Assessing Cross-Cultural Equivalence Through Confirmatory Factor Analysis

Western Journal of Nursing Research http://wjn.sagepub.com

Assessing Cross-Cultural Equivalence Through Confirmatory Factor Analysis Kuei-Hsiang Hsueh, Linda R. Phillips, WenYun Cheng and Sandra J. Fulton Picot West J Nurs Res 2005; 27; 755 DOI: 10.1177/0193945905276585 The online version of this article can be found at: http://wjn.sagepub.com/cgi/content/abstract/27/6/755

Published by: http://www.sagepublications.com

On behalf of: Midwest Nursing Research Society

Additional services and information for Western Journal of Nursing Research can be found at: Email Alerts: http://wjn.sagepub.com/cgi/alerts Subscriptions: http://wjn.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations http://wjn.sagepub.com/cgi/content/refs/27/6/755

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

Western Journal of Nursing Research, 2005, 27(6), 755-771 Western Journal of Nursing Research October 2005, Vol. 27, No. 6

10.1177/0193945905276585

Assessing Cross-Cultural Equivalence Through Confirmatory Factor Analysis Kuei-Hsiang Hsueh Linda R. Phillips WenYun Cheng Sandra J. Fulton Picot This article presents a study in which Picot’s caregiver rewards scale (PCRS), originally developed in English, was cross-culturally validated with 137 Chinese adult children family caregivers in the United States using confirmatory factor analysis. A one-factor structure of the 21item revised Chinese PCRS was supported as indicated by goodness-of-fit index = .94, adjusted goodness-of-fit index = .93, standardized root mean square residual = .09, and chi-square to df ratio = 2.7. Chi-square for this model was (c2 [189, n = 137] = 514, p < .05). The standardized alpha was .90. All factor loadings were significant (p < .00) and above .30. Construct validity was supported by significant associations with caregivers’ filial beliefs (r = .32, p < .01) and caregivers’ filial affection toward parents (r = .23, p < .05). Results suggest that the Chinese PCRS is a summative measure of perceived caregiver rewards with potential for evaluating interventions to reduce perceived burden among adult children family caregivers. Keywords: family caregiving; confirmatory factor analysis; cross-cultural equivalence; Chinese Picot caregiver rewards scale

With the recent increased focus on family caregiving, substantial attention has been directed at positive family caregiving experiences across cultures (Jones, Zhang, Jaceldo-Siegl, & Meleis, 2002; Kramer, 1997; Picot, 1995; Picot, Youngblut, & Zeller, 1997; Shyu, Archbold, & Imle, 1998; Yamamoto-Mitani et al., 2004). For Chinese caregivers, in particular, Kuei-Hsiang Hsueh, Ph.D., R.N., Assistant Professor, University of Missouri—St. Louis, College of Nursing; Linda R. Phillips, Ph.D., R.N., F.A.A.N., F.G.S.A., Professor of Nursing and Medicine, University of Arizona, Co-Director of the Arizona Center on Aging, and Director of the Arizona Geriatric Education Center; Wen Yun Cheng, R.N., Ph.D. candidate at the University of Minnesota, Senior Instructor and Clinical Nurse Specialist, National Cheng Kung University, School of Nursing, Taiwan; Sandra J. Fulton Picot, Ph.D., R.N., F.A.A.N., Associate Professor, University of Maryland, Baltimore School of Nursing, Department of Organizational Systems and Adult Health. DOI: 10.1177/0193945905276585 © 2005 Sage Publications 755

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

756

Western Journal of Nursing Research

positive experiences in providing care for dependent relatives are major predictors of satisfaction with the family caregiving role (Jones et al., 2002; Shyu et al., 1998). Existing studies, however, have used only qualitative methods. Picot’s caregiver rewards scale (PCRS; Picot et al., 1997) is the only selfreport questionnaire that clearly conceptualizes positive experiences in relation to family caregiving and operates those concepts in an instrument that has been empirically validated using confirmatory factor analysis (CFA); yet it is available only in English. No cross-culturally validated instrument is available for assessing Chinese caregivers’ positive perceptions of the family caregiving role. The establishment of a quality, cross-culturally validated instrument is needed to evaluate positive caregiving experiences among Chinese family caregivers. THEORETICAL AND OPERATIONAL CONSTRUCT OF PCRS

The PCRS was originally designed as a 27-item, two-facet scale to capture intrinsic and extrinsic rewards based on qualitative work using thematic analysis, literature review, and panel experts (Picot et al., 1997). Intrinsic rewards were defined as personal or internal feelings of achievement and growth. Extrinsic rewards included perceptions of rewards through verbal or nonverbal communication with external sources such as God or health care providers. The term rewards was adopted from Nye (1978) and Thibaut and Kelley (1959) to recognize positive perceptions of the family caregiving role. The 27-item PCRS obtained a coefficient alpha of .89 with a pilot sample of 20 Black family caregivers (Picot et al., 1997). After deleting and revising items that were redundant or lacked clarity, a revised 24-item PCRS was generated, and its content relevance was supported by two content experts using the content validity index (Lynn, 1986). The revised 24-item scale was then tested on a sample of 83 Black female caregivers and had a coefficient alpha of .83. Construct validity was supported by significant correlations with caregiving demands and palliative coping. The 24-item PCRS was later revised to 25 items by separating one item into two for conceptual clarity; the instrument was then further validated on a random sample of 256 Black and White, female and male caregivers (Picot et al., 1997). Using CFA, a revised 16-item scale was generated with a coefficient alpha of .88. The instrument was then proposed as a summative measure of perceived caregiver rewards because of the high correlation (r = .96) between intrinsic and extrinsic rewards, suggesting that

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

October 2005, Vol. 27, No. 6

757

more than 92% of variances was shared by the two factors. Construct validity was supported by associations with caregiving demands, coping, depression, and caregiver burden (Picot et al. 1997). Operational Definitions

Based on Picot’s work, the present study defined rewards as a single construct that reflects positive experiences, including a sense of personal growth, fulfilling payback, control, friendship gains, joy resulting from faith in God’s promises, integrative bonds with other family members, and verbal or nonverbal gratitude from the care recipient, praise from others, and financial rewards. Because the PCRS was used in this study for the first time with Chinese family caregivers, all 25 items were used for psychometric testing as suggested by the original developer (S. J. Picot, personal communication, October 13, 1999). PURPOSE

The purpose of the present study was to (a) translate and adapt the English PCRS for use with Chinese adult children family caregivers in the United States and (b) test the psychometrics of the Chinese version of PCRS using CFA. DESIGN AND SAMPLE

The present study used a cross-sectional, descriptive design with a convenient sample of Chinese adult children family caregivers (n = 137), who assisted an elderly family member (older than 55) with at least three activities of daily living for at least 3 months. The caregivers had to be able to read and write adequate Chinese, either simplified or traditional characters, or English but they could speak any of the major Chinese dialects such as Mandarin or Taiwanese. Sample size was at least five cases per item for a psychometric test (Tinsley & Tinsley, 1987). The Institutional Review Board of a large southwestern university approved all study procedures. The caregivers’ ages ranged from 21 to 70 with a mean of 41 (SD = 10.6), and years of education varied from 9 to 22 with a mean of 16.1 (SD = 2.7). About 89.7% (n = 123) of the caregivers were foreign-born and had been living in

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

758

Western Journal of Nursing Research

the United States from 2 to 58 years with a mean of 19 years. As in other studies (Arai, Zarit, Sugiura, & Washio, 2002; Jones, 1995; Kim & Lee, 2003), the majority of the caregivers were female (n = 96, 70.1%), either daughters (n = 62, 64.6%) or daughters-in-law (n = 34, 35.4%), caring for frail elders with one or more chronic illnesses. Most of the participants (n = 116, 85.4%) had annual incomes in the range of $40,000 to $149,999. METHOD Strategies for Ensuring Cross-Cultural Equivalence

When adapting an instrument from one language to another language, cross-cultural equivalence should include functional and linguistic equivalence (Berry, 1969, 1980, 1992; Phillips et al., 1996). Functional equivalence means that objects, concepts, situations, and behaviors described in the original instrument are applicable to another culture. Linguistic equivalence means that objects, concepts, situations, and behaviors described in the original instrument denote the same information or are semantically equivalent to the translated version. Before selecting an instrument for this study, a thorough investigation of instruments relevant to positive aspects of family caregiving was conducted to assure functional equivalence. This is the first step in applying an instrument developed in one language to a population who use a different language (Berry, 1992; Phillips et al., 1996). This is often a neglected step in the adaptation process. As noted earlier, PCRS, originally developed in English, was selected because it was the only available instrument that provided a comprehensive, theoretically based depiction of positive family caregiving experiences. To achieve linguistic equivalence, traditional techniques were integrated with innovative strategies such as clarifying semantic meaning with native speakers prior to translation and conferring with the original designer after back translation. Traditional techniques include translation, back translation, multilingual technique, and review panel (Brislin, 1970, 1980; Irvine & Carroll, 1980; Phillips et al., 1996). After creating the translated version of PCRS, its psychometrics were evaluated using CFA. CFA is an advanced psychometric technique for testing hypotheses about specific factor structures based on a prior theoretical or empirical basis (Kline, 1998). CFA is appropriate for this study because the study adopted and validated an instrument, PCRS, originally developed in English that has a clear theoretical

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

October 2005, Vol. 27, No. 6

759

basis hypothesizing the factor structure and empirical evidence supporting the factor structure. Figure 1 summarizes the procedures for evaluating linguistic equivalence. Clarifying semantic meaning. Prior to translating the PCRS, the first author clarified the intended meanings of each item in the original PCRS with native English speakers who were Ph.D.-prepared scholars in nursing and knew the caregiving rewards concept well enough to clarify the semantic meaning and context of each item. Translating and back translating. Following clarification, the first author translated the original PCRS from English into Chinese to create the first Chinese PCRS. The first author was bilingual (Chinese or English) and had lived in the United States for 7 years. The principles for instrument development (Brislin, 1980) were used to create simple, easily understood wording and a translation free of colloquialisms. Meaning-based translation (Irvine & Carroll, 1980; Larson, 1998) was applied to preserve the contextual meanings of words between the original and target languages. Modifications in sentence structure were employed in the translated version as needed because thought patterns and syntax differ from language to language (Larson, 1998). This ensured the accuracy of the translation and its fidelity to the intended meanings in the original language (English). After the first Chinese PCRS was generated, two Chinese Americans independently back translated it to English to generate two copies of the second English PCRS. The two back translators were master’s-prepared bilingual scientists who had lived in the United States for nearly 15 years and were fluent in both Chinese and English but not familiar with the original scale. Through an iterative process of translation and back translation, a consensus was reached that generated a final copy of the second English PCRS. Two items—“will be acknowledged in the care recipient’s will” (Q5) and “felt health care providers work harder” (Q7)—were noted as making little sense from a Chinese perspective by the first author and the two back translators. For the Chinese, it would be culturally unacceptable to use wills as a way of rewarding care for dependent elders or to expect health care providers to work harder because providing care for dependent family members is perceived as a family responsibility and a cultural virtue. Establishing content equivalence. After reaching agreement on the back translation, the original developer, Dr. Picot, compared the final copy of the

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

760

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

Figure 1: Procedures for linguistic equivalence. NOTE: PCRS = Picot’s caregiver rewards scale.

Step 1. Clarifying Semantic Meaning

a. Translate the original PCRS into Chinese to generate the first version of Chinese PCRS b. Back translate the first version of Chinese PCRS into English to generate the second version of English PCRS

Compare the second English PCRS with the original PCRS to confirm evidence of linguistic equivalence between the original PCRS and the first Chinese PCRS

Refine wording of the final copy of the first Chinese PCRS to create the second Chinese PCRS

Assess readability of the second Chinese PCRS with six Chinese with various educational backgrounds to create the first Chinese PCRS -- was also called the 25-item Chinese PCRS

Validate the 25-item Chinese PCRS with 137 Chinese adult children family caregivers to determine its psychometrics

Verify the meanings of each item in the original PCRS with two native English speakers

Step 2. Translating / BackTranslating******

Step 3. Establishing Content Equivalence**********

Step 4. Refining the Wording****************************

Step 5. Evaluating Face Validity*********************************

Step 6. Determining Psychometrics**************************************

October 2005, Vol. 27, No. 6

761

second English PCRS with the original PCRS. Dr. Picot agreed that all concepts were equivalent to those in the original PCRS, except for one item. The problematic item was reworded by the first author and re–back translated by another Chinese scientist who was a master’s-prepared bilingual and had lived in the United States for 25 years. Dr. Picot then accepted the revision. The original developer’s agreement provided evidence that each concept in the original PCRS was adequately translated and linguistically equivalent to each concept in the Chinese PCRS. Refining the wording. Two bilingual scientists (Chinese or English), one from mainland China and the other from Taiwan, each edited the wording of the final revised version of the first Chinese PCRS. The scientist from mainland China was a doctorally prepared bilingual and had lived in the United States nearly 7 years. The other scientist from Taiwan was a master’sprepared bilingual and had lived in the United States nearly 5 years. Although major dialects of spoken Chinese are similar or mutually intelligible, only mainland China adopted the simplified characters, and Hong Kong and Taiwan still use the traditional characters when writing in Chinese. Different written phrasing preferences are used in mainland China versus in Taiwan and Hong Kong. Two copies of the second Chinese PCRS were generated: One used simplified characters and easily understood mainland Chinese wording; the other used traditional characters and easily understood Taiwanese wording. Although traditional and simplified written characters have similar shapes and lines, traditional characters have more lines than the simplified characters, but they mean the same thing. This allowed the study participants to fully understand the instrument written in their preferred characters. For people of Chinese descent born in the United States, the original English version of the 25-item PCRS was provided because no matter which Chinese dialect they spoke (Mandarin or Taiwanese), they were usually able to read English. The two copies of the second Chinese PCRS were evaluated using the original PCRS as a reference by a panel of three Chinese reviewers, all of whom held advanced degrees. The panel member from mainland China evaluated the version using the simplified characters. The other two panel members, one from Taiwan and one from Hong Kong, reviewed the version using the traditional characters. According to the three reviewers, all items on the two copies of the second Chinese PCRS were well written, except for the two items that the first author had noted earlier as lacking cultural relevance.

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

Western Journal of Nursing Research

762

Evaluating face validity. Two copies of the second Chinese PCRS were administered to six Chinese persons, three from mainland China and three from Taiwan. Two of the three from each group were adult children caregivers. Group members had educational backgrounds ranging from middle school to graduate school. This process was used to assess the face validity (readability) of the two copies of the second Chinese PCRS to create the final Chinese version. After each participant completed the questionnaire, the first author checked each participant’s comprehension of the meaning of each item. The two copies of the final Chinese PCRS were then confirmed without changes, creating the 25-item Chinese PCRS. Determining psychometrics. The 25-item Chinese PCRS was administered to the study sample along with two scales originally developed in Chinese: the revised Yang’s filial beliefs subscale (YFVS-R) and the revised Yang’s filial affection subscale (YFAS-R; Yang, Yeh, & Huang, 1988). The YFVS-R was used to measure participants’ traditional filial beliefs. The YFAS-R was used to address participants’ filial affection toward parents. According to Holroyd and Mackenzie (1995), caregivers’ cultural values direct the meaning of family caregiving and influence their predisposition to engage in family caregiving. Chinese caregivers who highly valued filial obligation felt satisfied with family caregiving (Jones, 1995; Jones et al., 2002). On evaluation, the concurrent validity of the Chinese PCRS could be inferred from the relationships between the Chinese PCRS and the YFVS-R and VFAS-R. Instruments

The 25-item Chinese PCRS. The 25-item Chinese PCRS was used to measure positive experiences in relation to the family caregiving role, including a sense of personal growth, fulfilling payback, control, friendship gains, joy resulting from faith in God’s promises, integrative bonds with other family members, verbal or nonverbal gratitude from the care recipient, praise from others, and financial rewards. The instrument is a 5-point rating scale with possible responses ranging from 0 = not at all to 4 = a great deal; higher scores indicate greater rewards from family caregiving. The YFVS-R. The YFVS-R is a revised subscale adapted from Yang’s filial obligation scale (YFOS; Yang et al., 1988) that measures participants’ agreement with traditional filial values. The original subscale was designed

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

October 2005, Vol. 27, No. 6

763

from a Chinese perspective to reflect four main attributes of Chinese traditional filial obligation: paying respect to parents, obeying parents, taking care of parents, and protecting parents. The YFVS-R consists of eight items selected from the original scale that are relevant to family caregiving, including attributes of respect, obedience, caretaking, and protection. The eight items were confirmed by a panel of experts consisting of a male Ph.D. scholar, two female Ph.D. scholars, and a male medical doctor. They were all middle-aged, Chinese immigrants who were raised in traditional Chinese society in Taiwan or Hong Kong and knew the attributes of Chinese traditional filial obligations well. The YFVS-R is a 4-point rating scale with possible responses ranging from 0 = not at all to 3 = very much; higher scores indicate stronger agreement with traditional filial values. Internal consistency reliabilities for the original subscale were in the range of .65 to .90 with a sample of more than 10,000 young adults in Taiwan from various backgrounds (Yang et al., 1988). The YFAS-R. The YFAS-R is another revised subscale adapted from Yang’s filial obligation scale (Yang et al., 1988) that measures participants’ affection for their parents. The original YFAS was also designed from a Chinese perspective to reflect two aspects of parental affection, positive and negative. YFAS-R consists of eight items relevant to family caregiving. The criteria used for selecting items and evaluating content validity for the YFVS-R were also used to create the YFAS-R. The YFAS-R is a 4-point rating scale with possible responses ranging from 0 = not at all to 3 = very much; higher scores indicate greater affection toward parents. Internal consistency reliabilities for the original subscale were in the range of .66 to .93 with a sample of more than 10,000 young adults in Taiwan from various backgrounds (Yang et al., 1988). Data Collection

The first author and six representatives recruited potential participants and distributed the survey packages by mail or in person according to participants’ preference. The six recruiting representatives were female, master’sprepared community leaders. They were trained by the first author with a standard data collection plan to assure that (a) every participant understood the purpose of the study, (b) every participant’s questions were answered properly, and (c) every participant would receive an appropriate follow-up call, if needed.

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

Western Journal of Nursing Research

764

DATA ANALYSIS

The Statistical Package for Social Science 12 for Windows was used to examine the descriptive statistics to ensure the accuracy of data entry by identifying and correcting out-of-range values and to describe the magnitude of each variable and correlations. The percentage and distribution of missing data were examined. The beta version of EQS 6.1 and Windows was used for CFA to assess reliability and validity. The one-factor, 25-item Chinese PCRS was an unknown scaling factor with multiple items. For the estimate to be identified, one-factor loading was fixed at 1.0 (Kline, 1998). Both maximum likelihood (ML) and elliptical least square (ELS) were used because ML requires normality of data and ELS is useful for data with unequal variances (Bentler, 1995). Criteria for Psychometric Testing

Because the Chinese PCRS was a new scale, an acceptable standardized alpha of .70 was applied for internal consistency reliability (Edwards, 1970). Item standard deviations are ideally expected to be one half the means (Nunnally & Bernstein, 1994). Larger standard deviations suggests greater sampling bias (Walsh & Betz, 2001). Item-total correlations are required to be from .30 to .70 (Cronbach, 1951; Ferketich, 1991). The statistical significance for rejecting the chi-square test ( 2 [df]) was set at p < .05 typically, but this could be affected by many factors (Byrne, 1994). Other fit indices were reported: goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI), and standardized root mean square residual (SRMR). GFI is the percentage of covariances explained by the covariances implied by the model. It is used to index the ability of the empirical model to account for the patterns in the correlation matrix. AGFI has the same capacity as GFI but adjusts for the number of parameters estimated in the model. SRMR represents the average, standardized residual value derived from the fit between the hypothesized model and the data. The criteria for accepting the model included a value < 3 for relative chi-square ( 2 to df ratio) and a value of .90 for GFI and AGFI (Byrne, 1994; Gefen, Straub, & Boudreau, 2000). Although SRMR < .05 has been suggested for an exact fitting model, a value of less than .1 for SRMR is considered a reasonable fit (Tabachnick & Fidell, 2001). Factor loadings were expected to be in the range of .30 to .70; if < .30, a lack of scale homogeneity is indicated, and if < .70, redundancy is suggested (Ferketich, 1991). Model modification was performed based on statistical evaluation and theoretical considerations. An item was dropped if

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

October 2005, Vol. 27, No. 6

765

it obtained < .30 on the factor loadings, contained large residuals, lacked conceptual clarity, and inadequately depicted the construct of perceived rewards for Chinese adult caregivers. FINDINGS Descriptive Information

There were less than .05% random, missing data for all responses. With such a small amount of missing data, the mean replacement procedure was used for the scale scores (Bollen, 1989). Table 1 displays the item means, item standard deviations, and skewness. The item mean was 1.59 (SD = 1.22) with scores ranging from .26 to 2.53 on a 5-point scale of 0 to 4. The results showed large discrepancies between and within items similar to those of the original English PCRS (Picot et al., 1997). The distribution of responses for the items showed wide variations in skewness, ranging from –.07 to 3.12. For two items, Q5 and Q7, which made little sense from a Chinese perspective, item standard deviations were either almost twice or more than twice their item means. The same problem appeared with another two items, Q19 and Q25. In addition, Q5 and Q25 showed severe skewness, as indicated by an absolute value of the skew index being almost 3 or greater than 3. These four items—Q5, Q7, Q19, and Q25—were considered potentially unfavorable items for psychometric evaluation. Psychometrics Estimate

The one-factor, 25-item Chinese PCRS obtained a standardized alpha of .89. The criterion for a reliable scale was achieved by the YFVS-R with a standardized alpha of .89 and by the YFAS-R with a standardized alpha of .81 as well. The one-factor, 25-item model failed on the ML method but presented fairly well on the ELS method, as indicated by GFI = .90, AGFI = .90, SRMR = .11, and relative chi-square ( 2 to df ratio) = .29. The factor loadings for Q5, Q7, Q19 and Q25 were below .30. Also, Q5, Q19, and Q25 failed to meet the minimum criterion for interitem correlations. Another two items, Q14 and Q15, showed conflicting results; the item-total correlation between them was .88, indicating redundancy. Factor loadings were adequate with .71 for Q14 and .70 for Q15. A one-factor, 21-item revised model was proposed by deleting Q5, Q7, Q19, and Q25. These items performed poorly in descriptive statistics and

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

Western Journal of Nursing Research

766

TABLE 1: Mean, Standard Deviation, Skewness, and Reliability and Factor Loadings of the 25 Item and 21 Item Chinese PCRS (n = 137)

25-Item Chinese PCRS a = .89 Item

M

SD

Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 Q13 Q14 Q15 Q16 Q17 Q18 Q19 Q20 Q21 Q22 Q23 Q24 Q25

1.68 1.72 2.10 1.61 0.29 1.37 1.45 2.01 1.21 2.11 2.53 2.20 1.37 1.93 1.93 1.47 1.80 2.11 0.73 0.86 2.14 1.82 1.82 1.32 0.26

1.47 1.27 1.21 1.31 0.78 1.19 1.25 1.48 1.09 1.29 1.24 1.37 1.37 1.28 1.31 1.25 1.25 1.29 1.14 1.08 1.18 1.23 1.31 1.19 0.71

21-Item Chinese PCRS a = .90

Interitem Item Total Factor Item Total Factor Skewness Correlation Correlation Loading Correlation Loading 0.27 0.20 –0.10 0.20 2.86 0.36 0.32 –0.09 0.57 –0.09 –0.48 –0.23 0.60 0.01 –0.08 0.35 0.06 –0.18 1.52 1.03 –0.33 0.09 0.13 0.56 3.12

Y Y Y Y (< .30) Y (< .30) Y Y Y Y Y Y (> .70) (> .70) Y Y Y (< .30) Y Y Y Y Y (< .30)

.42 .61 .62 .53 .15 .41 .29 .32 .44 .68 .53 .41 .45 .71 .70 .46 .50 .41 .22 .43 .41 .53 .60 .56 .20

.49 .68 .69 .58 .16 .45 .29 .37 .48 .74 .61 .47 .49 .78 .76 .51 .55 .47 .24 .45 .47 .60 .67 .61 .21

.42 .62 .62 .52 D .39 D .33 .43 .70 .56 .42 .44 .72 .71 .46 .50 .41 D .37 .40 .54 .61 .54 D

.49 .69 .69 .58 D .44 D .38 .47 .75 .63 .49 .48 .78 .77 .51 .55 .48 D .42 .46 .61 .68 .60 D

NOTE: = Standardized Cronbach’s alpha for internal consistence reliability. D = Item dropped; Y = Yes, interitem correlation was in the range of .30 to .70; PCRS = Picot’s caregiver rewards scale.

consistently failed in reliability and validity estimates and the ability to measure the construct of perceived rewards. Q14 and Q15 were kept because they were important items for depicting personal growth as a positive caregiving experience (Picot et al., 1997). They performed well on itemtotal correlations and the factor loading analysis. The one-factor, 21-item revised Chinese PCRS was reevaluated with the ELS and closely fitted with the data, as indicated by GFI = .94, AGFI = .93, relative chi-square ratio ( 2

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

October 2005, Vol. 27, No. 6

767

to df) = .27, and SMRM = .09. The chi-square test ( 2(189), n = 137) = 541 was lower but it was still rejected at p < .05. Because chi-square is affected by many factors (Byrne, 1994), instead of regarding chi-square as a test statistic, it was regarded as a GFI in the sense that large 2 values correspond to a bad fit and small 2 values to a good fit. Reliability estimate. Table 1 presents the means, standard deviations, skewness, reliability, and factor loadings of the 25-item and 21-item Chinese PCRS. After reducing the number of items from 25 to 21, the standardized alpha increased from .89 to .90. The average mean, on a scale of 0 to 4, improved from 1.59 (SD = 1.22) with scores ranging from .26 to 2.53 to 1.77 (SD = 1.27), with scores ranging from .86 to 2.53. The overall variation between and within items improved. All 21 items had interitem correlations in the range of .30 to .70, except that the item-total correlation between Q14 and Q15 was > .70. Validity estimate. All factor loadings were significant (p < .00) and above .30 (see Table 1). Construct validity was supported with relatively weak, yet still significant correlations between filial beliefs (r = .32, p < .01) and filial affection (r = .23, p < .05). DISCUSSION

The validation of the Chinese PCRS was based on the principles and procedures suggested by Berry (1992), Brislin (1970), Irvine and Carroll (1980), and Phillips et al. (1996). Four items—Q5, Q7, Q19, and Q25— were deleted, leaving a total of 21 items. As with the original English PCRS (Picot et al., 1997), improvement was noted after item reduction by obtaining a standardized alpha of .90 and adequate factor loadings. Construct validity of the one-factor, revised 21-item Chinese PCRS was supported by CFA to be a single dimensional measure. Results support positive experiences in the theoretical construct of the English PCRS, including a sense of personal growth, fulfilling payback, control, friendship gains, joy resulting from faith in God’s promises, integrative bonds with other family members, verbal or nonverbal gratitude from the care recipient, praise from others, and financial rewards. The four items that were deleted had obvious problems. Two items that the first author and bilingual scientists were concerned about—“will be acknowledged in the care recipient’s will” (Q5) and “felt health care

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

768

Western Journal of Nursing Research

providers work harder” (Q7)—seemed inadequate and culturally irrelevant for explaining perceived rewards in the context of Chinese culture. As noted earlier, providing care for dependent family members is a filial responsibility as well as a cultural virtue for the Chinese; thus, using wills and expecting health care providers to work harder would not be a common way of rewarding caregivers of dependent elders. The other two items—Q19 and Q25—were considered less than adequate for reflecting the positive experiences of Chinese caregivers because they referred to socially desirable or undesirable behaviors or values that might introduce a social desirability bias (SDB; Crowne & Marlowe, 1960). SDB is the tendency to give an answer that makes the respondent look good or shows that the respondent has respectable characteristics (Paulhus, 1991). SDB was a major concern in the present study because of the influence of traditional Chinese culture on Chinese caregivers. The Chinese customarily viewed family caregiving as a familial expectation, a moral obligation, and a cultural virtue (Ho, 1994). In traditional Chinese culture, providing support to a dependent elder is valued as a socially desirable behavior, whereas materialism is viewed as a negative, socially undesirable characteristic (Yau, 1994). It would not be unusual for Chinese caregivers to hide their true feelings when answering questions related to materialism, such as Q19 “do not need to hold a job” and Q25 “felt financial gain from providing care.” As with the original study (Picot et al., 1997), the four items removed from the Chinese PCRS were also dropped from the English PCRS. In contrast, five other items—Q8, Q9, Q12, Q13, and Q21—which were not retained in the English PCRS, performed fairly well with the present study sample. Social desirability bias was the main concern for some of the failing items such as Q5, Q13, Q19, and Q25. Sampling bias—including gender, income, education, and racial heterogeneity—was the concern for the rest of the failing items (Picot et al., 1997). The two personal growth items—Q14 and Q15, which failed but were kept in the original study because they depicted personal growth as a positive experience—performed well in the present study also. The study results are consistent with the major themes identified in qualitative work with Chinese caregivers, including personal growth, fulfilling payback, close bonds with other family members, and joy resulting from faith in God’s promises (Jones, 1995; Jones et al., 2002; Shyu et al., 1998). The study’s limitations included using a convenience sample of Chinese adult children caregivers who were predominantly foreign-born, well educated, and with moderate to high incomes. In summary, the one-factor, revised 21-item Chinese PCRS appears to be a simple, culturally adequate

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

October 2005, Vol. 27, No. 6

769

tool that is useful for assessing the positive experiences of Chinese adult children family caregivers. It could be useful for understanding cultural influences on caregivers’ perceived rewards and help identify caregivers who do not perceive rewards. This assessment may serve as a foundation for developing or tailoring interventions to reduce perceived burden among Chinese caregivers. NOTE 1. The first author developed and implemented the study as supervised by the second author. During the implementation stage, the third and fourth authors assisted with cross-cultural validation. In the development of the manuscript, the second author offered scientific critique and the third author participated in literature review. Copies of the complete 25 items of the original Picot caregiver rewards scale may be obtained from Dr. Picot. Below is the contact information: Sandra J. Fulton Picot, Ph.D., R.N., F.A.A.N., Associate Professor, Department of Adult Health Nursing, University of Maryland, 655 West Lombard St., 3rd Floor, Baltimore, MD 21201, Tel: 410-706-4434, Fax: 410-706-0344, e-mail: [email protected].

REFERENCES Arai, Y., Zarit, S. H., Sugiura, M., & Washio, M. (2002). Patterns of outcome of caregiving for the impaired elderly: A longitudinal study in rural Japan. Aging & Mental Health, 6, 39-46. Bentler, P. M. (1995). EQS Structural equations program manual. Encino, CA: Multivariate Software. Berry, J. (1969). On cross-cultural comparability. International Journal of Psychology, 4, 119128. Berry, J. (1980). Introduction to methodology. In H. D. Triandis & J. W. Berry (Eds.), Handbook of cross-cultural psychology (pp. 1-28). Boston: Allyn & Bacon. Berry, J. (1992). Methodological concerns. In J. Berry, Y. Pootinga, M. Segal, & R. Dasen (Eds.), Cross-cultural psychology (pp. 210-246). New York: Cambridge University Press. Bollen, K. A. (1989). Structural equations with latent variables. New York: John Wiley. Brislin, R. W. (1970). Back-translation for cross-cultural research. Journal of Cross-Cultural Psychology, 2, 185-216. Brislin, R. (Ed.). (1980). Translation: Applications and research. New York: Wiley-Halsted. Byrne, B. M. (1994). Structural equation modeling with EQS and EQS/Windows: Basic concepts, applications, and programming. Thousand Oaks, CA: Sage. Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 274-334. Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24, 349-354. Edwards, A. L. (1970). The measurement of personality traits by scales and inventories. New York: Holt.

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

770

Western Journal of Nursing Research

Ferketich, S. (1991). Focus on psychometrics: Aspects of item analysis. Research in Nursing & Health, 14, 165-168. Gefen, D., Straub, D. W., & Boudreau, M. C. (2000). Structural equation modeling and regression: Guidelines for research practice. Communications of The Association for Information Systems, 7(7), 1-78. Ho, D. Y. F. (1994). Filial piety, authoritarian moralism, and cognitive conservatism in Chinese society. Genetic, Social, & General Psychology Monographs, 120, 349-365. Holroyd, E. A., & Mackenzie, A. E. (1995). A review of the historical and social processes contributing to care and caregiving in Chinese families. Journal of Advanced Nursing, 22, 473479. Irvine, S. H., & Carroll, W. K. (1980). Testing and assessment across cultures: Issues in methodology and theory. In H. D. Triandis & J. W. Berry (Eds.), Handbook of cross-cultural psychology: Methodology (pp. 181-244). Boston: Allyn & Bacon. Jones, P. S. (1995). Paying respect: Care of elderly parents by Chinese and Filipino American women. Health Care for Women International, 16, 385-398. Jones, P. S., Zhang, X. E., Jaceldo-Siegl, K., & Meleis, A. I. (2002). Caregiving between two cultures: An integrative experience. Journal of Transcultural Nursing, 13, 202-209. Kim, J., & Lee, E. (2003). Cultural and noncultural predictors of health outcomes in Korean daughter and daughter-in-law caregivers. Public Health Nursing, 20, 111-119. Kline, R. B. (1998). Principles and practice of structural equation modeling. New York: Guilford. Kramer, B. J. (1997). Gain in the caregiving experience: Where are we? What next? The Gerontologist, 37, 218-232. Larson, M. L. (1998). Meaning-based translation: A guide to cross-language equivalence. Lanham, MD: University Press of America and Summer Institute of Linguistics. Lynn, M. R. (1986). The determination and quantification of content validity. Nursing Research, 35, 382-385. Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York: McGrawHill. Nye, F. I. (1978). Is choice and exchange theory the key? Journal of Marriage and the Family, 40, 219-233. Paulhus, D. L. (1991). Measurement and control of response bias. In J. P. Robinson, P. R. Shaver, & L. S. Wrightsman (Eds.), Measurement of personality and social psychological attitudes (pp. 17-59). San Diego, CA: Academic Press. Phillips, L. R., Luna, I., Russell, C. K., Baca, G., Lim, Y. M., Cromwell, S. L., et al. (1996). Toward a cross-cultural perspective of family caregiving. Western Journal of Nursing Research, 18, 236-251. Picot, S. J. F. (1995). Choice and social exchange theory and the rewards of African American caregivers. Journal of National Black Nurse Association, 7, 29-40. Picot, S. J. F., Youngblut, J., & Zeller, R. (1997). Development and testing of a measure of perceived caregiver rewards in adults. Journal of Nursing Measurement, 5, 33-52. Shyu, Y. L., Archbold, P. G., & Imle, M. (1998). Finding a balance point: A process central to understanding family caregiving in Taiwanese families. Research in Nursing & Health, 21, 261-270. Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). Boston: Allyn & Bacon. Thibaut, J., & Kelly, H. (1959). The social psychology of groups. New York: John Wiley. Tinsley, H. E. A., & Tinsley, D. J. (1987). Use of factor analysis in counseling psychology research. Journal of Counseling Psychology, 44, 414-424.

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010

October 2005, Vol. 27, No. 6

771

Yamamoto-Mitani, N., Ishigaki, K., Kuniyoshi, M., Kawahara-Maekawa, N., Hayashi, K., Hasegawa, K., et al. (2004). Subjective quality of life and positive appraisal of care among Japanese family caregivers of older adults. Quality of Life Research, 13, 207-221. Yang, K. S., Yeh, K. H., & Huang, L. (1988). A social attitudinal analysis of Chinese filial piety: Concept and assessment. Bulletin of the Institute of Ethnology, Academia Sinica, 65, 117227. Yau, O. H. M. (1994). Consumer behavior in China: Consumer satisfaction and cultural value. London: Routledge. Walsh, W. B., & Betz, N. E. (2001). Test and assessment (4th ed.). Englewood Cliffs, NJ: Prentice Hall.

Downloaded from http://wjn.sagepub.com at UNIV OF MISSOURI ST LOUIS on June 30, 2010