall

with

included

Following

religious

entire

the

study

Li Alpert 1999

1991

ton

at

in

sampling

proposed

to

overlapping

Strycker

rameter

aged

her

for

as

been

necessary

time

collected

ied

may
for

measures and

such

have

longitudinal

data

which

outdated
designs

the

over

from longi

studies

Longitudinal

These

Under

i.e

lost

on

pro
partic

implications

time

and

accelerated

adulthood

and aging
of

reliability

those

and
the

This example

development

and

2001

with

of

Meador
between

had been

1996

Alwin

of

trajectory

can

data

adults

to

George

participants

vival

amount

data

section

possibility of

design

longitudinal

tal

and

the

time

over

attrition

Reasons

chapter
the

becoming

supplementing

by

study

information

histories

life

of adult

only

health

and

relationship

retrospective

community-dwelling
on which

older

participants

periods

in older

course Their

life

can

data

asking

Flint

adults

health

on

of

the

by

of the

study

in the

informa

Longitudinal

Hays

in older

section

data

of

this

with

monetary

individuals

waves

earlier time

George

health

of

group

retrospectively

about

later

collected

prospective

ftiture

of individuals

conducted

2004

type

basic

section

this

in

of data

studying

information

in

far

presented

of collecting

group

Campbell

so

considered

is

begins

obtained

recall

are

The

designs

sequential

the

beyond

exist

designs
discussed

in

are

in the

Time-efficient

the

to

results

time

of

period

is

longitudinal

to study

participants

addressed

questions

it

most adoles

for

extensive

the

study

time consuming

desires

of

for

longitudinal

and

use

all

at

Collecting

expensive

due

the
are

the

of

re

faulty
the

out

data

level

by

advantages

many

when one

studies

influence
trition

its

range Also

age

tudinal

an

an

over

wide

mother

for

typical

meaning

be

biased

no

time

over

current

pointed

to

preferable

disadvantages

cess

Gorman 2004

different

very

without

data

may have

2004

despite

change

the

having

was

of

not

today

Other

i.i

home

Of course
study

is

be

may be

periods

including

may

it

al

et

data

retrospective

ear

these

different

1930s

George

In

For example

and

Hayward

have

cents

in history

Yet

on

style

are

as

col
been

call

or

57

ISSUES

time

earlier

at

states

involvement

implication

the

for

that

have

also

and

focused

personality

outcomes

they

had

lime period

by

religious

by

lime

this

success

Questions

may

study

the

by

Caspi

labor

current

influenced

prerequisite

job

physical

and

impact

have been

poorer

as

during

where

confrontational

having
Ii

workplace

lime

lier

skills

involvement

influenced

outcomes

emphasis

greater

religious

to

As

partially

in society

that

interpersonal

in the

cess

It

least

Specifically

out

pointed

placed

at

was

main
result

tended

significantly

had occurred

that

changes

1928

adulthood

into

style

men experienced

these
as

born around

boys

this personality

AND METHODOLOGICAL

DESIGN

enough

testing

Second

the

that
recall

and

oc
of

Data

acquisitions

cross-sectional

can

be

extended

sequences

Exhibit 3 - 6

that
or

into

Baltes

are

initially

single-cohort
cross-sectional

1968

Sehaie

structured
longitudinal
or

as either
study

longitudioal

Baltes

1975

FOUNDATIONS

58

Sequential

studies

velopment

to

time frame

cross-sectional

from

tudinal

uals

every

into

2020
could

the

who were

uals

this

assess

45

In

gin

in

viduals

with

The
then

be

time

point

and

each

of the

of individuals

cal

included

longitudinal

in

cannot

original
the

rate

be obtained

most

tudinal

cohort

effects

effects

First

lime of

the

domly
with

at

the

an

age

lime chosen

under

problem

which

design

in

tween
from

midlife

capture
of

range

have
years

using

in

of

differ

information

are

cohorts
times
the

between

sign

are

rich

in that

analysis

changes

In

might include

can

they

effects

on

the
are

velopmentalists
age

and
analysis

be

variable

the

being

the

years

Second
of

widths

between

old

from

participants
that

match

the

at

this

full

age

with

1965

several
to

and

amount

measurements

of

Using

age

time
the

of

Many de

studied
in the

analysis

of

in

cohort-

that

time-of-

performed

assumption

Cmup

of

Example

Sample

Most

Data

Time

Time

data

collection

Bas ed on

Collection

Efficient

Design
Time

Time

Cohort

35...39years

to

consider

this

be

40-l4years

pass

case

period

of

range

con

time

mea

specific

individuals

from

de

efficient

the

who

to

age
are

lime point

drawn

are
the

first

to

new

of

data
most

cohort

interested

six

adding

analyzed

age

changes

under

from

testing

by Sehaie

of

most

cohort

Schaies

identical

individuals

midlife

relative

measurement

using

proposed

strategies

the

trast

by

initial

pro

multiple

repeated

cross-sectional

generated

as

64

to

was

ran

relationships

adulthood

to the

data

well

35

that

subjects

previous

as

each

would

we

individuals

be

can

of dif

at

from

ages
interval

data

The

sequential

sampled

example

as

of

added

are

example

included

process

retesting

longitudinal

samples adding

at the

successive

to study

in the

with

have

entire

shown

new samples

covers

testing

within

also

is

1965

defined

must be

study As an

older

The

that

5-year

tested

and longi

optimally

used

the

posed and would

TAaLE

is

data

collected

age

process

this

in

age

interval

measurement

be

to

where

in

of

are

longitu

were

Each

lime

years

measure

collection

there

age group

Time

year By

incorporating

is

and

wanted

interest

the

from each

people

as

individuals

3.1

of

4549years
49

to

ples

that

to pass

ages

criti

that

is

set of

in table

example

ferent

cohorts

second

within

short-term

The resampling

group

data

first

collection

the

at

lime

providing

data

initial

new

age

the

many

as

resampled

time-of-measurement

development

of

had

that

Schaie

interest

one

into

range

this

sequences

way

and

learning

each

Age

The age range

surements

also

new sam

evaluation

which

collection

intervals

age

and

range

data

initial

se

interindividual

systematic

of

years

The

cross-sectional

effects

indi

repealing

design

these

in

sequences

by

cross-sectional

efficient

1994 combines

1977

35

from

3539

age groups

about

be

cross-sectional

the

and

turned

of

is

at the

testing
to

plan

study

approaches

permits

in

might

investigation

two

change

of

they

by drawing

change

age

Schaies

the

in the

sequence

traindividual
ences

in

the

tervals

from

sam

the

4044

years

be

elapse

intervals

age

second

at the

restudied

of

studies

in the

of individ

with

years

obtained

between

difference

age

3034

i.e

by

dinal

5-year

e.g 3539

range

to

years

measurements

in

Then

and

retrieved

providing

point

formed

sequence

for the

of

individ
old

retest

cross-sectional

data

or

longi

until

years

grotips

age

once
two

group

cohort

2010

later

at

been

in

every

simple

years

be

age

participants

at

years

first

then

would

quence

ples

the

cross-sectional

in three

study

At

larger

second

drawn

rnent

these

45

the

be

years

only

retest

should

ples

for

planned
and

first

4549

example

additional

old

years
also

group

4044

and

30

contrast

2005

an

study

to

researcher
the

tween

random

studying

were

they

2015

sequence

longitudinal
cluding

planning

until

years

2010

in

retests

by

the

if

whereas

groups

age
For

might begin

2005

in

same
time

in

points

sequence

30-year-olds

observed

indi

of

sample

repeatedly

independent

each
the

covering

different

same

long

design

longitudinal

the

use

individuals

the

involving

cohorts

sequences

of

cohorts

more

or more

de

about

questions

true

use

sequences

from two

samples

allow

without

for

required

Longitudinal
viduals

also

can

be answered

sam

interest
intervals

expected

midlife

with

to pass

example

Note

Each

Exhibit 3 - 7

represents

for

particular

sample

RESEARCH

measurement
being
ferent

cohorts

would

than

sequential

and

group

age
40-

to

this

Time

and

44-year-old

analysis
until

pending

on

between

waves

wait
In

of

data

the

time-sequential

the

study hecause

sequential

no

occurred

have

on

cross-sequential

Time

contrasted

The

the
or

changes

year-old

age

same
the

i.e

would

be

ples
of

first

each

porate

found

those

from

tested

at

of these
of

tests

in Schaie

to

the

these

age

Time

i.e

dctailcd

that

effects

practice

and

attrition

and Caskie

age

also

cluded

be

2004

individuals
in the

cause
to

be

to

the

situation

and

SAMPLING

AND

of

after

other

The

way

viduals

researcher

who

will

has

important

the

studys

obtains

participate
implications

findings

In

other

the

actual

in his or her
for the

words

group

of

with

type

of

of

sam-

learning

more

investigation

generalizability
the

indi

graduation

students

ified

to

random
ensure

Exhibit 3 - 8

could
in

currently
disabilities

complex
sample
that

Pro
911

the

in

type

of

The

more

are

of

2001

will

for

applied
for

to

those

schools

high

Stratification

sample

preparation

designed

are

applied

their

itninediately

random

be
likely

findings

directly

of

ap

study

processes

Stanovich

curricula

sample

in

Frank

and

regarding

Iowa

use

important

characteristics

lack

in

eventually

study

research

be

high

graduates

Disabilities

sensory

findings

ed

curriculum

Sitlington

students

learning-disabled

life

2476

were

environment

or

from

graduated

be immediately and

Franks

year

special

the

Frank

on

of

af

former

of

be more

e.g

study

one

individuals

Learning

to

by sample
to

likely

for

2476

education

of applied

findings

Sitlington

GENERALIZABILITY

Of

as the

research

influenced

some

used

list

and

tends

such

research

more

also

was

1990

disabilities

Sitlington

sampling

in basic

with

entire

sample

First

ma

participated

all

this

to the

individuals

studied

years

findings

participants

part of their special

random

than

incor

can

had
as

of these

plied

the

in

for

from

Frank

who had

Iowa

in

likelihood

equal

and

from

selected

students

partic

the

school

high

ob

be

which

of learning-disabled

randomly

ucation

gram

sam

those

analyses

were

1090

descriptions

and

analyses

completing

age

second

the

success

ter

obtain

the

generalized

Sitlington

to

sampling

school

Time

interest

particular

could

was

research

from

Sup

of potential

Because

an

has

sample

group

sample

be

sample

from

this

records

random

be

interest

study

names

selecting

example

the

39-

to

these

at

from

groups

More

from

to

Time

and

tested

data

35-

49-year-old

samples drawn

with

stable

we might want

Time

first

whether

random

of

ran

nonprobability

of

graduates

population

of

For

gen

as

random

simple

sample

sample can

population

pop

sample

in

in the

study

into
of the

are

greater

method such

population

university

representative

drawn

heing stud

the

members

in

effects

remains

from

ef

age

examines

45at

the

groups

age

between

the

samples

contrasted

with

to

hy randomly

selection

that

larger

imply

involved
In

the

random

ipants

the

methods

samples

representative

inclusion

tained

which

to

of the

result

sampling

1986

wanted

one

pose

as

who

representative

sampling

generate

biases

of
of

university

of

changes
simple

variable

example

difference

first

with

analysis

groups

age

or different

drawn

samples

For

and

group

from

data

groups

the

the

group was

The

on

the

time

examine whether

the

two
cross-

assumption
age

of the

strategy

between
over

chance

time-of-measurement

effects

time-sequential

difference

equal

generalize

study

the

Babbie

every

interest

time-sequential

with

cohort

assuming no
ied

in

avoids

is

characteristics

and

find

individuals

of

individuals

ensures

sampling

ability

Certain

to

59

of the

to

sample

of

probability

be done

be conducted

from two

data

Finally

are

fects

3.1

of

could

of

it

the

the

for

member

the

the

uniform

least

variable

analysis

Time

and

table

the

at

Use

the

ISSUES

validity

sample

applied

interest

cralizability

samples

contrasted

under

effects

or

changes

age

are

representative

dom

that

likely

of

cross-

only

In

be

more

of

the

population
to

much

cause

datal

require

changes

time-of-measurement
that

both

cohort

analysis

mean

can

minimum

use

are

ulation

external

The

beyond

requires

larger

findings

De

to

strategy

analyses

as

collections

could

cohort-sequential

per

passed

analyzing

begin

be

proposed

this

collections

one could
to

have

has

to

the

interdependent

findings

the

are

the

drawback

cannot

analysis

one

use

from

data

practical

collections
that

and

in

data

data

interval

before

sequential
earlier

that the

is

three
the

contrast

One

would

of

and

study

participated

cohort-

45- to 49-year-old

the

Time

and

age group

strategy

formed

long

Time

the

the

be performed

from

data

longitudinal
of

one

are

ings

studys

validity

in

used

pie

dif

for

external

single-cohort

could

variable

found

is

greater

example

that

the

change

age

by

midlife

analyses

influenced

have

results

provided

our

Time

the

the

be

not

consistent

If

For

design

have

effects

studied

AND METHODOLOGICAL

DESIGN

sample
is

most

include

is

the

often

strat

used

adequate

FOUNDATIONS

60

numbers

ables

that

the

to

tIe

of

individualafrom

in the

terest

are

under

problem

1956 by

numbers

equal

cohort

be

also

may

lion

which

one

recruit

groups

must

bers

of individuals

of

7090

methods

many

in

have

90-year-old
than

tween

the

as

those

adults

by

Yet the

be

science

bility

use

more

commonly

obtained
for

groups
and

by

from

e.g
consist

including

person

every

volunteers

haphazard

to

sampling

Ccneralizability

samples

of

volunteers

and

sentativeness
individuals
lion to

who

which

Minke

Haynes

potential

volunteer

we may want

may

their

for

bias

differ

to

simply

ples

by

of the

lack

gay

that
of

use

repre

Specifically

from

the

popula

studys

find-

homeless
and

lesbian

uals

in

sample
more

Exhibit 3 - 9

these

could

often

people
If

large

groups
then

all

be

can
taken

identified

the

as

who can

be

might be

be

accessed

particularly

1998

Rose

disabilities

number

frnm

those

of

Kalton
of

individ

random

identified

who

the

Factor

identified

individuals

con

Exam

1989

Rothblum

enough

by

who

2002 members

with

rep

begins

group

target

Pruchno

Patrick

and

referral

minorities

community

1989

Anderson

by

ran

Minke

known

also

that

include

for

identify

Anderson

of populations

samples

group

members

group

Woods-Brown

2002

Aaron

of

Kalton

sampling

members

elderly

2003

on

types

snowball

the

members

2003

populations

needed

sampling

the

indicate

Haynes

to

sampling

Repli

or in other

or

to recruit

frame

that

increase

findings

available

difficult

or

other
so

use

Snowball

few

identify

as

be

either

to

points

2001

Minke

or hard

or

necessar

samples

samples

can

studys

sampling

sampling

tacted and

or

studies

of

to generalize

of

then

known

is

2003

identifying

survey

in testing

because

and

researcher

to

responds

with

samples

to

the

can

sample

utational

rare

likely

because

Haynes

participate

for the

concern

dom

the

research

with

especially

existing

these

to

situations
of

of the

Research
is

directly

or

of

types

im

rather
of

types

or

is

it

not

Stanovich

other

generalizability

partly

of surveys

examined

environments

limitations

or

with

study

memory
1987

available

of

but

findings

studies

may have

Yet

does

participants

studys

situations

be

also

advertise

through

participate

is

should

churches

centers

that

variables

samples

from

The procedure

who

available

groups

of

studies

use

re

Additionally

types

1993

the

that

un

the

in their

al

et

for
to

of

Boatwright

mind
of

pursuits
opportunity

ability

1991

decline

Dodge
in

keep

invalidate

other

known

are

Such

Thus

or

nonproba

or

samples

volunteer

study

for that

groups

behavioral

samples

case

particular

use

to

samples can

most

pools

convenient

available

to

portant

the

West

conditions

cating

random

senior

or

are

who

who

various

volunteers

passersby

eligible

of individuals

courses

training

of

in the

health

al

prevention

about

Schleser

abilities

ily

samples

participant

training

so

inter

volunteer

the

for certain

or

Higher

requirements

et

volunteer

concerns

courses

nonrandom

local

example

not

even

or

and

Wagner

and

greater

the

al

et
to

scientific

who do

those

al 1993

who

to

time allowing

purpose

project

individuals

for taking

of the

true

soliciting

search

convenient

convenience

media or

in the

of

study

than

the

to strat

result

volunteer

or

Dodge

greater

sample

in

research

et

Wagner

1993

at

et

related

be important

job

types

example

university

ments

as

or

samples

group
are

These

the

in

Dodge

leisure

Janz

Hecht

to transportation

may be

participate

for

be

match

instance

may

that

As

samples

as

be compared

will

studies

social

of

stratification

groups

For

motivation

obtain

to

sampling

incomes

access

status

more

Clark

Crothaus

supporting

in

having
to

Omenn

Dodge

higher

site

testing

derstand

individuals

sizes

the

Rosnow

Patrick

educated

better

in

middle-aged

Rosenthal

1988

Thornquist

have

have

cioeconomic

study

inclusion

maintain

representation

remains

fact

difficult

the

group
to

it

other

and

self-selected
participate

he

to

likely

1993 Wagner

1991

LaCroix

est

more

1959

better

that

who

Individuals

are

Schork

range

the

spanned

be

num

equal

70-year-old

versus

that

population

in

various

or

age

self-improvement

this

to

that

as

used
of

obtain

that

unlikely

to college

maintain adequate
in

1999

population

sample

e.g

Baltes

equal

be

proportions

in the

returning

the

ify

Kiumb and

resulted

ensuring

altematively

hard-to-

or

ethnic

indjyiduals

could

in

certain

sample

is

rare

Schaie

99l

characteristics

Rodgers

l-Ierzog

from

Stratifica

the

studies

studies

1992

in

in situations
of

research

the

Liang

method

of

terms

into

groups

included

equal

them

1975

individuals

age groups

It

years
would

Rather

old-old

random

stratified

also

in

ings

in

enrolled

this

were

inclusion

ensure

e.g

used

With

Seat-

began

from

important

For example

minorities

sex

were

sizes

especially

related

of individuals

men and women

of

study and

the

and

cohort

by age

stratified

2004

organization

vari

The

1986

l996a

samples

maintenance

health

are

expects

Babbie

Schaie

random

taking

one

study

Study

Longitudinal

in

which

on

differ

in

of

subgroups

subgroups

or

related

various

These

population

are

but

svilling

.-

to participate

from

e.g

study

social

family

service

of

1998

found

time

staff

ously

that

but

or

may

proach

discuss

search

or

purposive

more

used

lects

certain

the

researcher

acteristics

or behaviors

selection

may

be

servation

Rabbit
the

increasing

or

of

study

measure

selected

based

judgment

be

is

best

to

for

terms

ability

produce

to

of

sample

the

use
not

are

that

not

subject

of college

generalize

results

its

is

Replication

generalizability

of

to

of

are

and

alumni

im

restrictions

may

limit

in

using
studys

to noncollege-educated

important

findings

to

from

Sharma

2003

of

have

studies
the

on

representa

Mof

Gottsclialk
of

Loss

in studies

waves

variables

e.g

that

use

participants
inter

longer

testing

ISSUES

or

of

abilities

inferred
other

by

Thus

it

by

that

is

constructs

dressed

that

of

change

in

consider

to

able

the

For

amounts

one

that

the

of

study

particularly

validity

also

in

the

and

an

terms

un

reliabil

uses
that

the

and

method

these

and

question

development
is

behav

scales

to assess

population

fits

construct

of

rating

adequate

learning

traits

participants

e.g

using

directly

of interest

ensure

to

is

for the

appropriate

measurement

of

the

means

has

constructs

personality

important

measurement

ity

these

observing

indirect
is

researchers

developmental

cannot be observed

about

intellectual

must be
ior

that

in studying

information

Rather

observed

sampling

example

of the

interested

are

tests

that

address

large

any
For

the

purposive

automatically

to

their results

wide

individuals

wishes

question

of

increased

Schaie

2004

other

impact

MEASUREMENT

Many

sam-

developed

study

aim

be

time between

of

vals

Cooney

1996a

Fitzgerald

Tang

Johnson

also

may

sample

the

remaining having

jobs

whereas
no

sample

those

that these
toward

may

with

the

The

samples does

sample

individuals

of

1998

fits

function

participants

one

the

of generalizing

random

Thus

of the

have found

for

purposive

newly

tiveness

elder

are

including

1988

had

attrition

reasons

2001 John

the

with

1986

Brant
that

lost

includ

participate

income and

1988

represen

Campbell

hias

Schaie

the

may be

participants

studies

to

more

to

re

began

indi

broaden

to

ob

views

individuals

purpose

be noted

studies

on

The

program

group

when

researcher

random

or

that

the

the

should

It

frames

on

of

at

se

char

may identi

used

1986

and

studied

Alternatively

as

tested

Babbie

ability

sampling
will

be

to

is

of

range

the

such

found

if

of

alter

less

it

for several

2003 Some

even

may

attrition

Aiwin

tend

of

attrition

study

Participants

or

class

Tobin

extended

of

conservative

specifically

characteristics

pies

ply

groups

may be

sampling

with

re

Even

of

if

rendering

refusal

attrition

Willis

is

influence

especially

education

With

the

display

being

adults

isolated

socially

individuals

Republican

target

of

which

2003

For example

skills

better

will

middle

sample

caregiver

by

we

targets

period

illness

relocation

refusal

types

or qualitative

interest

ly

or

should

generalizations

generalizability

previously

occasions

testing

death

sample

of

the

on

sample

of the

was

it

Tang

that

1-laynes

after

1986

social

who were

viduals

of

made

than

tative

of

sampling

they

characteristics

sani

demo

the

research

representative

ap
al 2002

et

specifically

with

with
the

Thus

or

be considered

also

son

cost-effective

sampling

because

individuals

e.g
organi

Rothblum

must

61

made

better

of testing

waves

ISSUES

replicated

applications

be

can

be

individuals
characteristics

that

longitudinal

sults

ing

rare populations

and

In

between

more

Simultane

service

target

Minke

al

et

methods

in descriptive

1986

Babbie

method

to

Com

required

formal

judgmental

commonly

this

1986

of nonprobability

type

is

Rabbit

lists

al 1998

et

most in

successful

most

Patrick

final

the

health
so

research

the

should

health

fortunate

and

between

this

Patrick

better
less

reported

any

identitirsg

sampling

groups

support

the

to

where

cost-effective

of

graphic

be

about

recruitment

multiple

wanted

receiving

techniques

highly

media mailing

be

not

list

the

to

jobs and
pIes

information

group

snowball

was

using

snowball
zations

of

one

if

research

e.g

recruitment

other

used

identified

counter

is

reveal

to

study

members

to

may

were

descriptive

used

is

under

process

pared

for

sampling

fluential

who

caregivers

helpor

formal

be

also

formally

agencies

An

Kalton

study

samples

studyfor example

of the

objective

the

in the

Snowball

random sample from

when

type

included

are

1989

derson

AND METHODOLOGICAL

DESIGN

RESEARCH

__________________

scale
is

ad

assessment

important concept
of

identiljiing

reli

change

establishing
studies

using
Validity

probability

2001

or

Studies

of people

with

nonprobability
of adult

greater

samples

development

income more

that

Stanovich
use

education

samples
better

tests
text

of

validity
the

Exhibit 3 - 10

is

perhaps most important

intended

purpose

of

the

test

in the

Thus

corn
test

62

FOUNDATIONS

could

validation

accumulating
iii

various

done

are

test

measure

particular
of

struct

proaches

the

to

contributes

toward
valid

is

This

interest

sharc

the

the

that

three

each

ap

which

of

toward

of information

that
are

could

have

How

est

it

well

of

reflections

tal

the
of

the

made based

on

the

reflection

rate

1990
lidity

and

obtained

items

validity of

either

proached
test

or

have

hypothesized

measure
analysis

ducted

considered

is

e.g
the

would

have

other

several

might be

the

that

fhetor

strong
of

that

is

that

are

validity

on

being
the

but weak

construct

were

been

can

of

tal

the

con
in

validated

be

the

sirability

may need
essary

to

to

on

the

The

methods

analysis

are

of

by

tests

of other

and

exploratory

discussed

in

traits

more

be

in

intent

of

factor

later section

during

the

is

an

issue

development

it

de

social

the

cases

in

e.g men

or

honesty

may be nec

items

in

resulting

validity

Criterion-related

validity

correlated

strongly

Two

the

is

predictor

Validity

with

main

be

can

measure

criterion

types

of

assessed

being

by

measure

criterion-related

how
i.e

validated

Suen
validity

Validity

have
validity

studies

e.g

the

va

of Ikee

in

on

1990

Content

development

Issues

of personality

In these

va

already

or attributes

the

Content

has

the

mea

Face

sensitive

biases

to

measure

factors

loadings

confirmatory

detail

to

con

measure

appears

1994

Criterion-Related
factors represented

validity of
scale

but

to
In

validity

during

may

certain

be avoided

face

person

related

is

considerations

measures

disguise

reduced

of

be appropri

not

measure

than

matter

where

bias

face
the

after

important

or

that

Bernstein

subject

health

pro

community

measure

intended

is

rather

Nunnally

lidity

the

whether
it

developed

stage

Suen

for assess
disease

is

content

with

determined

also

on

may

concept

validity

with

construct

is

spread

be

same

the

sure

is

be confused

content

to

items

of adults

study

concerned

Or

of
or

purpose

or context

patient

childhood

in

validity
not

which

approach

test

loadings

not

factor

Alternatively

is

lidity

other

do

measures could

the

ap

within

they

items

rnultitrait-inultimethod

expectation

tests

trast

construct

poor

from

of scores

which

the

Face

as

domains

or

measure

many domains

across

ate for the

the

particu

one

based

vary

or institutionalized

development

ality

for

appropriate

whether

by

validity

must be

test

representation

purpose

may

dwelling

valid

analysis

burden

are

con

analysis

with

the

find

particular

validity

items

possible

the

items

Thus

of

of the

good

another

gression

fhctor

test

for

or

construct

items

all

example

caregiver

should

factor

if

For

1990
ing

of

may

to

from

content

the

content
be

per
the

measures

forms

purpose

do

same

construct

items

but not

context

the

which

sample

of the

establish

help

in establishing
of

collection

to

the

in

the

on

loadings
to

items

the

intended

comparing

comparing
if

cases

alternate

Finally

sample

lar

by examining

issue

this

va

measure

are

that

Using

that

Suen

results

The

establishing

internally

example

strong

for

noted

externally

For

tcsts

construct

accu

same

of the

established

measures

method

in the

1994

considered

be

an

is

convergent

be

that

construct

good

others

to

of

so

interest

measures

can

other

1990

Suen

ity

with

Bernstein

same

the

possible

of

true

Construct

scorc

test

c6nstruct

other

measure

similarity with

measure

as

and

it

individual

similar results

to yield
with

Deter

subjective

in

good

represents

also

may

can

collected

which

in

one

with

In

entire

validity

1994

typically
of the

items samples

be expected

content

the

is

study

the

rcpresent

judgment

validation

of possible

be

to

measured

is

that

related

is

Nunnally

struct

one

the

need

interpretations

measure

that

one

main

its

of

idea

Specifically
is

that

and

information

the

validity captures

the

similar results

ing

construct

meaningful

on

of inter

construct

Bernstein

representativeness

based

set

pool

items

determines

Nunnally

this

foniiing

selected

that

items

potential

to assess the

items

measure

the

Validity

in research

conclusions

used

the

of

measure

The measures we employ

been

of possible

process

of

new mea

on

inclusion

for

sample

pool

would

Construct

selected

are
only

mining

con

the

describes

process

is

purpose

of

Items
sure

1999 con

establishing

section

types

that

establishing

assessment

validation

test

different

to

and

interrelated

measure

of

something

approaches

evidence

contributing

as

of

process

use

McDonald

1990

the

the

than

rather

of

all

continual

as

about

Suen

once

that

validity of
of

viewed

contexts

only

cluded

be

infonuation

that

phase

is

typically
of

considered

new

measure

been

when

the

dictor

measure

Exhibit 3 - 11

These

discussed
criterion
in

is

measured

types

differ

in

in relation

question With

predictive

terms

to the

of

pre

validity

.------

DESICN AND METHODOLOCICAL

RESEARCH

the

criterion

with

measured

is

concurrent

measures

cal

related

assessed

for

the

why

pecially

used

make

to

used

to

be
to

cally

if

the

as

cognitive

is

be

measure

test

same

it

to

current
implied

is

that

Or

than

1994

be

also
the

as

These

reduce

the

future

itself

test

not

the

measure

nally

on

may
the

good

attenuate

the

criterion

validity of

the

of

criterion

and

their correlation

the

measure

adult

learners

have

more

scores

on

measure

than

sample

and middle-aged
terion
the

as well

age-heterogeneous

assessment

limited

of

as older

young

adults
it

if

of

range

involving

lower than

may appear

validity

more

adults

measures

For example

older

adminis

test

or developmental

effects

result

can

effects

in

changes
which

administration

even

coefficient

Suen

change

the

though

1990

or Split-Half

Reliability

Fi

that restriction

or

two

factors

state

validity

predictive

predictor

of the

esti

other

study

research

to consider

important

either

in

participate

have

must

also

it is

range

to

ability

Nun

reliability

several

by

Alternate-Forms
as the

such

higher

intervals

in the

validity
predict

did

produce

Test-retest

reliability

inter

test

second

the

at

the

by

shorter

it

that

will

nature

of

types

scores

test

mind

in

influenced

longer

practice

measure
because

example

influenced

parallel

such

changes

For

it

administrations

test

reduce

trations

can

state

to assess

the

keep

be

can

estimates

mates can

to

63

reliability

of

administration

important

is

Bernstein

in the

Specifi

first

reliability

used

reliability

that

this

would

measure

to predict

used

may
select

to

validity

concurrent
are

or

of

between

vals

It

test-retest

administration

of the

test

estimate

this

second

form

parallel
the

nally

scores

study For

the

scores

used

good

scores

Nunnally
test

research

impairment

has
test

are

impairmeot

for

is

cognitive

where

certain

es

measure

that

criterion-related

good

measure

case

theoreti

becomes

person

ecample

of the

use

have

from

about

individuals

valid

need

For

indicate

or exclude

the

is

should

measure

measure

scores

decisions

1994

Bernstein

such

choices

called

is

reliability

the

Obviously

predictor

Good

of

procedure
of

validity

important when

to

and

be guided by some
predictor

This form

measure

predictor

to it

Criterion-related

be

the

criterion

simultaneously
will

measures

rationale

the

validity

are

of criterion

after

ISSUES

the

cri

would

have

in

sample

To

the

avoid

issues

another

ity

construct

two

However

the

be

difficult

created

1990

by

the

measure

the

this

of

par
be

two

to

halves

assess relia

from

items

process

may

truly

are

into

forms needed

number
in

mea

it

forms can

alternate
items

test

and

two forms

scale

the

two

prohibitive

the

the

halved

is

of

of constructing

parallel

Because

bility

cost

Instead

dividing

creates

this

forms

to

is

reliability

may be
that

to ensure

Suen

allel

and

reliabil

test-retest

assess

to

alternate

equivalent
effort

with

used

than one

rather

sures

associated

approach

the

total

estimates

reliability

Reliability

be

will

The

well the
level

measure

of

the

test

i.e

individual

are

over

time

needed
liability

furms

scale

stability

The

next

which

of

parallel
section

differ

measure

are

assessed

measure

terms

defined

how

in
are
of

that

how

to

halves
various

selected
single

test

and

on

pending

reliability

participants

take

same

the

correlation

squared

procedure

items

produce

on

how

be

can

constructed

estimates

item

split

to

the

items

may then

is

two
are

from
vary

de

was performed

re

Cronbachs Alpha

created
the

of

descriptions
or

test-retest

split-half

and

reliability

have

reliability

demon

Reliability

measure
the

the

compli

potential

split-half

test

halves

reliability

how

the

the

formula

Nunnally

e.g

One

Depending

test

parallel

strated
to

with

split

see

1990

and

measure

total

Spearman-Brown

details

remains

exactly

the

the

Suen

1994

alternate-forms

One way

with

computational

As

Test-Retest

relative

corrected

equivalent

assess

types
the

for

cation

reliability

measure

several
of

or

of

reliability

of

presents

in

of

researchers

must be

Bernstein

Bernstein

definition

forms

how

ability

measurements

when
To

as

alternatively

Nunnally

concern

of

reliability

of

of

is

defined

unobserved

stable

second

the

Typically

be

true

being

how

stability

terms

can

the

reflects

reflect

score

1994

measure

may

reliability

the

of

reliability

attenuated

test

of the

is

to

have

twice

two

sets

and
of

group
then
test

of

find

scores

creating

simple

seem
Id

more
of

Exhibit 3 - 12

forms

Cronbachs

process

like

forms

parallel

extreme

test

With

of

alpha

approach
this

test

may

takes
to

approach

creating

each

be

not

what

may
paral-

item

is

64

FOUNDATIONS

considered
item

pair

much

that
the

when

the

considered

is

cient

provides

of

item-pair

all

advent

number

Institute

of

computers

scale

coeffi

zero

for the

be

can

SM

as

SAS

as

has

With

included

viduals

in

on

had been

American

skated

that

no

of

measure

or

to

other

any

test

ual

test

of correct

norms

is

be used

participants

Suen

Scales

Variables

be

can

mnemonic
include

two

In

visual

individuals
ues

on

not

decisions

on

one

For

in

measurement

the

when

the

correctly

scales

eg

an

the

ordinal

some continuum

ested

type

task

or

ordered

on

variable

For

age groups
categories

the

example
the
yet

of

scores

change

is

measurements

than

loci

strated that the

values

orders

were

based

the

val

do

between
study

may be

may be unevenly

1956

generated
vidual

This

However

existed

unlikely

when
over

change
as

having

the
three

many more

Exhibit 3 - 13

--

individual

time
score
test

or

are
are

more

analysis

time

options

be

to

single

to

less

occasion

for

just

e.g
demon

of

change

research

studies

amount

the

estimates

sometimes

points
than

reliable

was

reliability

1988

of the

indi

in

better

oc

has

no

that

amount

most

two

inter

conclusion

this

differences

much

that

1988 1995

in the

Rogosa

itself

any

allowed

are

critique

an asumption

situation

may

at

change

from which
on

items

able

typically

tend

they

on

of

collected

are

of

Rogosa

an

change
the

that

by defin

being

as

well-known

differences
is

Thus

able

amount

taken

of

exam

this

Change

construct

construct

this

in

vari

scale

ability

researchers

the

examining

on

of

categories

in

in

of

the

interpretation

memory

same

Lord

distances

age groups

use

the

on

time points

rhymes

rank

differences

equal

might
verbal

Although

ranks

data

more

curred

method

two

or

of

scale

ordered

about

abil

memory
Thus

number

meaningful

conceptually

conclusions

this

does

this

individual

ratio scale
the

as

However

recalled

the

variable

as

Although

10

interval

an

represents

interest

from an

test
of

of

exam-

For

generated

of 10

the

to say

much

as

10

as the

good

operationalized
of

of

score

true that

is

twice

of

as

or other

has

ability

as

with

recall
It

be

as

When

of

recall

are

of

ratio

the

and

-on

twice

Person

as

score

recall

in

and

accurate

as scores

memory

is

variable

be

not

categories

categories

ordered

for delayed

could

individ

20

individual

ing

and

acronyms

considered

able

the

ple

four

example

memory

represent

spaced

of

scale

that

of

the

as great

Measurement

acrostics

three

ity

that
of

be

delayed

twice

are

score

the

of

comparing

with

may

about

unordered

trait

necessarily

people

not imply

and

interval

ordinal

any

20

of

scores

not

were

scores

of

Measurement

mnemonics

along

an

between

or

variable

contrast

gender

that

unordered

eg

mnemonics
and

of

used

strategy

the

an

these

administration

make

represent

construct

for

1990

measured

variables

that

Questions

important

nominal ordinal

measurement

particular

mke

to

assume

ethnic

appropriateness

particularly

pie

are

weight

learning

of 20

score

not

statements

of

person with

of

are

they

much

as

may

it

with
as

demon-

differences

group

person
construct

be

to

that

development

and

sample

relate

of

the

between

also

pects

with

measure

use

construct

may

use

being stud

could

it

this

Measure

will

scores

Nominal

on

important
scores

to

differences

appropriateness

influence
use

unless

reasonable

expected

to

validated

Caucasian

sample

groups would be

been

want

not

on

validated

be considered

to

similar to that

wuld

one

of indi

type

appropriate

has

it

population

ied For example

African

when

sample

particular
standardized

most

is

the

for

needs

also

study

measure

Specifically

that

measure

of

twice

of

conducted

as

measurements

weighs

With measurements

valid

score

absence

studies

height

ra

ratio

variables

in

for

and

point

words

learning

e.g

physical

Person

as

zero

Such

and

sciences

Measure

of

are
appropriateness

meaningful

as

depend

interval

complete

measured

or

represent

the

on

in other

point

represents

being

physical

such

zero

development

length

The

of the

common

the

Appropriateness

definition

true

adult

individuals

between

ratio scale

measurement
scale

difference

variable

on

between

of

on

in the

is

on

the

average

and

1990
such

programs

scales

tio

values

The

variable

of reli

1999

statistical

differences

alpha

estimate

between

equal
ent

ratio scales

and

intrial

distances

forms

other

of Cronbachs

Suen

large

is

both

For
equal

became

items

test

each

of

Bernstein

process

to the

preferable

correlations

by

and

of high-speed

conservative

computed

correlations

intensive

The computation

ability

the

Nunnally

this

with

easier

especially

and

examined

noted

1994

and

test

separate
are

1995
of

reli

Although

data

two

as

of
for

the

provides

change

RESEARCH

should

score

reliable
In
ate

or

not be

poor

addition

to

change

individual

change

We

change

over

other
the

ual

time due

scores

SEM

defined

2002

those

by

cant

as

be

for

scores

would be

greater

and

classified

as

considering

of

the

tests

issue

also

be

tests

have changed

resent

considered

the

Thus

the

time

for

time

sue

factor
in

factor

firmatory

measurement

of

Another

from

than

tors

factor

lated
more

must

mated

and

among

the

return

longer

of

more

1991

in the

section

method

the

analysis

This

con

on

used

is

to

test

invariance

In

contrast

factor

Bollen

factor

ray

rovide
Lu

the

often

is

basic

determine
to

from

performed

analysis

use
texts

With

oveniew

specific

on

as

which

whether

the

there

be

of the

this

statistical

confirma
shortly

information

would

details

is

unknown

are
to

described

method

that

new measure

is

precursor

will

factor analysis

EFA

method

is

dimensions

underlying

be

to

as

uncor

solution

rotation

varirnax

Suen

is

1990
fac

the

among

an

of

type

corre

oblique

an

If

rotation

promax

solution

is

oblique

factor

esti

is

low one may

are

solution

Factor

Analysis

the

at

Corsuch

provides

measure

has

structure

see

factor

the

of

fit

that
this

framework

researcher

structure
benefit

indicators

of

to the

and

use

can

data

to

that

CFA frame

of the

the

CFA

details

modeling

Another

multiple

Sorbom

Jareskog

computational

equation

collected

that

is

confirmatory

when

1983

and

statistics

fit

analysis

used

hypottesixed

least

structural

been

factor
is

same

underlying

Analysis

EFA

when

in

factors

factor

rotation

intercorrelations

statistical

construct

which

such

CFA

1989

for

work

appropriate

scores

The

the

is

expects

correlation

use

to

analysis

known or

has

analysis

EFA

one

hypothesized

orthogonal

assess directly

SCALE DEVELOPMENT

articularly

where

to exploratory

thus

factor

in

of

or set

be used

or

factor

uses

Exploratory

meaningful

where

orthogonal

appropriate

1993

Exploratory Factor

in

is

comparisons

Horn

depth

addi

the

rep

in

invariance

interpretable

domains

also

may

plots

constructs

underlying

longitudinal

75%

least

at

where

made

solution

to the

Opti

plots

for

uncorrelated

can

in cases

bat

of

explained

point

the

result

measure

expected

is

that

proper number

Scree

be

orthogonal

estimated

is

in
the

of scree

use

cases

to
En

most popular

However

classified

less

that

decision

the

related

and

variance

1983

not

the

made by

explained

obtain

to

the

fac

nature

guidelines

account

where

does

factors
in variance

solution

the
will

Corsuch

determine

to

of

crease

and

solution

variance

be used

the

percentage

factor

extracted

signifi

invariance

may no

they

or

making

discussed

for

be

time

over

of factors

of

Confirmatory

demonstration

across

defined

example

scores

relationships

factors

same

important

is

over

the

and

decline

temporal

set

scores

value

would
change

change

of

If

For

had

al

the

that

tion

range
et

stable

as

had

significant

the

this

common

the

by

values

Ball

as

than

When

of

Two

are

in

are

determining

for

this

common

structure

themselves

65

reduce

to
of

exploratory

is

factor

factors

of the

factors

of the

change

measure

used

best

researcher

mally

as having

improvement

change

significant

tery

of

of

set

ISSUES

intent

procedure

the

used

often

individ

allows

improvement

in

change

then

about

the

smaller

into

this

terpretation

are

witlv

analyzed

of scores

Because

tors

details

e.g

classified

classified

or significant

SEM

then

to

value

be

to

the

at

range

change

amount

significant

SEM

decline

the

if

This method

randomly

be

can

outside

those

only

reliable

only

with

the

and

Use

are

set

individual

will

whereas

change

the

1996a

Schaie

that varied

to

having

as

variables

decisions

of reliable

computational

compared

cre

scores

SEM
change

for

1979

are

formed by

meaningful

meaningful

to

change

fluctuations

measurement

Dudek

see

Change

are

of

error

amount

individuals

random

to

values

examine

to

of the

some

show

reliable

level

subtracting

that

will

two

also possible

is

level in terms

individuals

define

to

it

know

standard

un

an

as

automatically

largcr

simply

score

the

at

dismissed

choice

AND METH000LOCICAL

DESICN

can

We

necessary

be appropri
be

among

by

Schaie

best

that

heifer

of

estimate

construct

observed

1991

in

which

the

by
is

than

of

six

indicated

be

pro

example

determined
Seattle

an indi

could
For

measure

Willis

represented

each

or factors

form

used

battery

was

that

the

Longitudinal

cognitive

by

at

domains
least

three

measures

CFA

of

and

Dutta

whether
set

on

single

cognitive

Study

to

true score

vided

obtained

e.g Corsuch 1983

interrelationships

viduals

used

are

and

can
the

the

Exhibit 3 - 14

be

useful

relationships

latent

constructs

method
between
they

for

determining

observed

represent

variables

remain

FOUNDATIONS

66

invariant

across

Joreskog

1979

been

Baltes

see

regression

the

ences

measured

lion

factorial

of

that

showing
and

with

latent

single

invariance

the

between

relations

remain

constructs

troduction

of

lationships

Schaie

also

is

interventions

differ

variables

minimum

only

requires

have

same

the

The

1993

variance

or

invariance

the

in

out being

re

samples

The

samples

where

Schulenberg

weak

level

the

factor

loadings
the

of

but

levels

of invariance

complex
strate

studies

partial

may

it

for

in

in the

many

demon
Shavel

invariance

of the

words by an
tant

if

factors of

participants
of

whether

in

activities

For

does

increased

ability

cal

from

trained

of

2002

cobson

and

2002

Clark

clinical

significance

an

ona

of

arises

followed

over

analysis

metric

in

variable

more
clini

result

Karlawish

and

Ja

estimates

of

Atkins
on

method As Tucker

to estimate

of

interest

CFA

some uew

within

has

Study
test

intelligence

added
been
to

to

used

examine

recently

The

relations

battery

established

battery

Schaie

Willis

the

the

are

tive

of

neu

the
of

amount
the

to

gression

been

Caskie

the

statistic

of

R2

by

addition

size

variance

statistic

value

the

as

is

analysis
of

each

been

has

In

the
the

several

done
the

change

block

can

of

these

about

amount

of linear

R2

also

that

re

of variance
in the

or

re

have

of predictors

hierarchical
in

rela

explained

context

blocks

in

It

variance

Both

included

of predictors

If

used

of

analysis

information

that

describes

set

linear

typically

is

be used

can

presents
For

statistics

provide

variance

equation

regression

Exhibit 3 - 15

12

entered

psycho

used

for

section

this

exist

R2

the

effect

total

the

explained
gression

size

commonly

Cohen 1983

Cohen

fre

Seattle

more
analysis

regression

measures

extension

in the

to the

of

ANOVA

new measures and

study

two

vari

samples

of effect

Many measures

known

situation

because

the

the

permits

observed

This

space
studies

regression

previously

long time periods
often

its

fac

use

to

Size

imple

to

appropriate

However

aging

are

ropsychological

not

is

location
of

method

Longitudinal

to

of the

confirmatory

procedure

this

extension

construct

latent

quently

constructs

of

variable

the

or variables

factor

use

it

latent

observed

estimation
able

the

demonstrated

tor scores

on

is

Dwyer 1937

the

1971

an

This

11

of

application

important

analysis

ment

show

effect

an

Effect
factor

into

list

abilities

indicator

information
of

impor

everyday

grocery

research

of

list

be

translated

McGlinchey

provide

life

recall

adults

psychometric
useful

on

training

individual

short

be

can

abilities

focus

to

and

has

the

sta

assess

of daily

older

the

recall

to

meaningfulness

and

the

example

way

would

it

increase

this

to

has

to

ability

performance

signifi
not

is

memory

words

five

it

performance

aspects

improved

applied

of

practical

average

assess

to

level

the

For example

activities
increased

study

that

may be

significance

small

practical

so small

is

Urdan 2001 One

clinical

observed

everyday

transfer

Byrne

subset

only

error

to

longitudinal

size

sample

change

stricter

in

meet

possible

whether

may have

finding

large

in very

true

time

across

Another

unique

to

pro
which

Because

invariance

where

1989

be demonstrated

interest

be

only

further

the

time

difficult

measurement

Muthen

son

be

in

factorial

invariance

of

across

can

Weak

Meredith

equality

Mered

unstandardized

of the

or

practical

been

metric

is

invariance

factorial

factors

the

in

particularly

be

also

with

significant

statistically

can

reverse

significant

tistically

be

meaningful

practically

loadings

1983

invariance

time

intercepts

of the

nonzero

Mason

of

strong

requires

and

variances

and

equality

across

posed

invariance

indicators

zero

factorial

requires

additionally

the

of

invari

longitudinal

Significance

can

findings

such

configural

McArdle

next

level

of

tjut

pattern

Horn

lime

across

of

requirement

demonstration

the

is

which

can

Research

affect

1987

ith

Clinical

in

cance

ance

1975

OF RESULTS

MEANINGFULNESS

following

might

Be

of

in Schaie

THE

EVALUATING

con

important

Hertzog

Willis

Test

the

in

age

observed

stable

that

Personality

to the

Parham

Schaie

The demonstra

markers

NEO

of

McCrae

Rigidity

the

1992

relations

2004

al

restrictions

and

age changes

the

Costa

Inventory

et

of

construct

latent

significant

impose
of

interpretability

the

and

2004

havioral

trajecto

Shifts

1973
on

has

quantitative

underlying

1970

variables

would

found

if

struct

observed

that

developmental

in the

changes

time

across

invariance

assume

in

Nesselroade
of

factorial

one

of differences

truly reflect

ries

can

or

groups

when

Only

demonstrated

comparisons

on

multiple

staged

occurs

be examined

with

For

DESIGN AND METHODOLOGICAL

RESEARCH

ANOVA

i2

model

the

be

were

assessed

and

group

age

For

in

on

group

impact
the

if

ANOVA

the

in

each

of

ability

this

to

study

an

true

null

false

noted

that

how

one

size

value

to

of

ability

the

Cohen

article

and

George

the

that

the

dependent

is

his

in

power

It

and

exists

these

particularly

1986

Babbie

hypothesis

intersection
sure

that

or

hypothesized

The

practice

CA

in

the

social

of

Helmers

was

controlled
Association

popu

research

with

trial

Lev
Effects

older

Journal

1980 In

Stewart

home

the

of

some

research

ological

important

posing

and

adult

learning
of

ments

and

and

have

sources

that

of interest

1968

data

gitudinal

in

changes

are

data

of

many

herein

lyzing

should

sociated

when
The

with

be

the

sure

use

are

need

sures

is

should

for

valid
in

implicit

address

include

con

of

Eds

are

but

of

the

that
biases

Byrne

and

and mean

re

about

Finally

as

research

to

significance

of

the

work The

multi-

\V

Nesselroade

Reese

Met

psychology

New

York

hod

Academic

Structural

with

equations

latent

York Wiley

the

1989

Muthen
of

equivalence

The

structures

invariance

Campbell
and

mea

factor
of

issue

Psychological

covariance

partial

measure

456

lOS

Bulletin

1963

Stanley
quasi-experimental
In

teaching
search on

researchers

their study

develop
on

466

research

appropriate

The

differences

Shavelson
for

Testing

samples

reporting

1973

individual

.1

1989

teaching

designs

Cage

pp

17

Ed
1246

Experimeotal

for

research

Handbook
Skokie

of

in

re

IL Rand

McNally
findings

determine

Elder

the
against

practical

ana

Psychology

Drelopmenta.l

developmental

New

variables

difficulty

implies

potential

and

research

the

for

change

generational

pp 219251

issues

Bollen

nonrepresentative

information

readers

Multivariate

sequences

Press

Ion

age-related

the

but

In

Life.span

ological

Caspi
allow

and

of

analysis

measures

pie

samples ensure

samples

reliable
all

1970

Nesselroade
mental

type

differences

sufficient

to address
of

se

effects

the
Baltes

types

results

designing conducting

cross-sectional
generation

cross-sectional

note

Develop

145171

11

ontogenetic

ment
searchers

and

methodological

When

of study

Random

random

true

obtaining

in

choice

to

studys

and
and

Nesselroade
longitudinal

of

Behavioral

of

of age

study

Human Development

treat

for the

the

required

of

generalizability

found

drawn When age

construct

area

Journal

Longitudinal

the

in

Baltes

pro

operant-observational

162168

implications

cross-sectional

the

detailed

referenced

prolect

can be

in

More

be

can

topics
other

important

clusions

research

development

research

designing

new

when

considered

behaviors

self-care

489500

quences

of

method

and

design

be

to

conducting

these

textbooks

need

that

topics

an overview

in

An

residents

International

Ameri

the

of

of

adults

288 22712281

and dependence

nursing

4th

Jobe

Burgess

ment

to provide

51

Family

2002

al

et

interventions

training

Baltes
of this chapter

public
female

Wadsworth

study

aim

the

Acade

male and

and

randomized

adequate

CONCLUSION

it

children
of

Marsiske

cognitive

in

of

CA

differences

The

and

aging

fliego

of Marriage

Berch
eck

of adult

expected

Binstock
of

for

cares

can Medical

be

to

learning

detect

to

as the

for

important

hu

of

study

In

San

ap

Quantitative

observation

Journal

ed. Belmont

of

type

caretakers

dependence

will

strengthen

in the

Handbook

1988 Who
Naturalistic

Baltes
lation

generate

981990

ef

large
of

meaning

the

on
is

new areas such

development
power

of

solid

Press

places

estab

previously

pp 2243

Ainato

researchers

for

medium and

small

notes

used

Eds

sciences

correct

Ball

in

studies

only

2001
methods

Longitudinal

man development and aging

Co
social

chooses
be

reviews

for

being conducted

studies

can

that

Campbell

proaches

long-term

hypothesis

confusion

of

point

conventions

designations
test

learning

findings

for

mic

often

size

fect

will

field

burgeoning

Alwin
the

i.e

difference

effect
is

his

lished

research

research

and

consideration

describes

rejecting

1992

analysis

development
robust

new

into

and

proportion

be computed

important

Power

of

population

In

also

is

detect

probability

hen

elements

key

References

size

calculations

power

these

adult

and

compared
Effect

of

of

effects

reasoning

both could

to

included

effect

relative

example

two-way

attributed

each

the

training

examined

of variance

for

and

separately

can

effect

calculated

is

67

ISSUES

incorporation

the

children

Exhibit 3 - 16

1987

Bern
world

Life-course

Developmental

Moving

of

explosive

patterns

Psychology

23 308313

68

FOUNDATIONS

Center

Human

for

handbook

Resource

Human

of personality

Psy

for

Cohen

The

111
power

1992

primer

Cohen

1983

Cook

Applied
the

for

analysis

NJ Lawrence

Hillsdale

multiple

behavioral

Design

Rand

analysis

and

research

longitudinal

Costa

McCrae

Jr

NEO

of

1992

women

the

for

national

sample

version

and

studies

of

Bar

in

and

recall

York

of

Essential

pp 282295

aging

Factors

heart

self-management

error

on

continuing

Aging

of

data

longitudinal

longitudinal

designs

within

of

at

Duncan
curve

growth

of

loadings

of other

ings

of

analysis

biased

from

Human

press

Eds

Religious

being

of

in

panel

Income

33

25

and

Schaie
influences
elderly

data

The

Dynamics

life

the

health

Jour

course

pp 246282

New

British

at

in

the

of structural
investiga

Eds

Baltes
the

pp 303351

an

Family

Columbia
estimation

developmental

in

esti

stud

test

Council on

National

research

study

of behavior

New

York

and

Academic

guide

Eds
730
the

York

into

Exhibit 3 - 17

1989

Special

methods

research

Amityville

NY

Aging

disease

23 10431049

rare

Hcrzog
for

gerontology

pp

2002

of transforming

Alzheimers

Sampling

Baywood

Clark
challenges

users

Software

Lawton

In

Karlawish

well

of

Scientific

Chicago

Anderson

Kalton

Springer

1993 LISREL

Sorbom

reference

per

Booth
and

presented

Statistical

Are

panel

empirical

Nesselroade

populations

Krause
on

In

long-term

Press

1299

in

of the

longitudinal

Joreskog

Mcador

health

South

invariance

An

study

Vancouver

in

is

look

scientists

Paper

attrition

panel

from

development

load

1998 An

Flint

Religion

the

of

Study

In

spective

attrition

Associa

November

2003

Longitudinal

factor

factor

Best

1983 When

of factor

processes

1979

models

173178

Moffitt

Resources

Hays

George

known

the

in

Eds

179188

meeting

tions
the

of

pp 114125

practical

concept

by

Relations

Erlbaum
of

psy

factorial

Psychological

Mason

of family

Joreskog

and appli

issues

Psychometrika

sample

Panel

Michigan

in

tests

ethereal

twenty-year

variable

latent

determination

Gottschalk

Fitzgerald

of

test

given

Concepts
Lawrence

1937 The

Dwyer

sal

modeling

to

of

change

oJ

invariant

nnt

Tang

ies

Li

introduction

MahwahN

cations

the

mates

accelerated

Stryckcr

1999 An

in

Journal

change

Horn

American

McArdle

nual

Alpert

analysis

DC

ern Psychologist

236248
Duncan

the

for

Horn

Methods

Psychological

of

markers

hon

1996

Hops

assess

73

assimilation

rating

Collins

Johnson

Duncan

Analysis

In

Washington

Psychological

86 335337

Bulletin

Duncan

of

Psychology

50 189198
Comments on issues

1991

methods
older

misinterpretation

measurement

conditions

Psychology

invariance

of

on

based

Horn

project

Research

involvement

influencing

standard

New

of

of

Counseling

psy

Liang

Nonparticipation

disease

15 220237
Dudek
1979 The
the

the

York

Jan
1993

in

New

In

elderly

Greenberg

method

variance

Clark

Schork
adults

on

and

Educational

of

chotherapy

Press

University

Dodge

Lawton

pa pen

1981

in the

Stiles

2003

differ

Age

In

recognition

Eds

Saithouse
chology

1998

of

573586
Honos-Webb

McDowd

re

Baltes

performance

variability

Journal

and

Journal

Age

surveys

43 S200S205

Willis

ment

extra-

and Aging

Psychology

1988
sample

Sciences

intelligence

traindividual

in

of Personal

Journal

176186

53

interview

to

ag

change

Personality

midlife

Psychology

Viidik

and

health

Serdi

1987
to

of

condi

41 87107

status

Paris

study

arm

long

Steen

Functional

Social

Hofland

in

personality

144149
Craik
ences

rates

Hills-

social

Demography

Rodgers

Fluid

neuroticism

in

Stability

openness

Social

sponse

Larson

sectional

and

ity

edj

early-life

Schroll

college

Herzog

Zonderman

Lebowitz
Cross

from

2nd

2004 The
of

mortality

Moane

Helson

Gerontology

Manual

Cultural

Psychology

analysis

influence

NORA

Gerontology

Jr McCrae

bano

1986

on cog
attrition

subject

exclusively

Erlbaum

Berg

ing The

TX WAR

Odessa

Costa

and

Journal

It It

mens

on

Heikkinen

1988

43 P12P 17

Sciences

Psychological

The

Eds 1997

prior functioning

variables

personality

set

field

Willis

between

relationship

tions

Quasi-

for

Minority

Corman

Hayward

McNally

Schaie

nitive

in

IL

issues

be

to

Black

compari

he were

wishes

Factor

Lawrence

childhood

1979

and

Chicago

tings

sciences

refuses

who

Ethnic

1983

dale

regres

Erlbaum

Campbell
experiments

and

Gorsuch

Cohen

who

Jacqueline

and Adulphus

people

182196

Bul

Psychological

2001

Throne
black/white

biracial

Diversity

1551 59

sion/correlation

of

black

183204

Review

112

letin

architecture

in

identity
son

2004

chological

Cohn

Gillem

Re

Center

Research

source

Cervone

Cooney
The

NLS

1992

Research

OH

Columbus

Addressing

laboratory

advances

treatments

Neurobiology

RESEARCH

Klumb

1999

Baltes

and

old

very

tive

Berliners

activities

functions

as

Social

Gerontology

Lachman

of

and

adults

old

consump

resources

1984

intellectual

for

elderly

of

use

and

Journal

of

54B S271S278

Sciences

Jelalian

attributions

Time

Productive

and

of

in

young

39

Gerontology

5775 82
Lachman
differences

age

causal

in

physical

and

Aging

1271

social

York

attributions

performance

for

cognitive

and

Psychology

The

seasons

mans

of

New

life

McArdle

McDonald

1999

McGlinchey
Clinical

and

use

useful

and

analysis

factorial

ing

one

Therapy

pp 6995

ed. New

3rd

D.C

SmithA

recall

by

Psychological

and

participants

1988

search

In

recruitment

of

the

strategies

Schaie

Campbell

Eds

pp

research

aging

re

longitudinal

17

New

1209

York

1995

366

Mahwah

subject

and

analysis

of

change

pp

volunteer

hear

and

about

of
53

the

39

the

study

Or

how

women

of diverse

attainment

bisexual

educational

2002 How

for

Gay

and

Lesbian

to

ages

rcscareli

Medical

reach

les

ethnicity
pmujects
Associa

Willis

2005

Extending

into

primary

the

neu

men
and

Neuropsychology

Willis

1991

The

rigidity-flexibility

and

cognitive

adulthood

in

Psychology

and

2001

Longitudinal

re

Aging

1383

Schaie

Hofer
in

research

aging

Eds

I-Iondbook

Schaic

ing

of

Birren

the

Ck

Palo

Alto

of

Psychological

The

age

changes

tween
chology

Exhibit 3 - 18

in

the

and

Cognition

cross-

cognitive

behav

2004

Caskie
Study

Longitudinal

personality

and

for

Consult

70 67 1680

Bulletin

Willis
Seattle

CA

1968

Strother

Schaie

pp

aging

Press

Psychnlngists

study

of

Press

1975 Manual

Rigidity

stud
Schaie

psychology

Acadensic

Parham

sequential
ior

In

San Diego

Sehaie
Aaron

Hand
science

Blaekwell

Aging

between

Test of Behavioral

The

Ed

developmental

assessments

Outta

lationship

377

Erlbaum

1975

Factor

loumal
tion

The

NJ Lawrence

about

18 384390
Methodological

cvell

space

sequential
Description

12 245277

Cognition

questions

New York Wiley

Rothbluoi

you

Myths

supplensental

Rosnow

Rosenthal

bian

plus

Ed

Gottman

In

did

Myths and methods

research

On

Teti
in

Ten

ability

Study

In

MA

MaIden

ropsychological
tal

on cog

2004

Caskie

ies

longitudinal

139

pp

influences

research

methods

of aging

Longitudinal

Development

research

gerontol

Springer

1975

developmental

aging

Schaie

37

Springer
Rogosa

in

York

Press

University

of research

abilities

Methodological

Seattle

Caskie

Schaie
about

Rawlings

in

comparison

five

Myths

Meredith
issues

1998 Re

New York

Human

in

Encyclopedia

Developmental

Kaszniak

38 295302

Gerontologist

Rogosa

of

of

45 P52P57

Rose

effectiveness

integra
Journal

Ed

Baltes
in

strategies

issues

adults

methods

Research

Oxford

Schaie

book

of contextual

Sciences

York

or explanation

Puglisi

older

New

adult

in

New

Press

81281

2004

de

revisited

Erlbaum

Study

development The

Schaie

Psychometric

York McGraw-Hill

Effects

research

cruiting

Erlbaum

Morrell

Pruchno

Patrick

psychology

1994

1990

on

Gerontology

costs

Lawrence

of pictures

Dudley
tion

NJ

Understand

counseling

Bernstein

theory
Park

and

clinical

Mahwah

Nunnally

Eds

Hersen

in

issues

Sampling

issues

Longitudinal

Maddox

cd pp

isitive

2003

Theoretical

Life-span

NJ Lawrence

University

In

revisited

designs

Eds

development

ttle

re

Springer-Verlag

Intellectual

1996b

Schaie

Schaie

Psychometrika

Sea

Cambridge

2nd

factor

invariance

invariance

Thomas
research

New York

Hillsdale

1996a

in

development
development

Developmental

psychology

The

Howe

cognitive

Reese

studies

in

In

Cognitive

Cohen

pp

aging

threats

validity

in

1994

of

Reinhold

Eds

pp 241272

search

Schaie

the

in

Schaie

psychology

development

Progress

of

study

designs

Nostrand

Internal

cognitive

Brainerd

hood

Which

Behavior

they

Measurement

58 525543
Minke
Haynes
In

1988

Schaie

ogy

1993

the

of

of

Bulletin

Birren

New York Van

Schaie

methods

are

In

pp 4564

treatment

unified

33 529550
Meredith

Handbook

In

Experimental

Jacobson

significance

how

us

Erlbaum

Atkins

2002

Eds

the

for

the

in

Journal

Psychological

Quasi-experimental

aging

velopmental

Test theory

NJ Lawrence

Mahwah

models

data

145166

18

Research

Aging

6491107
1977

Schaie

methods
aging

model

problems

developmental

of

69

version

guide

of

aspects

14208215
1965
general

adulthood

in

Modeling

cross-sectional

structural

growth

421437

16

1992

and

longitudinal

latent

Schaie

users

Cross-sectional

psychological

Gerontology

Educa

growth

Measurement

Hamagami

complete

of

measurement

SAS/STAT

ISSUES

Institute

1959
of

of adult

Psychological

SAS

Schaie

3958

1956 The
and

tional

NC

Gary

psychology

Knopf

Lord

ing

Adulthood

32

1978

Levinson

to

1986

McArthur

1999

Institute

study

Self-efficacy

performance

Journal

SAS

AND METHODOLOGICAL

DESIGN

cognition

11

Relationsisip

Aging

3043 24

be

Neuropsy

70

FOUNDATIONS

Schaie

Willis

1987

Hertzog

Effects

mary mental

Schnlenberg

cognitive

structure

ability

training

upon
and

Psychology

Ag

wonien

of

older

initial

adults

ory training

SharmaS

entry

International

of

Shidy

Brant
in

Aging

ofAging

Journal

24 5566

Development

attrition

niem

in

the

Experimental

Factors

their

with

terly

Stake
sand

disabilities

adult

into

1397151
1995 The
Oaks

CA

chology

6th

life

Learning

adolescents
crossing

the

Quar

Disability

art

of case

study

research

gist

senior

Principles

Erlbim

think

Allyn
of

test

straight

about

psy

theories

Hillsdale

health

plain

associated

promotion

estimates

Mah

English

LaCroix
with

participation
Gerontolo

program

and

with

Eds

pp 78108

the

San Diego

homeless
62

ill

elderly

Handbook

2002
mentally

lntensational

issues

Methodological

Woods-Brown

Exhibit 3 - 19

score

427436

Hecht

research

Schaie

less

ways

Erlbaum

Factors

2001

Willis

homeless

Bacon

in

New

Springer-Verlag

31 598-602

aging
to

behavior

factor

36

Statistics

ef

593597

31

York

of

Crothaus

IL

Thou

Sage

Relations

2001

in

New

and

1991 Hu

criminal

Psychometrika

NJ Lawrence

intervention

ed. Boston

1990
NJ Lawrence

Are

successfully

2001 How

Stanovich

Suen

1990

Frank

bridge

use

1991

learning

and

knowledge

tirdan

329340
Sitlington

retinol

Farrington

1971

Tucker

Wagner

Gerontology

and

CARET Gerontologist

of advancing

wah

Longitudinal

men

Ohlin

to

L.Jl986

Baltimore

older

the beta-carotene

to

man dnelopmental

1992

among

adherence

recruited

trial

Tonry

increasing

and compliance

K.TobinJ

affecting

for

strategies

recruiting

program

Human

1987

Boatwright

comparison

OmennG

Patrick

and

Participation

ficacy

West

Schleser

M.D

Uhornquist

1xi-

233242

ing

and

of

families

3460

behavioral
IL Bitten

of the psychology

CA

Academic

Ethnographic

persons

10-A

in

In

Single
Dissertation

study
adult

of

Press
of

home

Abstracts