Content uploaded by Miyong To Kim
Author content
All content in this area was uploaded by Miyong To Kim on Nov 05, 2015
Content may be subject to copyright.
Nursing Research
September/October 2002 Vol 51, No 5
309
The Psychometric Evaluation of Korean
Translation of the Personal Resource
Questionnaire 85-Part 2
Hae-Ra Han
▼
Miyong T. Kim
▼
Clarann Weinert
Background:
Although the Personal Resource Questionnaire
(PRQ 85-Part 2) has been widely used to measure per-
ceived level of social support, results of psychometric evalu-
ation of the measure with a Korean population are not cur-
rently available in the literature.
Objective:
To assess the psychometric properties of the
Korean language version of the PRQ 85.
Method:
An empirical validation of the theoretical relation-
ships among the construct, social support, as measured by
the PRQ 85, and other related constructs was conducted. In
addition, factor structure comparison with a cross-validation
technique was utilized on a combined sample of 555 Korean
adults from three independent studies.
Results:
Construct validity for the Korean PRQ 85 was evi-
denced by statistically significant correlations of perceived
social support with the theoretically relevant variables such
as depression and psychosocial adjustment. Using principal
axis factoring with an oblique rotation, the two-factor solution
was found to be most satisfactory in the first random sample
and was cross-validated in the second sample, accounting
for 52% and 54.1% of the total variance, respectively.
Inspecting the distribution of variables within the factors,
however, only the first factor appeared to be a construct-
related factor; the second factor, which was exclusively rep-
resented by negatively keyed items, was method-related.
Conclusion:
The findings suggest that a one-factor solution
can be used to accurately describe a substantive (or con-
struct-related) factor pattern of the Korean version of the
PRQ 85. Although researchers generally agree that nega-
tively worded questions are important in order to minimize
the “response bias,” this approach appeared to be problem-
atic for Koreans who used the translated PRQ 85. Further
research is warranted to explicate this important method-
ological issue in cross-cultural instrumentation.
Key Words:
cross-cultural measurement
•
Korean
•
Personal
Resource Questionnaire (PRQ)
•
social support
he increasing need for development and adaptation
of culturally appropriate measures has been well
documented (Carlson, 2000; Devins, Beiser, Dion, Pellitier,
& Edwards, 1997). A major challenge associated with
cross-cultural administration of any instrument is estab-
lishing the equivalence of the instrument across the groups
studied (Ellis, Minsel, & Becker, 1989). Because the use of
translated instruments has been a frequent feature of cross-
cultural studies, researchers (Calson, 2000; Cho & Kim,
1998) have emphasized the importance of conducting rig-
orous psychometric evaluations of such translated instru-
ments to establish validity and equivalency before using
them in cross-cultural settings. The purpose of this article
is to report the process of validation of the newly trans-
lated version of the Personal Resource Questionnaire (PRQ
85-Part 2) (Weinert & Brandt, 1987), one of the most psy-
chometrically sound measures of perceived social support,
in a Korean sample drawn from three independent
research studies.
Background
Since the 1970s, interest in the concept of social support
(Cohen & Syme, 1985) has increased dramatically within
the field of nursing research. This increasing attention may
reflect the buffering role it plays in the process of coping
with stressful events such as illness. Evidence points to the
central role that social support plays in alleviating this
stress, so that individuals can achieve better adjustment
outcomes (Badger, 1993; Braden, Mcglone, & Pennington,
1993; Eliopoulos, 1997; Han, 2000).
There are a wide variety of definitions of social support
in the literature. Early studies of social support focused on
Hae-Ra Han, PhD, RN, is Instructor, The Johns Hopkins Univer-
sity School of Nursing, Baltimore, Maryland.
Miyong T. Kim, PhD, RN, is Associate Professor, The Johns Hop-
kins University School of Nursing, Baltimore, Maryland.
Clarann Weinert, SC, PhD, RN, FAAN, is Professor, Montana
State University–Bozeman College of Nursing.
T
310
A Korean Translation of the Personal Resource Questionnaire
Nursing Research
September/October 2002 Vol 51, No 5
the structural aspects of the concept such as marital status
or social network size (House, Robbins, & Metzner, 1982;
Berkman & Syme, 1979). However, later work has empha-
sized the role of social relationship or the functional con-
tent of the relationship (e.g., feeling loved or supported
and having a confidant) (House & Kahn, 1985; Chappell
& Badger, 1989). Development of the PRQ 85 (Weinert &
Brandt, 1987) is based on a functional definition of social
support. According to Weinert and Brandt (1987), social
support is defined as composed of the following five
dimensions: (a) provision of attachment/intimacy, (b)
social integration (being an integral part of a group), (c)
opportunity for nurturant behavior, (d) reassurance of
worth, and (e) the availability of informational, emotional,
and material help.
The PRQ 85 consists of two parts: Part 1 is an assess-
ment of the social networks individuals rely on for situa-
tional support, an aspect of the questionnaire which will
not be addressed in this paper. Part 2 was constructed
according to Weiss’ (1974) theoretical formulation of the
relational provisions of social support. The PRQ 85-Part 2,
hereafter referred to as the PRQ, is a self-administered
norm-referenced instrument measuring the respondents’
perceived social support. The PRQ consists of 25 items
representing each of the five categories listed above on a 7-
point Likert scale. It requires approximately 10 minutes to
complete. Total scores can range from 25 to 175, with
higher scores indicating higher levels of perceived social
support.
The reliability of the PRQ has been assessed by evalu-
ating test-retest reliability and internal consistency reliabil-
ity. For a sample of 100 adults aged 30 to 37 years, the
test-retest reliability yielded a score of 0.72 over a 4- to 6-
week period (Weinert & Brandt, 1987). Cronbach alphas
for the total scale have ranged from 0.85 to 0.93 (Gau,
1996; Long & Weinert, 1992; Weinert, 1988; Weinert &
Brandt, 1987).
Construct validity of the PRQ has been tested by a fac-
tor analytic approach. The number of factors identified in
previous studies has varied from three to five (Weinert,
1987; Yarcheski, Mahon, & Yarcheski, 1992; Gau, 1996).
Weinert (1987) used combined data from three studies by
different researchers. The overall sample consisted of 248
men and women who were primarily White, married, and
middle-class. Using a principal component analysis with an
oblique (oblimin) rotation, a three-factor solution was
determined to be the most satisfactory, with 43.4% of the
variance being explained. The three factors were labeled as
intimacy/assistance, reciprocity, and integration/affirma-
tion; however, the results of the study were not consistent
with the five subscales originally proposed by Weinert and
Brandt (1987). Yarcheski et al. (1992) reported that the
five-factor model was not supported. In a validation study
of the PRQ for adolescents, Yarcheski et al. reported that
a four-factor solution best represented the PRQ structure
in 325 adolescents aged 12 to 21 and that it had 17 items.
The majority of the adolescents were White, with a mean
age of 16.2 years. In a principal component analysis with
an oblique rotation, the four-factor structure accounted for
48.9% of the total variance in social support. The four fac-
tors were labeled as relational provisions of intimacy,
social integration, assistance/guidance, and sense of
alliance, worth, and nurturance. Using the confirmatory
factor analysis approach, Gau (1996) tested the multidi-
mensionality of the PRQ in 345 pregnant women aged 18
to 44 years (mean ⫽ 29.2 years), with the majority being
White. The original five-factor structure was not supported
in the sample. Gau eliminated seven items from the origi-
nal scale that had double- or triple-loadings to latent fac-
tors, low-item reliabilities, or similar wordings. The
trimmed five-factor model with seven items deleted was
found to best fit the data (
2
⫽ 295.01, df ⫽ 130, good-
ness-of-fit index ⫽ 0.92, adjusted goodness-of-fit index ⫽
0.89, standardized root mean square residual ⫽ 0.05, and
critical N ⫽ 199.72).
The construct validity of the PRQ has also been
assessed by correlating the scores for the scale with the
scores from other instruments measuring depression, anxi-
ety, neuroticism, and extroversion (Weinert & Brandt,
1987; Weinert & Tilden, 1990). Weinert and Brandt
(1987) obtained Pearson correlations between perceived
social support and a number of mental health and person-
ality measures. Perceived social support was significantly
related to mental health measures and the personality indi-
cators. A significantly positive correlation (r ⫽ 0.32, p ⬍
.001) was obtained between the extroversion and per-
ceived support, indicating an association between feeling
supported and socially oriented personality traits. A nega-
tive correlation (r ⫽⫺0.28, p ⬍ .001) was found between
neuroticism and perceived support, demonstrating that
persons who are emotionally labile are more likely to per-
ceive less support. Negative relationships were also
observed between perceived support and anxiety (r ⫽⫺0.
37, p ⬍ .001) and depression (r ⫽⫺0.42, p ⬍ .001), indi-
cating that a high score on perceived social support was
associated with low anxiety and depression. Weinert and
Tilden (1990) studied the relationship between the two
social support measures, the PRQ and the Cost and Reci-
procity Index (CRI), and their relationship to other theo-
retically relevant variables in two samples, one consisting
of 333 adults and the other of 99 adults. Evidence of con-
vergent validity was found in a positive relationship
between the PRQ and the support subscale of the CRI in
both samples (r ⫽ 0. 58 and 0 .53, respectively). Conver-
gent validity was also demonstrated by significant correla-
tions between the PRQ and CRI and two measures of fam-
ily well being (the Family APGAR and the Dyadic
Adjustment Scale) in the sample of 333 adults (0.37 to
0.55). Discriminant validity was substantiated by moder-
ate inverse correlations between the PRQ and the Profile of
Mood State (POMS) total as well as subscales (⫺0.28 to
⫺0.48) in the sample of 99 adults.
Although there are numerous measures of social sup-
port, the PRQ has been widely used in the nursing disci-
pline because of its brief, self-rated nature and sound psy-
chometric properties (Dilorio, Faherty, & Manteuffel,
1992; Lee, Graydon, & Ross, 1991; Long & Weinert,
1992; Weinert & Long, 1993). As of October 2001, data-
base searches using MEDLINE, CINAHL, PsycINFO,
Health and Psychosocial Instruments, and HealthSTAR
indicated that the PRQ has been cited in more than 150
journal articles, dissertations, and book chapters. As a
Nursing Research
September/October 2002 Vol 51, No 5 A Korean Translation of the Personal Resource Questionnaire
311
cross-cultural measure, the PRQ has been
translated into Chinese, Japanese, Thai,
Spanish, and Dutch. The results of psy-
chometric evaluation of these language
versions are not yet available in the litera-
ture with the exception of the Chinese
version (Wang & Liu, 2000). Wang and
Liu (2000) reported alpha coefficients of
0.86 for the entire scale and from 0.65 to
0.78 for the subscales in a sample of 284
Taiwanese women aged 60 and older.
Methods
Translation and Equivalence Assessment
of the Korean Personal Resource Ques-
tionnaire
In this first translation of the instrument
into Korean, a systematic procedure was employed. First,
following the guidelines suggested by Brislin (1976), and
Jones and Kay (1992), a back-translation procedure was
used by a team of bilingual researchers (Kim, 1995). Both
functional and conceptual equivalence (Berry, 1981; 1992)
were assessed during the back-translation period. In addi-
tion, the translation team assessed the appropriateness,
representativeness, and readability of the items on the
Korean version of the PRQ. Metric equivalence (Berry,
1981; 1992) was also assessed by comparing empirical
indices generated from the original
English version used with sample of
Anglo Americans.
Samples
Data from three Korean samples
(Table 1) provided the basis for the
current analysis. The first data set
came from 152 first-generation
Korean immigrants whose age
ranged from 18 to 82 years with the
mean age of 43.4 years (SD ⫽ 13.7).
The majority were women (63.6%),
married (78.6%), and had at least a
high school education (88.9%).
About two-thirds (66.2%) of the
respondents had lived in the United
States (US) for more than 10 years (M
⫽ 15.1, SD ⫽ 10.5). This group completed the Korean
PRQ as part of a study (Kim, 1995) investigating depres-
sion experiences in Anglo Americans and Korean Ameri-
cans.
The second data set was obtained from 203 Korean
elderly immigrants aged 60 to 89 years, with a mean age of
69.6 years (SD ⫽ 6.5). The majority were women (63.4%),
married (68.8%), and had less than a high school educa-
tion (61.3%). Most (81.5%) had lived in the US for more
than 10 years, with a mean length of residence of 15.9
The Personal Resource
Questionnaire has been
cited in more than 150
journal articles,
dissertations, and book
chapters
Korean- Korean- Korean Mothers
Americans American of Children
of All Ages Elderly With Cancer Total
(
n
= 152) (
n
= 203) (
n
= 200) (
N
= 555)
Sex (percentage)
Male 36.4 36.6 — 23.2
Female 63.6 63.4 100.0 76.8
Age (yrs)
⬍
30 12.3 — 19.3 10.3
ⱖ
30 and
⬍
40 31.2 — 62.9 31.0
ⱖ
40 and
⬍
50 22.7 — 17.8 12.7
ⱖ
50 and
⬍
60 20.1 — — 5.4
ⱖ
60 and
⬍
70 10.4 56.1 — 23.4
ⱖ
70 3.2 43.9 — 17.1
Marital status (percentage)
Married 78.6 68.8 97.0 81.8
Not married 21.4 31.2 3.0 18.2
Education (percentage)
Less than high school 11.0 61.3 9.1 28.8
High school graduate 26.6 24.0 55.6 35.9
Some college or college 44.8 13.2 32.8 29.0
More than college 17.5 1.5 2.5 6.3
TABLE 1. Demographic Profile of the Respondents
312
A Korean Translation of the Personal Resource Questionnaire
Nursing Research
September/October 2002 Vol 51, No 5
years (SD ⫽ 6.6). These Korean American elderly com-
pleted the Korean version of the PRQ as part of a study
(Kim, Juon, Hill, Post, & Kim, 2001) exploring the preva-
lence of hypertension and barriers to its care.
The third data set included in the analysis came from
200 Korean mothers living in South Korea who had a child
undergoing cancer treatment during the course of a study
(Han, 2000) examining maternal psychosocial adjustment to
childhood cancer. The mothers’ ages ranged from 23 to 46
years, with a mean of 37.4 years (SD ⫽ 5.2). Most of them
were married (97%) and had completed at least a high
school level of education (90.9%). The Korean language ver-
sion of the PRQ was used as a measure of social support,
which had been hypothesized to moderate the effect of stress
on maternal adjustment.
For cross-validation of the factor solution of the Korean
language version of the PRQ, the three Korean samples were
combined to create a new sample pool, which was then ran-
domly split into two groups by computer-generated random
codes. The rationale for the data aggregation was two-fold:
(a) the primary purpose of this analysis was a psychometric
assessment of the Korean version of the PRQ, which required
a large sample size rather than finding contextual differences
in each sample; and (b) the analysis was also based on the
premise that the Korean American participants were first-
generation immigrants who were born, educated, and social-
ized in Korea and, therefore, assumed to share the same lan-
guage and relatively similar cultural backgrounds and values
with those living in Korea. This assumption was partly vali-
dated by comparisons of key demographic variables which
found no differences in (a) sex, (b) age, (c) marital status, (d)
educational level, (e) sample origin (i.e., from which study
the subject was obtained), or (f) PRQ scores between the two
randomly split groups.
Instruments
The Center for Epidemiological Studies-Depression Scale
(CES-D) (Radloff, 1977) is a self-reported, 20-item Likert-
type instrument that was used to measure symptoms associ-
ated with depression in Kim’s (1995) study. The scale consists
of items whose content relates to lowered mood, hopeless-
ness, sadness, crying spells, and alterations in sleep and
appetite patterns. Each item is scored on a 4-point scale, with
possible scores ranging from 20 to 80; the higher the score,
the higher the reporting of depressive symptoms. The Korean
version of the CES-D, first translated by Kuo (1984), has
been used in a number of Korean studies, with alphas rang-
ing from 0.87 to 0.89 (Hurh & Kim, 1990; Cho & Kim,
1998) and a test-retest reliability coefficient of 0.68 (Cho &
Kim, 1998). The coefficient alpha was 0.85 for the Korean
American sample of all ages (Kim, 1995).
The Kim Depression Scale for Korean Americans
(KDSKA) (Kim, 2002) is a 25-item self-report instrument
that measured depression symptoms among Korean Ameri-
can elderly (Kim et al., 2001) in four areas: (a) emotional, (b)
cognitive, (c) behavioral, and (d) somatic. The 4-point Lik-
ert-type scale can range from 0 to 75, with higher scores indi-
cating more depressive symptoms. Evidence of the reliability
and validity has been found (Kim, 2002). The Cronbach
alpha was 0.93 for the Korean American elderly sample (Kim
et al., 2001).
The Psychosocial Adjustment to Illness Scale (PAIS)
(Derogatis, 1986) is a 46-item Likert-type instrument that
measures parental psychosocial adjustment to the current
medical condition of their child via seven domains of adjust-
ment: (a) healthcare orientation, (b) vocational environment,
(c) domestic environment, (d) sexual relationships, (e)
extended family relationships, (f) social environment, and (g)
psychological distress. This 4-point scale can range from 0 to
138, with lower scores indicating better adjustment. Evi-
dence of reliability and validity of the scale has been reported
(Derogatis, 1986). The instrument was translated into
Korean for the purpose of Han’s (2000) study. The alpha
coefficient obtained for the sample of Korean mother was
0.90.
Analysis
A multiple analytic approach was employed. First, to evalu-
ate the internal reliability and item homogeneity of the
Korean translation, Cronbach alphas were calculated for
each of the subscales and for the PRQ total scores in three
samples. In addition, item analysis was performed to see if
any items showed inter-item correlations (⬎ 0.30 or ⬍ 0.70)
and item-total correlations (⬎ 0.30) that consistently fell
below the desired criteria (Nunnally & Bernstein, 1994)
across the three samples.
Construct validity was substantiated by examining the
relationship between social support and a variety of adjust-
ment outcomes to stress. According to a theoretical proposi-
tion by Cohen and Syme (1985), social support plays a role
in alleviating the impact of stress so that the individual can
achieve better adjustment outcomes. Therefore, for the
Korean American samples (Kim, 1995; Kim et al., 2001), it
was hypothesized that there would be an inverse relationship
between perceived social support and psychological distress
(depression). For the sample of Korean mother (Han, 2000),
it was hypothesized that there would be a positive relation-
ship between perceived social support and better psychoso-
cial adjustment.
Finally, using two randomly split samples, factor analy-
ses were conducted. Using a principal axes analysis with an
oblique rotation, the optimal number of factors was estab-
lished in Group 1. The principal axis approach uses free-esti-
mated communalities to identify the extent to which the vari-
ables (items) can be attributed to the common factors, while
the principal component approach assumes the initial com-
munality estimate to be 1.0 (i.e., the variance is perfectly
accounted for by the set of factors underlying the matrix)
(Gorsuch, 1983). Oblique rotation was used to allow rota-
tion of the initial factor axes to best represent clusters of vari-
ables and to be consistent with previous studies (Weinert,
1987; Yarsheski et al., 1992). The generalizations of a social
support structure was cross-validated in Group 2, using the
same extraction and rotation techniques.
Results
Descriptive Psychometric Statistics
Table 2 summarizes means, standard deviations, and
Cronbach alpha reliability coefficients for three different
samples on the Korean PRQ. Further item analysis
revealed item-total correlations ranging from .08 to .71.
Nursing Research
September/October 2002 Vol 51, No 5 A Korean Translation of the Personal Resource Questionnaire
313
One item (item p, “I feel no one has the same problem as
I”) consistently fell below the .30 criterion of item-total
correlation in all three samples. Inter-item correlations
showed an additional item (item x, “People think that I’m
not as good a friend as I should be”) that did not meet the
desirable standard of ⬎ .30 or ⬍ .70 across samples.
Construct Validity Testing
Construct validity for the Korean PRQ was assessed by
testing the hypothesized relationships between perceived
social support and adjustment outcomes to stress. The
scores on the Korean PRQ were correlated with the scores
from the CES-D for Korean Americans of all ages (Kim,
1995), the KDSKA in Korean American elderly (Kim et al.,
2001), and the PAIS in Korean mothers of children with
cancer (Han, 2000). Statistically significant inverse relation-
ships were obtained between perceived social support and
depression for Korean Americans of all ages and Korean
American elderly (r ⫽⫺.495 and ⫺.228, respectively; for
both, p ⱕ .001). In addition, the social support scores on the
Korean PRQ were positively correlated with better adjust-
ment (lower PAIS scores) in Korean mothers of children
with cancer, as hypothesized (r ⫽⫺.462, p ⱕ .001).
Factor Analyses
Using a combined sample, a principal axis factoring with
an oblique rotation was performed in Group 1 (N ⫽ 290).
Four factors with an eigenvalue ⬎ 1 were extracted,
explaining 57.5% of the variance. The scree plot, however,
indicated that there were only two factors with adequate
eigenvalues before the breaking point, the point where the
steep descent stops and the eigenvalues start to level off.
When a two-factor solution was performed, the extracted
factors explained 52% of the total variance in social sup-
port. Factor I was the strongest, accounting for the major-
ity of the variance (44.5%). Factor II explained an addi-
tional 7.5% of the variance. When the original five-factor
solution, based on Weiss’ (1974) conceptualization of
social support, was ordered, the factor matrix of the five-
factor solution failed to rotate.
Items were selected based on the following criteria: (a)
an item-factor loading of at least .45 and (b) a difference of
at least .20 between an item’s loading on the primary factor
and any other factor (Carmines & Zeller, 1979). The two-
factor rotated solution had 20 and 3 items, respectively,
that met the above criteria for item selection. Two items
(items p and x) did not meet the acceptable range of factor
loadings (⬎ 0.45). Inspection of the distribution of vari-
ables among the factors revealed that most of the items
loaded on the first factor, while only the negatively worded
items loaded saliently on the second factor (Table 3).
To check the consistency (reliability) of the factor struc-
ture obtained from the initial analysis, a principal axis fac-
toring with an oblique rotation was done on the second
random sample (N ⫽ 265). Three factors were extracted
at this time, explaining 57.2% of the total variance. How-
ever, the scree plot and distribution of the variables within
the factors also supported using two factors. The two-fac-
tor solution accounted for 54.1% of the total variance in
Group 2, with Factor I explaining 47% of the variance and
Factor II 7.1%. When the loadings of the extracted factors
were compared with those found in Group 1, very similar
patterns were emerged. The majority of items loaded on
Factor I, while negatively keyed items loaded on Factor II
(Table 3). Two items (f and p) appeared to double-load on
both factors; the factor loading (.439) of one of the items
(item f, “I spend time with others who have the same inter-
est that I do”) on Factor I marginally met the desired level
of .45, but the remaining factor loadings fell below the
acceptable range (see Table 3).
With the first factor appearing to be the only construct-
related factor, we further examined a factor solution of the
Korean PRQ after excluding five negatively keyed items
from the scale (items d, g, j, p, and x). Using a principal
axis factoring with an oblique rotation, two factors were
extracted from the first random sample, explaining 58.9%
of the variance. The first factor explained most of the vari-
ance (55%). In addition, the scree plot clearly showed that
there was only one clear factor. Therefore, a single-factor
solution was conducted. The results of this unidimensional
factor solution revealed a much clearer factor structure for
the Korean sample, yet did not reduce the amount of total
variance explained (54.8%), when compared to the two-
factor solution from the exploratory factor analysis. The
factor matrix showed that every item loading was above
the desired criterion of .45. The unidimensional structure
was cross-validated in the second random sample. At this
time, 58% of the variance was explained, with all item
loadings above the criterion (⬎ .45) except for item f,
which had a factor loading value of .425. Coefficient
alphas were calculated for the 20-item PRQ; the value
obtained was .96 for both samples.
Mean (
⫾⫾
SD) Cronbach Alpha
Korean Americans of all ages (
n
= 152) 130.9 (20.2) 0.90
Korean American elderly (
n
= 203) 88.3 (9.8) 0.91
Korean mothers of children with cancer (
n
= 200) 123.0 (22.4) 0.88
Tot a l (
N
= 555) 112.5 (25.9) 0.93
TABLE 2. Means, Standard Deviations (SDs), and Internal Reliability
Coefficients (Cronbach Alpha) of Personal Resource Questionnaire
Scores in Korean Samples
314
A Korean Translation of the Personal Resource Questionnaire
Nursing Research
September/October 2002 Vol 51, No 5
Discussion
While many researchers have demonstrated the multidi-
mensionality of the original English version of the PRQ
(Weinert, 1987; Yarcheski et al., 1992; Gau, 1996), the
results of this study indicate that the five hypothesized
domains by Weiss (1974) are not supported in a Korean
sample, when a Korean language version of the PRQ is
used. Rather, the factor structure that emerged from the
Korean sample suggested that the Korean PRQ may be
used as a unidimensional scale, since only one substantive
factor and one potential method bias factor were
extracted. The first factor, which contributed to the major-
ity of explained variance, included items representing inti-
macy/assistance, reciprocity, and integration/affirmation
dimensions of social support. The second factor, which
explained the small but non-negligible amount of variance,
could not be interpreted as a substantive factor because no
coherent commonality was present, except that entire
items had negative or double negative anchor. The unidi-
mensionality of the Korean PRQ became clearer when the
20-item PRQ was factor-analyzed, after excluding nega-
tively worded items. The amount of total variance
explained increased with the 20-item scale in both samples,
as compared to that for the 25-item scale.
These findings are not surprising, on both empirical
and theoretical grounds. First, the high internal consis-
tency (Cronbach alphas ranging from .88 to .91) for the
total scale and high inter-item correlations that exceeded
the desirable standard (with exception of some negatively
worded items) provide partial evidence that the scale could
be used as a unidimensional scale in this population. Sec-
Group 1 (
N
= 290) Group 2 (
N
= 265)
Factors Factor I Factor II Factor I Factor II
Perceived social support
a 0.801 0.808
b 0.672 0.726
c 0.758 0.758
e 0.813 0.788
f 0.498 0.439 0.352
h 0.744 0.761
i 0.777 0.739
k 0.645 0.720
l 0.637 0.669
m 0.699 0.739
n 0.713 0.704
o 0.743 0.727
q 0.703 0.781
r 0.824 0.838
s 0.866 0.851
t 0.693 0.661
u 0.743 0.828
v 0.828 0.869
w 0.803 0.845
y 0.775 0.866
Negatively worded items
d 0.573 0.601
g 0.550 0.550
j 0.678 0.597
p 0.317 0.233 0.253
x 0.356 0.294
Eigenvalue 11.1 1.9 11.7 1.8
Variance (percentage) 44.5 7.5 47.0 7.1
TABLE 3. Item Loadings for the Two-Factor Solution on the Structure
Matrix
Nursing Research
September/October 2002 Vol 51, No 5 A Korean Translation of the Personal Resource Questionnaire
315
ond, the sum scores of the total PRQ, which were used in
the assessment of the theoretical relationships with other
related constructs, performed well. That is, the correlations
among the sum scores of the PRQ and the scores repre-
senting other theoretically related constructs were statisti-
cally significant. In fact, using the sum score of an entire
scale in analysis has an implicit premise of unidimension-
ality of a scale, in that the operational procedure (summa-
tion of scores) allows equal contribution (not weighted)
from each item without having subscales.
It is important for researchers to validate empirical
findings in light of the theoretical implication of unidi-
mensionality of the Korean version of the PRQ. Unlike the
multiple factor structure that has emerged from dominant
cultural group samples, it is clear that the Korean sample
showed very little distinction among different attributes of
social support (i.e., affective/assistance, reciprocity, and
integration/affirmation). It appears that people who
believed they had one type of social support seemed to
have other types of social support as well. Hence, it is
possible that the close family network and high reliance
on family as the predominant social support agent in
Korean culture (Howard & Berbiglia, 1997; Kauh, 1999)
influenced the unidimensionality of the Korean PRQ.
Another interesting result of this study was the poten-
tial response bias associated with negatively keyed items.
This finding suggests that the majority of Korean partici-
pants may have difficulty in comprehending questions in
the negative or double negative formats. In the classical
measurement theoretical framework, this difficulty is
translated into a systematic bias, which increases the por-
tion of measurement error along with the inherent ran-
dom measurement error. This theory concerning negative
item bias in the Korean population is in its embryonic
stage, since the finding itself was unforeseen. A measure-
ment error caused by negatively worded items in Korean
sample has also been observed in a depression scale vali-
dation study (Kim, 1995). Kim found that the Korean
participants constantly struggled with negatively keyed
items on several different scales, a problem that mani-
fested itself in the poor item-total correlations of those
items (Kim, 1995; 2001).
Various types of response tendencies may exist in dif-
ferent cultural groups. Researchers have found a number
of systematic patterns in the responses to questionnaires
that are made by members of various cultural or linguis-
tic groups. For example, Spanish speakers tend to
respond “good” to items more frequently than do English
speakers (Hayes & Baker, 1998). Asians, in general, man-
ifest strong social desirability traits in responding to ques-
tionnaires and tend to endorse milder or more subtle
forms of expression than the stronger forms of expres-
sions used by Westerners (Cheung & Bagley, 1998;
Devins et al., 1997; Iwata et al., 1998). Another valida-
tion study of the PRQ in a Native American group
(Navajo) reported a systematic bias related to its affective
terms (feelings) (Higgins & Dicharry, 1991). The authors
found that some of the items on the PRQ were inappro-
priate to Native American (Navajo) culture (i.e., personal
feelings). According to Navajo belief, feelings are not dis-
cussed or shared in public; this attitude is similar to the
Asian traditional value that considers public expression
of feelings inappropriate (Howard & Berbiglia, 1997).
Evidence of the construct validity of the Korean PRQ
was provided by statistically significant correlations
between perceived social support and the theoretically
relevant variables. However, the factor solution found in
the Korean samples was different from the hypothesized
five-dimensional structure by Weiss (1974). This study
provided further evidence of cultural influences on par-
ticipant responses. While the factor analysis approach
may be one way of looking at the structure of a construct
across various groups, the utility of the instrument that
measures the construct is often demonstrated by empiri-
cally substantiating the theorized relationships between
variables. The findings of this study suggest that the PRQ-
85 is a useful measure of social support in the study of
Korean samples if it is used as a unidimensional scale. A
construct validity assessment and interpretation of poten-
tial ethnic bias in a well-established cross-cultural mea-
surement for a new cultural group (i.e., Koreans) was illus-
trated in this study.
Accepted for publication April 2, 2002.
Corresponding author: Hae-Ra Han, PhD, RN, The Johns Hopkins
University, School of Nursing, 525 North Wolfe Street, Room 306,
Baltimore, MD 21205-2110 (e-mail: hhan@son.jhmi.edu).
References
Badger, T. A. (1993). Physical health impairment and depression
among older adults. Image - the Journal of Nursing Scholar-
ship, 25, 325-30.
Berkman, L. F., & Syme, L. (1979). Social networks, host resis-
tance, and mortality: A nine-year follow-up study of Alameda
county residents. American Journal of Epidemiology, 109,
186-204.
Berry, J. (1981). Introduction to methodology. In H. Triandis & J.
Berry (Eds.), Handbook of cross-cultural psychology: Volume
2 (pp. 1-28). Boston: Allyn & Bacon.
Berry, J. (1992). Methodological concerns. In J. W. Berry, Y. H.
Poortinga, M. H. Segall, & P. R. Dasen (Eds.), Cross-cultural
psychology: Research and applications (pp. 219-245). New
York: Cambridge University Press.
▼
People who believed they had one type of social
support seemed to have other types of social
support as well
Spanish speakers tend to respond
“good” to items more frequently
than do English speakers
316
A Korean Translation of the Personal Resource Questionnaire
Nursing Research
September/October 2002 Vol 51, No 5
Braden, C. J., Mcglone, K., & Pennington, F. (1993). Specific psy-
chosocial and behavioral outcomes from the systemic lupus
erythematosus self-help course. Health Education Quarterly,
20(1), 29-41.
Brislin, R. (1976). Translation: Application and research. New
York: John Wiley.
Carlson, E. D. (2000). A case study in translation methodology
using the Health-Promotion Lifestyle Profile II. Public Health
Nursing, 17(1), 61-70.
Carmines, E. C., & Zeller, R. C. (1979). Reliability and validity
assessment. Beverly Hills: Sage.
Chappell, N. L., & Badger, M. (1989). Social isolation and
well-being. Journal of Gerontology, 44, s169-s176.
Cheung, C. K., & Bagley, C. (1998). Validating an American
scale in Hong Kong: The Center for Epidemiological Studies
Depression Scale (CES-D). Journal of Psychology, 13, 169-
186.
Cho, M. J., & Kim, K. H. (1998). Use of the Center for Epi-
demiologic Studies Depression (CES-D) Scale in Korea. Jour-
nal of Nervous and Mental Disease, 186, 304-310.
Cohen, S., & Syme, S. L. (1985). Issues in the study and appli-
cation of social support. In S. Cohen & S. L. Syme (Eds.),
Social support and health (pp. 3-22). Orlando, FL: Academic
Press.
Derogatis, L. R. (1986). The Psychosocial Adjustment to Illness
Scale (PAIS). Journal of Psychosomatic Research, 30(1), 77-
91.
Devins, G. M., Beiser, M., Dion, R., Pellitier, L. G., & Edwards,
R. G. (1997). Cross-cultural measurements of psychological
well-being: The psychometric equivalence of Cantonese,
Vietnamese, and Laotian translations of the Affect Balance
Scale. American Journal of Public Health, 87, 794-799.
DiIorio, C., Faherty, B., & Manteuffel, B. (1992). Self-efficacy
and social support in self-management of epilepsy. Western
Journal of Nursing Research, 14, 292-307.
Eliopoulos, C. (1997). Chronic care coaches: Helping people to
help people. Home Healthcare Nurse, 15, 185-188.
Ellis, B. B., Minsel, B., & Becker, P. (1989). Evaluation of atti-
tude survey translations: An investigation using item
response theory. International Journal of Psychology, 24,
665-684.
Gau, M. (1996). The relationship between prenatal stress,
social support, spiritual well-being, and maternal-fetal attach-
ment for pregnant women. Unpublished doctoral dissertation.
University of Maryland at Baltimore.
Gorsuch, R. L. (1983). Factor analysis (2
nd
ed.). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Han, H. R. (2000). Maternal coping with childhood cancer.
Unpublished doctoral dissertation. University of Maryland,
Baltimore.
Hayes, R. P., & Baker, D. W. (1998). Methodological problems in
comparing English-speaking and Spanish-speaking patients’
satisfaction with interpersonal aspects of care. Medical Care,
36(2), 230-236.
Higgins, P. G., & Dicharry, E. K. (1991). Measurement issues
addressing social support with Navajo women. Western Jour-
nal of Nursing Research, 13(2), 242-255.
House, J. S., & Kahn, R. L. (1985). Measures and concepts of
social support. In S. Cohen & S. L. Syme (Eds.), Social support
and health (pp. 83-108). Orlando, FL: Academic Press.
House, J. S., Robbins, C., & Metzner, H. M. (1982). The associa-
tion of social relationships and activities with mortality: Prospec-
tive evidence from the Techmesh community health study. Amer-
ican Journal of Epidemiology, 116, 123-140.
Howard, J. Y., & Berbiglia, V. A. (1997). Caring for childbearing
Korean women. JOGNN: Journal of Obstetric, Gynecologic,
and Neonatal Nursing, 26(6), 665-671.
Hurh, M. H., & Kim, K. C. (1990). Correlates of Korean immi-
grants’ mental health. Journal of Nervous & Mental Disease, 178,
703-711.
Iwata, N., Umesue, M., Egashira, K., Hiro, H., Mizoue, T., Mishima,
N., & Nagata, S. (1998). Can positive affect items be used to assess
depressive disorders in the Japanese population? Psychological
Medicine, 28(1), 153-158.
Jones, E. G., & Kay, M. (1992). Instrumentation in cross-cultural
research. Nursing Research, 41(3), 186-188.
Kauh, T. O. (1999). Changing status and roles of older Korean immi-
grants in the United States. International Journal of Aging and
Human Development, 49(3), 213-229.
Kim, M. T. (1995). Depression manifestations among Korean and
Anglo Americans. Unpublished doctoral dissertation. University of
Arizona.
Kim, M. T. (2002) Measuring depression in Korean Americans:
Development of Kim Depression Scale for Korean Americans.
Transcultural Nursing, 13(2), 100-118.
Kim, M. T., Juon, H. S., Hill, M. N., Post, W., & Kim, K. B.
(2001). Cardiovascular disease risk factors in Korean Ameri-
can elderly. Western Journal of Nursing Research, 23, 269-
282.
Kuo, W. H. (1984). The prevalence of depression among Asian
Americans. Journal of Nervous and Mental Disease, 172, 449-
457.
Lee, R., Graydon, J., & Ross, E. (1991). Effects of psychological
well-being, physical status, and social support on oxygen-
dependent COPD patients level of functioning. Research in
Nursing & Health, 14, 323-328.
Long, K., & Weinert, C. (1992). Description and perceptions of
health among rural and urban adults with multiple sclerosis.
Research in Nursing & Health, 15, 335-342.
Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory
(3
rd
ed.). New York: McGraw-Hill.
Radloff, L. S. (1977). The CES-D scale: A self report depression
scale for research in general population. Applied Psychological
Measures, 1, 385-394.
Wang, H. H., & Liu, Y. Y. (2000). The relationship between social
support and well-being of rural elderly women. Kaohsiung
Journal of Medical Sciences, 16(12), 626-633.
Weinert, C. (1987). A social support measure: PRQ 85. Nursing
Research, 36, 273-277.
Weinert, C. (1988). Measuring social support: Revision and fur-
ther development of the Personal Resource Questionnaire. In
C. Waltz & O. L. Strickland (Eds.), Measurement of nursing
outcomes: Volume 1. Measuring client outcomes (pp. 309-
327). New York: Springer Publishing Company.
Weinert, C., & Brandt, P. A. (1987). Measuring social support
with the Personal Resource Scale. Western Journal of Nursing
Research, 9, 589-602.
Weinert, C., & Long, K. (1993). Support systems for the spouse
of chronically ill persons in rural areas. Family and Commu-
nity Health, 16, 46-54.
Weinert, C., & Tilden, V. P. (1990). Measuring social support:
Assessment of validity. Nursing Research, 39, 213-216.
Weiss, R. S. (1974). The provisions of social relationship. In Z.
Rubin (Ed.), Doing unto others (pp, 17-26). Englewood Cliffs,
NJ: Prentice-Hall.
Yarcheski, A., Mahon, N. E., & Yarcheski, T. J. (1992). Validation
of the PRQ85 social support measure for adolescents. Nursing
Research, 41, 332-337.
