See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235425781

Measurement and Analysis of Individualized Care Comparing Long-Term Care

Nurses and Care Aides

Article  in  The Gerontologist · January 2010

DOI: 10.1093/geront/gnp053.

CITATION READS

1 75

3 authors, including:

Norm O'Rourke Sienna Caspar

Ben-Gurion University of the Negev University of Lethbridge
183 PUBLICATIONS   6,386 CITATIONS    29 PUBLICATIONS   551 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Use and Abuse of Alcohol and other Substances in Bipolar Disorder View project

Hotel Study View project

All content following this page was uploaded by Norm O'Rourke on 27 May 2014.

The user has requested enhancement of the downloaded file.

Brief Report
The Gerontologist © The Author 2009. Published by Oxford University Press on behalf of The Gerontological Society of America.
Vol. 49, No. 6, 839–846
doi:10.1093/geront/gnp053
Measurement and Analysis of Individualized
Care Inventory Responses Comparing
Long-Term Care Nurses and Care Aides
Norm O’Rourke, PhD, RPsych,1,2 Neena L. Chappell, PhD, FRSC, CRC,3
and Sienna Caspar, MA4
Purpose: Motivating and enabling formal care-
givers to provide individualized resident care has
become an increasingly important objective in long-
term care (LTC) facilities. The current study set out to
examine the structure of responses to the individual-
ized care inventory (ICI). Design and Meth-
ods: Samples of 242 registered nurses (RNs)/
licensed practical nurses (LPNs) and 326 care aides
were recruited from 54 LTC facilities in 3 of 5 British
Columbia health authorities. Baseline confirmatory
factor analytic (CFA) models were computed sepa-
rately for RNs/LPNs and care aides; invariance anal-
yses were next undertaken to compare these CFA
models. Results: For both RNs/LPNs and care
aides, support was found for a 4-factor model of ICI
responses mapping onto a higher order individual-
ized care (IC) construct. This model was largely
equivalent between formal caregiver groups, al-
though the relative contribution of certain first-order
factors differed between the two. Of further note,
both groups appear to interpret and respond to 31 of
35 ICI items in a similar manner. Implica-
tions: The results of this study provide further sup-
1
Address correspondence to Norm O’Rourke, PhD, RPsych, Depart-
ment of Gerontology, Simon Fraser University–Vancouver Campus,
#2800-515 West Hastings Street, Vancouver, BC, V6B 5K3 Canada.
E-mail: orourke@sfu.ca
2
Department of Gerontology, Simon Fraser University–Vancouver
Campus, British Columbia, Canada.
3
Centre on Aging, University of Victoria, British Columbia, Canada.
4
Interdisciplinary Studies Graduate Program, University of British
Columbia, Vancouver, Canada.
Vol. 49, No. 6, 2009
All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org.
Advance Access Publication on June 11, 2009
Downloaded from gerontologist.oxfordjournals.org at Simon Fraser University on July 14, 2010
port for the psychometric properties of ICI responses.
Although further research is required, the ICI appears
to be an appropriate self-report measure. This instru-
ment may be used by researchers, policymakers, ad-
ministrators, and practitioners alike to assess strengths
as well as areas for improving the delivery of IC to
LTC residents by formal caregivers.
Key Words: Dementia care, Formal caregivers,
Individualized care, Long-term care, Scale reliability,
Scale validation
Individualized care (IC), also known as client-
centered, resident-directed, psychosocial, and
flexible care, has been described as the model of
choice for residents in long-term care (LTC) fa-
cilities (Calkins, 2001; Suhonen, Valimake, &
Katajisto, 2000). Yet, inconsistent operational
definitions of this construct have impeded devel-
opment of psychometrically sound measures. Re-
cently, however, Chappell, Reid, and Gish (2007)
developed a self-report instrument for LTC staff
measuring distinct facets of IC; thus far, responses
to this measure have demonstrated acceptable re-
liability. This brief report presents new data on
the latent structure of this Individualized Care In-
ventory (ICI), comparing responses between regis-
tered nurses (RNs)/licensed practical nurses
(LPNs) and care aides, and providing further psy-
chometric support for its use by researchers and
839
facility administrators wishing to improve care
(e.g., pre- and postintervention measurement).
The ICI was established through a review of the
literature, direct observation of care within LTC fa-
cilities, and ongoing consultation with an expert
panel. Six facets were identified: knowing the resi-
dent, opportunity for autonomy and choice for the
resident, open communication between staff mem-
bers and with residents, family involvement, resi-
dents connecting with others during activity
programming and in day-to-day life, and a home-
like physical environment. Support for the two fam-
ily involvement facets has previously been reported
by Reid, Chappell, and Gish (2007). Initial research
suggests acceptable reliability of responses to the
remaining four (Chappell, Reid, & Gish, 2007).
The current brief report compares ICI responses
between RNs/LPNs and care aides. As recently
noted by Caspar and O’Rourke (2008), RNs and
LPNs have relatively more education and higher
salaries yet have less resident contact than care
aides. There are, in other words, sound reasons to
examine whether the structure of responses to the
ICI differs between these occupational groups.
More precisely, the current study was undertaken
to examine the structure of ICI responses for both
RNs/LPNs and care aides, to ascertain whether the
structure of responses is invariant (or equivalent)
between the two, and to determine if both group-
ings of formal care providers interpret and respond
to ICI items in a similar manner.
Methods
Participants
Convenience samples of RNs/LPNs and care
aids were recruited from 54 LTC facilities within
three of five health authorities (or regions) in Brit-
ish Columbia (BC) as part of a larger study of staff
empowerment and IC (N = 568). To be eligible,
participants were required to work on a perma-
nent full-time or part-time basis (or as a casual in
an equivalent position), be proficient in English,
and have been employed in that facility for a mini-
mum of 6 months.
RNs and LPNs were categorized together
based on two initiatives implemented in BC due
to RN shortages: the Ministry of Health’s initia-
tive enabling LPNs to function in a full scope of
practice capacity (Harvey, Sams, Bosancic, &
Brunke, 2003) and a strategy developed to re-
place the majority of RNs working in LTC facili-
ties with LPNs (Greenlaw, 2003). Both initiatives
840
had been fully implemented prior to this study; as
a result, LPNs and RNs now have similar roles
and responsibilities within LTC (i.e., team lead-
ers and supervision of care aides), although dif-
ferences in skill and experience may persist.
Specific to ICI responses, however, no between-
groups differences were observed for any of the
subscales or combined ICI responses, Hotelling’s
T = .01, F(4,237) = .53, ns. Based on these find-
ings and the initiatives previously described, RNs’
and LPNs’ responses were combined for subse-
quent analyses.
Surveys were completed by respondents working
in facilities where managers agreed to allow their
staff to be approached to participate (n = 31) as well
Downloaded from gerontologist.oxfordjournals.org at Simon Fraser University on July 14, 2010
as at off-site educational sessions completed by par-
ticipants working in facilities where managers either
did not respond or refused to enable on-site partici-
pation (n = 23). The most common reasons for re-
fusal were a reported lack of time or resources.
All administrators were asked to provide de-
scriptive facility information even when we were
not invited to collect data on-site. Of note, there
were no significant differences between consenting
and nonconsenting facilities in terms of day- or
nighttime staffing ratios; nor did consenting and
nonconsenting facilities differ in numbers of resi-
dents, t(43) = .02, ns; unionization of care staff,
c2(df = 1) = .57, ns; or facility ownership (i.e., pub-
lic, private not for profit, private for profit), c2(df =
2) = 5.32, ns.
Responses to study variables were also com-
pared between participants recruited within con-
senting facilities and those who provided data at
off-site educational sessions. Of note, there were
no significant IC response differences between
RNs/LPNs or care aides working in consenting
and nonconsenting facilities, t(223) = 1.88, ns, and
t(324) = .92, ns, respectively. These findings sug-
gest that neither facility features nor reported abil-
ity to provide IC differ as a result of administrators’
willingness to facilitate this study. That is, admin-
istrators did not appear to bias responses (via con-
sent or refusal), providing greater confidence in the
generalizability of findings.
Individualized Care Inventory
The ICI (Chappell et al., 2007) is a 34-item scale
with responses provided along a 4-point Likert-
type key. The ICI consists of four subscales: know
residents (IC-KNOW, 11 items), resident autono-
my (IC-AUTONOMY, 11 items), communication
The Gerontologist
 Table 1. Descriptive Features and Psychometric Properties of Model Variables for RNs/LPNs and Care Aides

ICI responses M SD Range a Kurtosis Skewness

RNs/LPNs (n = 242)
Know residents 34.99 4.55 22–44 0.74 −0.45 −0.07
Autonomy 34.97 6.02 18–55 0.81 0.02 0.07
Staff–staff communication 31.17 4.53 16–40 0.82 0.25 −0.35
Staff–resident communication 8.41 1.80 4–12 0.68 −0.45 0.16
Care aides (n = 326)
Know residents 34.15 5.05 16–44 0.74 −0.05 −0.18
Autonomy 34.99 6.36 18–54 0.77 0.02 0.11
Staff–staff communication 31.10 5.47 13–40 0.86 −0.44 −0.34
Staff–resident communication 9.22 2.02 3–12 0.74 −0.35 −0.37
Note: ICI = individualized care inventory; RN = registered nurse; LPN = licensed practical nurses.

Downloaded from gerontologist.oxfordjournals.org at Simon Fraser University on July 14, 2010

staff–staff (IC-COMMUNICATION-SS, 10
items), and communication staff–resident (IC-
COMMUNICATION-SR, 3 items). Reported in-
ternal consistency of responses as measured by
Cronbach’s alpha is within acceptable to good
range for IC-KNOW (a = .77), IC-AUTONOMY
(a = .80), and IC-COMMUNICATION-SS (a =
.84). A somewhat suboptimal coefficient has been
reported for IC-COMMUNICATION-SR (a = .67)
possibly due to the small number of items (three)
within this subscale. As noted by O’Rourke, Hatch-
er, and Stepanski (2005), internal consistency can
be underestimated with fewer than eight items.
Results
The order of questionnaire presentation was
randomized creating two counterbalanced for-
mats. Comparative analyses indicated that re-
sponse levels did not significantly differ between
groups for any of the four ICI subscales; it is thus
unlikely that order effects confounded participant
responses. Table 1 provides descriptive informa-
tion for responses to study measures (e.g., internal
consistency, univariate normality) for RNs (n =
177)/LPNs (n = 65) and care aides recruited for
this study (n = 326).
Assessment of Between-Group Response Levels
Between-groups analyses were first computed to
compare response levels between RNs/LPNs and
care aides. There was an overall statistically sig-
nificant between-group difference in responses to
the ICI, Hotelling’s T = .08, F(4, 563) = 11.26, p <
.01. Additional analyses indicate that the only uni-
variate difference to attain significance (Bonferroni
adjusted alpha = .01) was IC staff–resident com-
munication, F(1, 566) = 24.56, p < .01, partial
h2 = .04. Care aides reported somewhat higher lev-
els of IC staff–resident communication (M = 9.22,
SD = 2.02) than RNs/LPNs (M = 8.41, SD = 1.08).
Baseline CFA Models
Separate CFA models were computed for RNs/
LPNs and care aides based on the assumed struc-
ture in which items loaded upon one of four a pri-
ori first-order factors (i.e., know the resident,
resident autonomy, staff-to-resident communica-
tion, staff-to-staff communication). Each of these
four factors was assumed to significantly contrib-
ute to measurement of a higher order IC latent
construct. Support for this measurement model
has previously been reported (Caspar & O’Rourke,
2008). CFA models were computed with the
AMOS 16.0 statistical program using the maxi-
mum likelihood method of parameter estimation.
Care Aides CFA Model.—The CFA model was
first computed for care aides (n = 326). Statistical
power was .99 for this model according to the
formula provided by MacCallum, Browne, and
Sugawara (1996). All items with the exception of
ICA07 loaded significantly upon their respective
factors (i.e., t values > 1.96). This resident autono-
my item (“Do you generally feel that you have
done things for residents when they could have
done it for themselves?”) was deleted from the care
aides’ CFA model for all subsequent analyses. For
care aides, it would seem that this item is unrelated
to measurement of resident autonomy.
Each of the four first-order factors contributed
significantly to measurement of an overarching
higher order construct. As presented in Figure 1, the
know the resident factor provided the greatest con-
tribution to measurement of this second-order

Vol. 49, No. 6, 2009 841

 Downloaded from gerontologist.oxfordjournals.org at Simon Fraser University on July 14, 2010

Figure 1. Confirmatory factor analysis of care aide individualized care instrument responses (n = 326). Note. Maximum likeli-
hood estimates (standardized solution and significance levels). Asterisks (*) denote parameters initially fixed to 1.0 for purposes
of scaling and statistical identification, thus significance levels cannot be computed for these five parameters. Numbers in paren-
theses indicate significance levels (statistically significant t values > 1.96).

construct, followed by staff-to-staff communication,
resident communication, and resident autonomy.
Correction was made for correlated error between
related item pairs based on AMOS modification in-
dexes (also known as Lagrange multipliers; Byrne,
2001). The CFA model for care aides subsequently
indicated effective fit of data, c2(df = 472) = 521.75,
p > .05). In accord with the significance thresholds
suggested by Hu and Bentler (1999), the Compara-
tive Fit Index (CFI) is within optimal parameters

842 The Gerontologist

(i.e., CFI > .95; CFI = .99) as are the Adjusted
Goodness-of-Fit Index (AGFI; i.e., AGFI ≥ .90,
AGFI = .90) and the Root Mean Square Error of
Approximation (i.e., RMSEA < .05, RMSEA = .018).
Of note, the full 90% confidence interval (CL90) for
the RMSEA statistic is within optimal parameters
(.00 ≥ RMSEA CL90 ≥.027).
RNs/LPNs CFA Model.—The CFA model was
next computed for RNs/LPNs. Although the sam-
ple size for this grouping was somewhat smaller
(i.e., n = 242), statistical power was the same as for
care aides (i.e., power = .99; MacCallum et al.,
1996). All items contributed significantly to mea-
surement of their respective first-order factors,
which, in turn, contributed significantly to mea-
surement of the overarching IC latent construct
(see Figure 2). Among RNs/LPNs, the know the
resident first-order factor again provided the stron-
gest contribution to measurement of IC; however,
resident autonomy emerged second followed by
staff-to-staff and resident communication. These
results suggest that RNs/LPNs perceive resident
autonomy as contributing more to IC (the least im-
portant factor among care aides), followed by the
communication latent constructs. In contrast, care
aides appear to place greater emphasis upon both
communication constructs rather than resident au-
tonomy. The invariance analyses to follow will as-
certain whether path coefficients significantly differ
between care provider groups.
As with the prior CFA model, strong goodness-
of-fit indexes emerged for the RN/LPN model,
c2(df = 493) = 512.38, p > .05, as CFI = .99,
RMSEA = .013, and the 90% range for the RMSEA
statistic is again fully within optimal parameters
(i.e., .00 ≥ RMSEA CL90 ≥.026). In this instance,
however, AGFI was .87, which is less than ideal.
Invariance Analyses
Comparison of Factor Structures.—Subsequent to
establishing baseline CFA models, care aide and RN/
LPN models were next compared to ascertain the
similarity of solutions. Comparison was undertaken
as a partial test of measurement invariance in this
instance due to deletion of item ICA07 from the care
aide model. In other words, no equality constraint
was imposed on this parameter due to between-
group differences in the computation of baseline
models (Byrne, Shavelson, & Muthén, 1989).
Factor structures were first compared to deter-
mine if the four-factor solution mapping onto a
Vol. 49, No. 6, 2009
second-order construct is equivalent between
groups. This solution emerged as viable, suggesting
that the structure of ICI responses may best be mea-
sured by a four-factor solution with each of these
mapping onto a higher order construct, c2(df =
965) = 1034.21, ns; CFI = .99, AGFI = .88, RM-
SEA = .011, .000 ≥RMSEA CL90 ≥.017 (see
Table 2). This finding suggests that the basic la-
tent structure of responses is equivalent for both
care aides and RNs/LPNs. These initial compari-
sons, however, do not tell us if the strength of
association between constructs is statistically
equivalent. This was next examined, comparing
parameter estimates between first- and second-
order constructs.
Downloaded from gerontologist.oxfordjournals.org at Simon Fraser University on July 14, 2010
Invariance of Factor Loadings.—The strength of
association between the ICI factors contributing to
measurement of the overarching IC second-order
construct was next examined. Although parameter
estimates linking staff-to-staff communication and
IC are indistinguishable between groups, Dc2(Ddf
= 1) = 2.31, ns, statistically significant differences
were observed for others. More precisely, the con-
tribution of staff-to-resident communication to
measurement of IC is significantly greater for care
aides than RNs/LPNs, Dc2(Ddf = 1) = 3.92, p < .05.
Even more pronounced is the between-group dif-
ference in the contribution of resident autonomy
to IC, Dc2(Ddf = 1) = 9.77, p < .01 (see Table 2). As
previously noted, the strength of association be-
tween resident autonomy and IC was second only
to know the resident for RNs/LPNs. Among care
aides, however, this factor emerged as least impor-
tant (although a statistically significant component
of IC). This finding is notable given the relative
importance of staff-to-resident communication as
reported by care aides in terms of both response
levels as well as the contribution to IC measure-
ment by this first-order construct. Overall, these
findings suggest similarity in terms of the structure
of ICI responses across formal caregiver groups,
although differences in the relative association
between first- and second-order constructs are
significant.
Item Analyses.—The preceding analyses provide
support for the factorial validity of responses to
the ICI. The final set of invariance analyses were
undertaken to ascertain if RNs/LPNs and care
aides interpreted and responded to ICI items in a
similar manner. This appears to be the case for
843
 Downloaded from gerontologist.oxfordjournals.org at Simon Fraser University on July 14, 2010

Figure 2. Confirmatory factor analysis of registered nurse (RN)/licensed practical nurse (LPN) individualized care instrument
responses (n = 242). Note. Maximum likelihood estimates (standardized solution and significance levels). Asterisks (*) denote
parameters initially fixed to 1.0 for purposes of scaling and statistical identification, thus significance levels cannot be computed
for these five parameters. Numbers in parentheses indicate significance levels (statistically significant t values > 1.96).

each know the resident item, Dc2(Ddf = 10) =
18.07, ns; staff-to-resident communication items,
2
Dc (Ddf = 2) = 5.74, ns; staff-to-staff communica-
2
tion items, Dc (Ddf = 9) = 14.55, ns; and most res-
2
ident autonomy items, Dc (Ddf = 6) = 5.45, ns. As
reported in Table 2, statistically significant re-
sponse differences were observed for the remain-
ing three resident autonomy items, Dc2(Ddf = 3) =
20.70, p < .01. Although the relative weight of fac-
tors differed for staff-to-resident communication,

844 The Gerontologist

 Table 2. Summary of Specifications and Fit Statistics for Invariance Analyses

RMSEA
Successive constraints applied c2 df Dc2 Ddf CFI AGFI (RMSEA CL90)
Unconstrained structural model 1034.21 965 — — .99 .88 .011 (.000–.017)
Resident communication upon individualized care 1038.13 966 3.92* 1 .99 .88 .011 (.000–.017)
Resident autonomy upon individualized care 1047.90 967 9.77** 1 .99 .88 .012 (.004–.017)
Staff–staff communication upon individualized care 1050.03 968 2.13 1 .99 .88 .012 (.004–.017)
Know-the-resident items upon latent variable 1068.10 978 18.07 10 .99 .88 .013 (.005–.018)
Resident communication items on latent variable 1073.84 980 5.74 2 .98 .88 .013 (.006–.018)
Staff communication items on latent variable 1088.39 989 14.55 9 .98 .88 .013 (.006–.018)
Resident autonomy items (nonsignificant) 1093.84 995 5.45 6 .98 .88 .013 (.006–.018)
Resident autonomy items (significant)
ICA 02 1098.27 996 4.43* 1 .98 .88 .013 (.007–.018)
ICA 03 1105.81 997 7.54** 1 .98 .88 .014 (.008–.019)
ICA 06 1114.54 998 8.73** 1 .98 .88 .014 (.008–.019)

Downloaded from gerontologist.oxfordjournals.org at Simon Fraser University on July 14, 2010

Notes: AGFI = adjusted goodness-of-fit index; CFI = comparative fit index; RMSEA = root mean square error of approxima-
tion; RMSEA CL90 = 90% confidence intervals for RMSEA values.
*p < .05. **p < .01.

this does not appear attributable to between-group
response differences to items measuring this first-
order construct.
Overall, item analyses suggest that care aides
and RNs/LPNs interpret and respond to 31 of 35
ICI items in a similar manner (excluding ICA07
from the care aide CFA model and item analyses).
In addition to previous findings supporting the un-
derlying structure of responses between these care-
giver groups, item analyses suggest reliability of
measurement.
Conclusions and Discussion
The results of this study provide additional sup-
port for this measure assessing formal caregivers’
reported ability to provide individualized care to
LTC residents (Chappell et al., 2007). This
instrument appears suitable for use with both RNs/
LPNs and care aides. As a result, it has both re-
search and practical utility. For researchers, it en-
ables brief and easy-to-administer measurement of
individualized care. For policymakers, LTC ad-
ministrators and practitioners, the ICI can be used
to identify whether individualized care is being of-
fered, to what degree, and where room for im-
provement might exist. More precisely, the ICI
may be used to identify where further training is of
benefit when deciding how best to utilize finite ed-
ucational resources.
The findings of this study support the four-
factor structure of responses to the ICI. Although
the relative contribution of first-order factors to IC
measurement differs somewhat between RNs/
LPNs and care aides, both caregiver groups appear
to interpret and respond to the majority of ICI
items in a similar manner.
Generalizability of study findings is limited by
various factors, however. First, only participants
who worked in LTC facilities in three of five BC
health authorities were recruited for this study. In
addition, differences between formal care providers
who agreed to participate in this study versus those
who declined cannot be ascertained. Also, respons-
es to the two family subscales were not examined.
This was not feasible for this study as participant
responses must be from the same respondents (to
all scales) to undertake analysis of covariance struc-
tures (e.g., CFA, invariance analyses).
Of further note, representative caregiver sam-
ples were not recruited for this study. We are thus
unable to generalize individualized care response
levels to the overall population of formal caregiv-
ers in this province (e.g., norms for subscales).
That was not the intent of this study; instead, mod-
els and invariance analyses were computed to ex-
amine the structure of ICI responses. Recruitment
of convenience samples underscores the need to
replicate these findings with formal caregivers de-
rived from other jurisdictions (e.g., where univer-
sal health care does not yet exist); however, these
findings add to a growing body of research provid-
ing support for the psychometric properties of ICI
responses (Caspar & O’Rourke, 2008; Chappell
et al., 2007). This instrument appears to be a sci-
entifically acceptable self-report measure of formal
caregivers’ reported ability to provide individual-
ized care to LTC residents.

Vol. 49, No. 6, 2009 845

Funding
Support for this study was received by Sienna Caspar from the Social
Sciences and Humanities Research Council of Canada (SSHRC Masters
Fellowship) and the Michael Smith Foundation for Health Research (MS-
FHR Trainee Award).
References
Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic
concepts, applications, and programming. Mahwah, NJ: Lawrence
Erlbaum.
Byrne, B. M., Shavelson, R. J., & Muthén, B. (1989). Testing for equiva-
lence of factor covariance and mean structures: The issue of partial
measurement invariance. Psychological Bulletin, 105, 456–466.
Calkins, M. P. (2001). The physical environment and social environment
of the person with Alzheimer’s Disease. Aging & Mental Health, 5,
S74–S78.
Caspar, S., & O’Rourke, N. (2008). The influence of care provider access
to structural empowerment on individualized care in long-term care
facilities. Journal of Gerontology Series B: Psychological Sciences and
Social Sciences, 63, S255–S265.
Chappell, N. L., Reid, R. C., & Gish, J. A. (2007). Staff-based measures of
individualized care for persons with dementia in long-term care facili-
ties. Dementia: The International Journal of Social Research and Prac-
tice, 6, 527–547.
846
View publication stats
Greenlaw, B. (2003). Licensed practical nurses: Current utilization. Re-
trieved May 5, 2008, from http://www.heu.org/~DOCUMENTS/2008_
NursingTeam/ LPN Utilization Study 2003.pdf
Harvey, A., Sams, C., Bosancic, Z., & Brunke, L. (2003). Process for
change in practice roles: A joint BC nurses’ union, college of licensed
practical nurses, hospital employees union and registered nurses asso-
ciation of BC protocol. Retrieved November 27, 2008, from http://
www.bcnu.org/
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indices in covari-
ance structure analysis: Conventional criteria versus new alternatives.
Structural Equation Modeling, 6, 1–55.
MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power
analysis and determination of sample size for covariance structure
modeling. Psychological Methods, 1, 130–149.
O’Rourke, N., Hatcher, L., & Stepanski, J. (2005). A step-by-step ap-
proach to using SAS for univariate and multivariate statistics (2nd ed.).
New York: Wiley-Interscience.
Reid, R. C., Chappell, N. L., & Gish, J. A. (2007). Measuring family per-
ceived involvement in individualized long-term care. Dementia: The
International Journal of Social Research and Practice, 6, 89–104.
Suhonen, R., Valimake, M., & Katajisto, J. (2000). Individualized care in a
Downloaded from gerontologist.oxfordjournals.org at Simon Fraser University on July 14, 2010
Finnish healthcare organization. Journal of Clinical Nursing, 9, 218–219.
Received August 12, 2008
Accepted January 20, 2009
Decision Editor: William J. McAuley, PhD
The Gerontologist