Preliminary Evidence for the Validity of the Family Caregiver Identity Scale

Abstract

The purpose of this study was to develop and evaluate the Family Caregiver Identity Scale (FCIS), an instrument designed to measure the extent to which an individual identifies with the family caregiver role. The process of instrument development outlined in the Standards for Educational and Psychological Testing was combined with Dillman’s four stages of pretesting. This was a multistage, iterative process, including several revisions based on feedback from experts, interviews, and pilot testing. Factor analyses were performed to test the hypothesized model of caregiver identity. A version of the FCIS consisting of 18 items was created and demonstrated initial evidence of validity. The FCIS will enable gerontological professionals to assess caregiver identity. The absence of caregiver identity is a factor in caregivers not accessing support services. This study contributes to the growing body of research connecting caregiver identity and support service utilization by caregivers.

Keywords

caregiving theory measurement factor analysis

Introduction

The term family caregiver refers to an individual providing unpaid care to an adult family member, partner, or friend with a chronic, disabling, or serious health condition (American Association of Retired Persons [AARP], 2015). The transition into the family caregiver role is often a subtle and gradual evolution of an existing relationship such as spouse or child. In this process, the person providing care becomes more than a spouse (wife/husband) or child (daughter/son) but rather evolves into a “caregiver” as well. For example, a daughter does not typically assist her father with toileting or bathing but may if she becomes his primary caregiver. The caregiver role, and the responsibilities that come with it, may not be welcomed by the person providing care. She may not identify with or willingly accept the role of “caregiver” (Dobrof & Ebenstein, 2003), even though she is performing the tasks typically associated with caregiving such as assisting with bathing, grooming, toileting, dressing, medication intake, mobility, and more. Beyond the work of Montgomery and colleagues (Montgomery & Kosloski, 2009, 2012), research is limited on the relationship between the transition into caregiving and caregiver identity development. Caregiver identity is the “self-understanding, self-objectification, or integration of information about the self” (Holland, 1997) as it relates to the caregiver role.

The AARP (2001) Caregiver Identification Study surveyed 4,037 adults to understand perception and identification of family members with the term “caregiver” as well as to establish prevalence of caregiving and care-related activities. Participants were asked two, dichotomous response questions to assess whether they were caregivers or not. The first question was straightforward—“Are you a caregiver?” The second question did not use the word “caregiver.” Instead, it provided a description of activities normally performed by caregivers—“Do you currently provide or have you provided in the last year unpaid help to a relative or friend who has a disability or chronic disease? This kind of help includes assistance with health or personal needs or household chores” (AARP, 2001). Interestingly, of the AARP sample, 15% of those who did not consider themselves caregivers in the first question acknowledged providing the kind of care described in the second question. From this study and others (O’Connor, 1999; Pruchno et al., 2008), there appears to be a disconnect between an individual self-identifying as a caregiver and recognizing that they are providing the type of care associated with being a caregiver.

Caregiving is not a role always embraced by all who occupy it. Caregiving is stigmatized (Werner et al., 2010), perhaps because of the attention given to the negative aspects of the caregiving experience in research and the media (Cohen et al., 2002; Horrell et al., 2015). Furthermore, many services designed to support caregivers focus solely on reducing the difficulties and fail to emphasize the positive aspects of the role (Carbonneau et al., 2010). This perpetuates the idea that caregiving is undesirable and not something which a person wants to be associated with. In addition, some cultures believe in filial piety, which demands providing care for one’s parents as a moral mandate. Filial piety does not consider the ability or interest of the child in actually providing care to parents (Wakui & Cheng, 2016).

In contrast, for other caregivers, “caregiver” is more than a functional or technical term but rather a deep-rooted aspect of themselves that is accepted. This distinction is significant because of the relationship between role identity and behavior (Biddle et al., 1985). Bronfenbrenner (1979) stated that “roles have a magiclike power to alter how a person is treated, how she acts, what she does, and thereby even what she thinks and feels” (p. 6). Several studies have demonstrated a positive association between caregiver identity—typically measured by the question, “Are you a caregiver?”—and the use of services such as health education, counseling, support groups, or personal care services (AARP, 2001; Henderson, 2001; O’Connor, 1999). In general, individuals who do not identify with the caregiver role will tend not to seek out or utilize support services developed to support caregivers. These services can mitigate the stress associated with caregiving that often lead to poor health outcomes and reduced life expectancy for the caregiver (Hong & Cadado, 2015; Lopez-Hartmann et al., 2012; Mason et al., 2007).

Because of the relationship between caregiver identity and support service utilization, a way to measure caregiver identity is needed, and presently, such an instrument does not exist. From a theoretical point of view, a theory-based, psychometrically sound scale designed to measure caregiver identity would contribute to the understanding of this construct. With regard to real-world application, determining where a caregiver is in the development of caregiver identity could have profound effects on a person’s ability to serve and support family caregivers. According to the most recent Caregiving in the U.S. report (National Alliance for Caregiving & AARP Public Policy Institute, 2015), 34.2 million Americans provide unpaid care to an adult. Using the percentage of caregivers who provide care but do not identify as caregivers from the AARP study (15%), we can estimate that there are 5.1 million caregivers who do not acknowledge or accept that they are indeed caregivers. Our contention is that by helping people who provide care to develop and embrace their caregiver identity, support service utilization by caregivers would increase and overall health and well-being of the caregiver would be improved. If caregiver identity continues to be rejected by those who technically occupy the role, “caregiving may continue to be hidden and potential sources of help and support may remain underutilized” (Hughes et al., 2013, p. 84).

Purpose

To address the gap in measuring caregiver identity, the purpose of the present study was to develop and evaluate the Family Caregiver Identity Scale (FCIS), an instrument designed to measure the extent to which an individual identifies with the family caregiver role. The first study aim was to create a pool of items to measure the theoretical domains of caregiver identity. The second study aim was to test these items to evaluate their factor structure. The study was approved by the University of North Carolina at Greensboro (Institutional Review Board [IRB] #14-0055).

Methods and Results

Study Design

A seven-step process was implemented to develop and evaluate the FCIS. This process was informed by a theory-driven systematic literature review to identify factors related to the development of family caregiver identity (Eifert et al., 2015). From the results of the systematic literature review, several themes related to the development of family caregiver identity emerged: (a) Role Engulfment and Reversal, (b) Loss of Shared Identity, (c) Family Obligation and Gender Norming, (d) Extension of the Former Role, and (e) Development of a Master Identity. A theoretical framework was created to explain the key factors, concepts, or variables related to caregiver identity (Figure 1). The theoretical framework informed the development of the FCIS and provided the basis for the domains to be measured.

Figure 1.

Theoretical framework for caregiver identity development.

The instrument development framework (Figure 2) used to develop and evaluate the FCIS was guided by the Standards for Educational and Psychological Testing or the Standards (American Educational Research Association [AERA], American Psychological Association [APA], and National Council on Measurement in Education [NCME], 1999) and Dillman’s (2000) stages of pretesting. The Standards guided the development of the purpose of the instrument (Step 1), test specifications (Step 2), and a pool of questions (Step 3). The questions resulting from Steps 1 to 3 of the study were pretested using Dillman’s (2000) process, including review by knowledgeable colleagues and analysts (Step 4), interviews to evaluate cognitive and motivational qualities (Step 5), a small pilot study (Step 6), and final review (Step 7). Because the study design included sequential steps where each step was based on the results of the prior step, the methods, analyses, and results are reported in combination.

Figure 2.

Instrument development framework.

Step 1: Purpose of instrument

The first step was to describe “the extent of the domains, or the scope of the construct(s) to be measured” (AERA, APA, & NCME, 1999, p. 37). The purpose of the instrument was to measure the extent to which an individual identifies with the family caregiver role based on five domains: Role Engulfment and Reversal, Loss of Shared Identity, Family Obligation and Gender Norming, Extension of the Former Role, and Development of a Master Identity. These domains provide the theoretical framework for the FCIS content and the basis for question development.

Step 2: Test specifications

The second step was to identify test specifications such as the format of questions and responses as well as scoring (AERA, APA, & NCME, 1999). Due to the subjective nature of identity (Fowler, 1995), the best measures were determined to be statements with response options on a continuum from “completely agree” to “completely disagree.” To ensure issues of fairness (AERA, APA, & NCME, 1999), questions were worded in such a way to lessen the likelihood of predictable differences in measures and outcomes for respondents due to race, ethnicity, gender, or other characteristics (Furr & Bacharach, 2014). In addition, issues of bias (AERA, APA, & NCME, 1999) as a result of improper or inapplicable content were addressed by engaging a panel of experts to review questions for content. In a later step, a panel of experts was assembled to review instrument questions for issues of fairness and bias.

Step 3: Pool of items

In total, 47 questions were created. Questions that measure the same domain were categorized together to create subscales, and a table of specifications was created (Fives & DiDonato-Barnes, 2013). A large, initial pool of questions was intentionally generated to address all of the domains with the knowledge that some questions would not survive the pretesting stage.

Step 4: Review by knowledgeable colleagues and analysts

Method

A panel of experts was asked to review the 47 questions. The panel consisted of eight professionals across the United States and Canada. Efforts were made to include, as panel members, scholars in the area of caregiver identity and general experts in gerontology. In addition, experts in survey and instrument development were asked to participate. Experts were given an outline of the instrument content that included the purpose of the FCIS, theoretical framework, descriptions of each domain to be measured, and plan for scoring as well as a link to a Qualtrics (Provo, UT, USA) survey which prompted them to rate each of the 47 questions on criteria of precision, strength, or clarity on a 5-point Likert-type scale from “very poor” to “very good.” They were also asked to indicate whether the questions should be included in the final instrument (“yes” or “no”) and were provided a text box for any feedback, suggestions, or comments on a given question. In addition, experts were given the opportunity to provide suggestions related to formatting and scoring of the FCIS.

Analysis

Data from the Qualtrics survey were analyzed with SPSS (SPSS Inc., 2009). The process for deleting a question involved a systematic examination of the responses. A mean score for each question was calculated. First, the binary responses of “should this item be included” (yes/no) were examined for each question. Questions which scored 40% or lower were excluded from further evaluation. For those questions which were retained, the ratings of precision, strength, and clarity were examined. In a second pass, questions were further considered for inclusion if the mean value of the expert ratings were 4.00 or above (“good” or “very good”) on any of the three criteria (precision, strength, or clarity) scales. If the mean value of the expert ratings fell below 4.00 (“very poor,” “poor,” or “fair”), the qualitative comments from the reviewers were assessed. If the question could be improved based on the feedback from the reviewers, the question was modified and included in the next version of the instrument. If the issues identified in the feedback could not be addressed or extensive modification of the question was required to adequately address the issues identified in the feedback, the question was removed.

Results

Based on survey responses and feedback, the pool of 47 questions was reduced to 21. Upon compilation and review of the 21 questions marked for pretesting, it was determined that key dimensions of two domains, Family Obligation and Gender Norming and Extension of Former Role, were not being adequately measured due to the elimination of certain questions. To address this issue, deleted questions were reexamined. One of the questions from Family Obligation and Gender Norming was rewritten based on the panel’s feedback and added back to the pool, increasing the total number of questions to 22. Two new questions were created—one for Family Obligation and Gender Norming and the other for Extension of a Former Role. These questions were based on deleted questions and feedback from the panel to fill the gap in the pool of potential questions. These additions raised the pool to 24 questions. Finally, one of the questions from Extension of Former Role was removed because it measured the wrong theoretical construct and was subsequently added back to the pool under Loss of Shared Identity. The final pool recommended by the experts resulted in a 25-question instrument, including five questions to measure the domain Role Engulfment and Reversal, five questions for Loss of Shared Identity, five questions for Family Obligation and Gender Norming, five questions for Extension of the Former Role, and five questions for Development of a Master Identity.

The majority of the panel agreed that the proposed format of the FCIS was appropriate, scoring of the FCIS was effective, and weighing of constructs equally was appropriate. In the final comments section, one expert recommended the removal of the “neutral/uncertain” option, changing the Likert-type scale from 5 points to 4 points. Due to debate about the usefulness of a midpoint (Garland, 1991; Kulas et al., 2008; Weems & Onwuegbuzie, 2001), this change was made to the instrument. In addition, feedback from the panel suggested that participants may have difficulty with certain phrases like “loved one” or “caring for.” These phrases were changed to “family member” and “providing care,” respectively, and a description of the terms was added to the beginning of the instrument.

Step 5: Interviews with family caregivers

Method

Step 5 involved conducting cognitive interviews (Willis, 2005), with 10 self-identified family caregivers to detect potential problems with individual questions and the overall instrument. Caregivers were given a preliminary version of the FCIS and asked to complete it. Caregivers were instructed to read each question of the FCIS out loud and then to verbalize their thoughts when responding. Concurrently, the interviewer asked questions designed to elicit further information from the caregiver. The purpose of the verbal probing was to find out more about a particular aspect of the question–response process. All interviews were audio-recorded. In addition, the interviewer took notes to record quotes and issues as the caregiver was talking.

Analysis

Recordings and notes were reviewed and summarized by question. Under each question were verbatim quotes of respondent’s interpretation of the question and any other relevant comments made regarding the question. In addition, pertinent interviewer notes were listed. Analysis entailed examining the summarized data for each question in its entirety. The information was used to assess individual questions and the instrument as well as a guide for decisions about keeping, deleting, or modifying questions.

Results

The cognitive interviews led to 25 changes being made to the instrument. The majority of changes were minor, but two questions were significantly altered due to poor word choice. Question 8 originally contained the word “obligated,” and Question 9 contained the word “responsibility.” The majority of interviewed caregivers expressed displeasure or unease with these words. Question 8 was revised from “I am obligated . . .” with the phrase “Others expect me . . . ,” and Question 9 was revised to “I am responsible . . .” from “It is my responsibility. . .” Finally, all interviewed caregivers completely disagreed with the question related to reminiscing with their ill family member. This outcome suggested that the question was not appropriate or did not capture the domain as intended. This question was flagged but kept in the final pilot study pool to confirm its unsuitability. A final instrument of 25 questions, including five questions for each of the five domains of caregiver identity development, resulted.

Step 6: Pilot study of the FCIS

Method

A pilot study to test the FCIS was conducted in partnership with local organizations. A variety of recruitment strategies were utilized, including flyers, face-to-face solicitation at appropriate events or services, and listservs. In addition to utilizing local organizations, snowball sampling was conducted via a variety of social media outlets. Potential participants were asked, “Are you a caregiver?” and “Do you currently provide or have you provided in the last year unpaid help to a relative or friend who has a disability or chronic disease?” This kind of help includes assistance with health or personal needs or household chores.” Respondents were included in the study if they answered “yes” to one or both of the questions. All participants were informed about the requirements, and the risks and benefits of participating.

Analysis

Descriptive statistics using SPSS were calculated on demographic variables. Confirmatory factor analysis (CFA) using Mplus was performed to evaluate construct validity and examine the fit of the observed data to the proposed factor structure. For analyses, robust-weighted least-squares means-and-variance adjusted (WLSMV) was used as the estimation method. WLSMV is only available in the Mplus program and has the best options for handling categorical data (Byrne, 2012; Muthén & Muthén, 2004). Given the Likert-type format of items and the relatively small number of scale points, the data were treated as categorical, rather than as continuous (Flora & Curran, 2004). This approach was implemented to address certain underlying assumptions regarding normal distribution and equal distance between scale points. The strategy was to first analyze the original items in a CFA in an attempt to replicate the conceptual model. If a reasonably parsimonious model with all 25 items was not obtained, additional analyses were planned using exploratory factor analysis (EFA; Netemeyer et al., 2003).

Results

Table 1 provides the demographic characteristics of the sample. The sample consisted of 512 participants predominately Caucasian (77%), female (77%), over the age of 45 (74%), and North Carolinians (100%). Approximately two thirds of the sample (68%) identified themselves as being a caregiver in the past 12 months.

Table 1.

Characteristics of Pilot Study Participants (N = 512).

Characteristic	n (%)
Gender
Female	396 (77.3)
Male	99 (19.3)
Missing	17 (3.3)
Age
Below 30	23 (4.4)
30–44	92 (17.9)
45–64	226 (44.1)
65 or older	155 (30.2)
Missing	16 (3.1)
Race/ethnicity
Asian or Pacific Islander	9 (1.7)
Black/African American	66 (12.8)
Caucasian/White	396 (77.3)
Latino or Hispanic	11 (2.1)
Native American	5 (<1)
Other	4 (<1)
Missing	21 (4.1)
Caregiver status (self-identified)^a
Yes	346 (67.5)
No	149 (29.1)
Missing	17 (3.3)

Participant answered “yes” to “Are you a caregiver?”

Factor analysis utilizes more than one index to determine the significance of the analysis and to determine the adequacy of the model fit to the data (Brown, 2006). Therefore, a range of goodness-of-fit statistics were examined, including chi-square (χ²) > 0.05 (Thompson, 2004), comparative fit index (CFI) and Tucker–Lewis index (TLI) ≥ 0.90 (Russell, 2002), root mean square error of approximation (RMSEA) ≤ 0.08 (MacCallum et al., 1996), and standardized root mean square residual (SRMR) ≤ 0.08 (Browne & Cudeck, 1993). In addition, factor loadings were studied to determine the strength of the association between the latent variable and the observed variable. The process of using factor analysis to estimate construct validity as well as an item-reduction technique to reduce overlapping or under-performing measures is described below.

Initial model

A review of the hypothesized model produced a variety of concerns (Table 2). First, the overall model fit attributes did not meet common standards (Browne & Cudeck, 1993; Hu & Bentler, 1999). The χ² value (1,403, p < .001) was large enough to reject the null hypothesis of a good fit. CFI and TLI values were 0.57 and 0.81, respectively. In addition, the RMSEA was 0.23, and the SRMR was 0.15. Second, several of the question loadings were low (<0.20) or negative, suggesting that the question provided little contribution to the instrument, potentially leading to model misspecification. Therefore, review of the CFA for the initially proposed, five-factor model led to the conclusion that (a) questions should be removed from the instrument and (b) an EFA of the revised structure should be conducted to better assess the factors that the instrument was capturing.

Table 2.

Mplus Output and Fit Indices of Confirmatory Factor Analysis for Initial Model (Model 1).

Factor and item	Loading	SE	Estimates/SE
Role Engulfment
1. I am the only person capable of providing care to my ill family member.	0.00	0.00	0.00
2. Because of providing care to my ill family member, I do not have time to do things that I would normally do.	0.07	19.82	0.07
3. I often feel isolated because I am providing care to my ill family member.	0.07	20.00	0.07
4. There are activities that I used to do that I no longer have time for because of providing care to my ill family member.	0.08	18.39	0.08
5. I see myself more as a care provider than as a child/spouse/other.	0.07	19.41	0.07
Loss of Shared Identity
1. I feel less like a daughter/wife/husband/son/other because of providing care to my ill family member.	1.00	0.00	0.00
2. The personal relationship I have with my family member will never be the same as it was before I began providing care to him or her.	1.07	0.04	27.37
3. Because I began to provide care to my ill family member, the dynamics or how we interact with each other has changed.	1.10	0.05	23.51
4. My ill family member and I no longer reminisce together or talk about the past.	0.86	0.05	18.82
5. I wish my life would go back to the way it was before my family member became ill.	1.02	0.04	24.81
Family Obligation
1. Providing care to others is part of a woman’s role.	1.00	0.00	0.00
2. Others expect me to provide care to my ill family member.	1.47	0.09	16.70
3. People in my family think that a woman, not a man, should provide care to an ill member of the family.	1.15	0.06	19.95
4. It is the responsibility of the family to provide care for ill members.	−0.48	0.08	−6.18
5. Being a part of a family means taking care of each other.	0.05	0.09	0.60
Extension of Former Role
1. Being a daughter/wife/husband/son/other means providing care to my family member if he or she becomes ill.	1.00	0.00	0.00
2. I am responsible for providing care to my ill family member because I am his or her daughter/wife/husband/son/other.	5.21	1.38	3.77
3. Part of my role as daughter/wife/husband/son/other includes providing care to a family member if he or she becomes ill.	1.38	0.32	4.31
4. Being a good child includes caring for elderly parents who can no longer care for themselves.	0.63	0.21	3.03
5. Providing care to my ill family member is a natural extension of my role as child/spouse/other.	4.54	1.21	3.75
Master Identity
1. Being a caregiver is a significant part of who I am.	1.00	0.00	0.00
2. Providing care is only a portion of my life.	0.25	0.06	3.94
3. Others identify me as a caregiver.	1.10	0.04	26.60
4. Being a caregiver is the most important role I have.	1.01	0.04	25.84
5. My ill family member would consider me his or her caregiver.	1.08	0.04	25.12

Note. Model fit indices: χ² = 1,403, df = 57, p = .00; CFI = 0.57; TLI = 0.81; RMSEA = 0.23; SRMR = 0.15. df = degrees of freedom; CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.

Revisions and final model

To improve the strength of the model, individual parameter estimates were examined, and alternative models were tested until a parsimonious factor structure was obtained. First, the model was identified, and latent variables were scaled by setting one of its factor loadings to 1 (Bollen, 1989). For revised models, the unit of measurement was set to the indicator variable with the highest estimate. For example, all questions for Factor 1 (Role Engulfment) were measured against RolEng4 due to its high estimate. Second, several questions were deleted from the model. FamOblig4 was removed due to its negative value (Byrne, 2012). Several other questions (RolEng1, LosShrd4, MastID5) were dropped because they fell below a 0.80 cutoff level of acceptability established by the researcher (Fornell & Larcker, 1981). Several questions for Factor 3 (Family Obligation) also fell below this criterion, but due to the factor needing at least three questions to adequately capture the construct (Straub et al., 2004), no other questions were removed. Finally, Factor 4 (Extension of Former Role) had several low scoring questions. The five questions were examined more closely to determine whether there were any issues. Two questions (ExFmrRo2, ExFmrRo5) showed correlated error and were removed (Brown, 2006).

A revised model was run with the remaining 18 questions (Figure 3). The overall fit of the model was improved dramatically and shown to be a better fit for the data than Model 1 (Table 3). The chi-square was significant, χ²(52)=259, p < .001. CFI and TLI values were 0.929 and 0.977, respectively, suggesting an acceptable fit. The RMSEA was mediocre with 0.093 but the SRMR was 0.056, indicating a good fit. Taken together, the results provide preliminary evidence for the validity of the FCIS.

Figure 3.

Path diagram of EFA for final model.

Table 3.

Fit Indices of Exploratory Factor Analysis for Revised Model (Final Model).

Chi-square	Value	259
	Degrees of freedom	52
	p value	.00
CFI/TLI	CFI	0.929
CFI/TLI	TLI	0.977
RMSEA	Estimate	0.093
SRMR	Value	0.056

Note. CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.

Step 7: Final check

In the final step, the 18-question instrument that resulted from the pilot testing stage was reviewed by three people who had no part in the development or revision process (Dillman, 2000). They were tasked with proofing the questions for wording or content problems only. No errors were identified, and no additional edits were made.

Discussion

The purpose of this study was to develop and to perform a preliminary evaluation of the FCIS. The FCIS can be used to measure the extent to which an individual identifies with the family caregiver role. A valid and reliable measure of family caregiver identity was needed to advance research and practice in the area of family caregivers. Much of the prior research approached caregiver identity as a one-dimensional factor, rather than as a multidimensional construct with a number of different facets. This study contributes to the growing body of research involving caregiver identity.

Recent funding from the National Institutes of Health (2015) identified “improved definitions and characterization of caregiving” (para. 4) as a critical need in caregiver research. They based this on reports from experts on the next steps for research on informal caregiving (National Institute on Aging, 2012). Among those reports was the suggestion that a gap existed in the “naming and framing” of the term caregiver, especially regarding the “descriptive label applied to family members in research” (Gitlin, 2012, p. 3). The creation of the FCIS to measure family caregiver identity would help close gaps and remove superficial labels limiting research on family caregiving.

Furthermore, caregiver identity may be closely related to caregiver stress, burden, and burnout. According to Montgomery et al.’s (2011) Caregiver Identity Theory, a major source of caregiver stress is caregiver identity discrepancy or “a disparity between the care responsibilities that he or she is assuming and his or her identity standard” (p. 641). To reduce caregiver incongruence, Montgomery and Kosloski (2009) suggest that one must (a) change their behaviors to bring them into line with their identity standard, (b) change their self-appraisal or perception of congruence, or (c) change their identity standard. When assisting individuals who are providing care to a family member, the first step may be helping that individual with the recognition of that role. The FCIS could be used with interventions aimed at helping individuals transition into the caregiver role, develop their caregiver identity, and reduce caregiving-related stress.

Instrument development is a continuous process. Results from this study are preliminary with further research needed with other samples and influential factors. Within this study, the model was modified to achieve adequate fit. CFA must be repeated using the same measure as the latent structure of the 18-item FCIS across different samples or populations.

Limitations and Future Research

The strengths of this study are the systematic process and variety of accepted practices utilized to develop and pretest the FCIS. The results and instrument appear to be valid, but it is important to consider the limitations involved. It is possible that an aspect of caregiver identity was not included. Caregiver identity research is still emerging and there may be other factors not stated in the literature and unbeknownst to others. Future research should explore the five domains suggested in the theoretical framework that provide the constructs measured by the FCIS and tested in this study, in particular, the domain family obligation and gender norming. Based on the results of the factor analysis, it is theorized that gender may moderate the relationship between family obligation and gender norming. This possibility needs to be examined and explored for further understanding and potential revisions to the theoretical framework and instrument. In addition, the relationship between the domains should be examined. Divergent validity is necessary to help establish construct validity by demonstrating that the five domains differ from each other.

Participants were recruited through community partner organizations in North Carolina and therefore comprise a convenience sample. North Carolina may have a regional belief system regarding family and caregiving that is different from other regions in the country (Kivett, 1976). This difference in the culture of family obligation should be taken into consideration when reviewing the results and examining the five-factor structure of caregiver identity development. The sampling plan limits the generalizability of the study beyond the sample. Future testing of the FCIS should be with a more representative sample. Furthermore, potential benefits of tailoring the FCIS based on certain cultural, behavioral, psychological, or other individual characteristics should be explored.

CFA has limitations as well. Alternative models are likely to fit the data equally as well as the one chosen. The choice of the appropriate model to understand the relationship between the latent variables was guided by theory and plausibility; however, it is possible that variations in the structural relationships could be found in future studies. Furthermore, the sample used for the factor analysis consisted of both current caregivers and past caregivers. The original model or an alternative model could work better with a sample of all current caregivers because their caregiver identity may be stronger. Future testing that examined differences in caregiver identity between current and past caregivers using the FCIS may be necessary. Finally, the modification of the original, 25-item model using CFA in the same sample could lead to a lack of generalizability. Although the instrument went through extensive testing, further research is needed to evaluate and refine the revised, 18-item FCIS. The logical next step in the ongoing process of assessing the measurement properties of the FCIS are additional tests of validity and reliability.

Conclusion

This research highlights the significance of a more global measure of family caregiver identity. A systematic literature review conducted previously revealed that there are multiple factors that influence caregiver identity development (Eifert et al., 2015). These layered and complex factors cannot be captured in a single yes/no question—“Are you a caregiver?” For this reason, the FCIS, an instrument designed to measure the extent to which an individual identifies with the family caregiver role, was developed and evaluated. Using an adapted instrument development framework from Dillman’s (2000) steps of pretesting and the Standards (AERA, APA, & NCME, 1999) resulted in an FCIS that is promising but needs further validation testing. If the FCIS performs well in the next stages of testing, it could prove to be a useful tool to measure family caregiver identity in research and practice. The full instrument is available upon request from the authors. The 18-item version that resulted from this study is included as a Supplemental Appendix.

Supplemental Material

Appendix_Instrument25_Final – Supplemental material for Preliminary Evidence for the Validity of the Family Caregiver Identity Scale

Supplemental material, Appendix_Instrument25_Final for Preliminary Evidence for the Validity of the Family Caregiver Identity Scale by Elise K. Eifert, William Dudley, James Eddy, Michael Perko and Rebecca Adams in Journal of Applied Gerontology

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Human Subjects

Basic ethical principles underlying the conduct of research involving human subjects were followed in accordance with the standards set forth in the Belmont Report. The study was approved by the University of North Carolina at Greensboro (Institutional Review Board [IRB] #14-0055).

ORCID iD

Elise K. Eifert

Supplemental Material

Supplemental material for this article is available online.

References

American Association of Retired Persons. (2001). AARP caregiver identification study. http://assets.aarp.org/rgcenter/post-import/caregiver.pdf

American Association of Retired Persons. (2015). Valuing the invaluable: 2015 update. http://www.aarp.org/content/dam/aarp/ppi/2015/valuing-the-invaluable-2015-update-new.pdf

American Educational Research Association, American Psychological Association, and National Council on Measurement in Education. (1999). Standards for educational and psychological testing. AERA Publications. http://psycnet.apa.org/doi/10.1037/e577932014-003

Biddle

B. J.

Bank

B. J.

Anderson

D. S.

Hauge

Keats

D. M.

Keats

J. A.

. . . Valantin

(1985). Social influence, self-referent identity labels, and behavior. The Sociological Quarterly, 26(2), 159–185. https://doi.org/10.1111/j.1533-8525.1985.tb00221.x

Bollen

K. A.

(1989). Structural equations with latent variables. John Wiley. https://doi.org/10.1002/9781118619179

Bronfenbrenner

(1979). The ecology of human development. Harvard University Press.

Brown

T. A.

(2006). Confirmatory factor analysis for applied research. The Guilford Press.

Browne

M. W.

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural equation models (pp. 136–162). SAGE.

Byrne

B. M.

(2012). Structural equation modeling with Mplus. Routledge.

10.

Carbonneau

Caron

Desrosiers

(2010). Development of a conceptual framework of positive aspects of caregiving in dementia. Dementia, 9, 327–353. https://doi.org/10.1177/1471301210375316

11.

Cohen

C. A.

Colantonio

Vernich

(2002). Positive aspects of caregiving: Rounding out of the caregiver experience. International Journal of Geriatric Psychiatry, 17(2), 184–188. https://doi.org/10.1002/gps.561

12.

Dillman

D. A.

(2000). Mail and internet surveys: The tailored design method. John Wiley.

13.

Dobrof

Ebenstein

(2003). Family caregiver self-identification: Implications for healthcare and social service professionals. Generations, 27(4), 33–38. http://www.asaging.org/generations-journal-american-society-aging

14.

Eifert

E. K.

Adams

Dudley

Perko

(2015). Family caregiver identity: A literature review. American Journal of Health Education, 46(6), 357–367. https://doi.org/10.1080/19325037.2015.1099482

15.

Fives

DiDonato-Barnes

(2013). Classroom test construction: The power of a table of specification. Practical Assessment, Research, & Evaluation, 18(3), 1–7. http://pareonline.net/getvn.asp?v=18&n=3

16.

Flora

D. B.

Curran

P. J.

(2004). An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. Psychological Methods, 9(4), 466–491. https://doi.org/10.1037/1082-989X.9.4.466

17.

Fornell

C. R.

Larcker

D. F.

(1981). Structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18, 39–50. https://doi.org/10.2307/3151312

18.

Fowler

F. J.

(1995). Improving survey questions (Vol. 38). SAGE.

19.

Furr

R. M.

Bacharach

V. R.

(2014). Psychometrics: An introduction (2nd ed.). SAGE.

20.

Garland

(1991). The mid-point on a rating scale: Is it desirable? Marketing Bulletin, 2, 66–70. http://rangevoting.org/MB_V2_N3_Garland.pdf

21.

Gitlin

L. N.

(2012). Research on informal caregiving: Gaps in knowledge and next steps. https://www.nia.nih.gov/sites/default/files/2017-08/gerald-summary_11-21-14_0.pdf

22.

Henderson

(2001). “He’s not my carer—He’s my husband”: Personal and policy constructions of care in mental health. Journal of Social Work Practice, 15(2), 37–41. https://doi.org/10.1080/02650530120090601

23.

Holland

(1997). Selves as cultured: As told by an anthropologist who lacks a soul. In Ashmore

Jussim

(Eds.), Self and identity: Fundamental issues (pp. 160–190). Oxford University Press.

24.

Hong

Cadado

B. L.

(2015). Caregiver stress: Does states’ expenditure on home- and community-based services matter? Home Health Care Services Quarterly, 34(2), 85–100. https://doi.org/10.1080/01621424.2015.1029186

25.

Horrell

Stephens

Breheny

(2015). Capability to care: Support the health of informal caregiver for older people. Health Psychology, 34(4), 339–348. https://doi.org/10.1037/hea0000144

26.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. https://doi.org/10.1080/10705519909540118

27.

Hughes

Locock

Ziebland

(2013). Personal identity and the role of “carer” among relatives and friends of people with multiple sclerosis. Social Science and Medicine, 96(100), 78–85. https://doi.org/10.1016/j.socscimed.2013.07.023

28.

Kivett

V. R.

(1976). The aged in North Carolina: Physical, social and environmental characteristics and sources of assistance. North Carolina Agricultural Experiment Station.

29.

Kulas

J. T.

Stachowki

A. A.

Haynes

B. A.

(2008). Middle response functioning in Likert-responses to personality items. Journal of Business & Psychology, 22(3), 251–259. https://doi.org/10.1007/s10869-008-9064-2

30.

Lopez-Hartmann

Wens

Verhoeven

Remmen

(2012). The effect of caregiver support interventions for informal caregivers of community-dwelling frail elderly: A systematic review. International Journal of Integrated Care, 12, e133. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3601532/

31.

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. https://doi.org/10.1037//1082-989x.1.2.130

32.

Mason

Weatherly

Spilsbury

Arksey

Golder Adamson

Drummond

Glendinning

(2007). A systematic review of the effectiveness and cost-effectiveness of different models of community-based respite care for frail older people and their carers. Health Technology Assessment, 11(15), 1–157, iii. https://doi.org/10.3310/hta11150

33.

Montgomery

R. J. V.

Kosloski

(2009). Caregiving as a process of changing identity: Implications for caregiver support. Generations, 33(1), 47–53. http://www.asaging.org/generations-journal-american-society-aging

34.

Montgomery

R. J. V.

Kosloski

K. D.

(2012). Pathways to a caregiver identity and implications for support services. In Talley

R. C.

Montgomery

R. J. V.

(Eds.), Caregiving across the lifespan: Research, practice, policy (pp. 131–156). Springer.

35.

Montgomery

R. J. V.

Kwak

Kosloski

O’Connell Valuch

(2011). Effects of the TCARE® intervention on caregiver burden and depressive symptoms: Preliminary findings from a randomized controlled study. Journals of Gerontology: Series B, 66(5), 640–647. https://doi.org/10.1093/geronb/gbr088

36.

Muthén

L. K.

Muthén

B. O.

(2004). Mplus user’s guide (6th ed.).

37.

National Alliance for Caregiving & AARP Public Policy Institute. (2015). Caregiving in the U.S. http://www.caregiving.org/wp-content/uploads/2015/05/2015_CaregivingintheUS_Final-Report-June-4_WEB.pdf

38.

National Institute on Aging. (2012). Next steps for research on informal caregiving. https://www.nia.nih.gov/research/dbsr/bsr-sponsored-publications

39.

National Institutes of Health. (2015). Research on informal and formal caregiving for Alzheimer’s disease. http://grants.nih.gov/grants/guide/pa-files/PAR-15-351.html

40.

Netemeyer

R. G.

Bearden

W. O.

Sharma

(2003). Scaled procedures: Issues and applications. SAGE.

41.

O’Connor

D. L.

(1999). Living with a memory-impaired spouse: (Re)cognizing the experience. Canadian Journal on Aging, 18(2), 211–235. https://doi.org/10.1017/S0714980800009788

42.

Pruchno

R. A.

Brill

J. E.

Shands

Gordon

J. R.

Genderson

M. W.

Rose

Cartwright

(2008). Convenience samples and caregiving research: How generalizable are the findings? The Gerontologist, 48(6), 820–827. https://doi.org/10.1093/geront/48.6.820

43.

Russell

D. W.

(2002). In search of underlying dimensions: The use (and abuse) of factor analysis in Personality and Social Psychology Bulletin. Personality and Social Psychology Bulletin, 28, 1629–1646. https://doi.org/10.1177/014616702237645

44.

SPSS, Inc. (2009). PASW statistics for Windows (Version 18.0). Chicago, IL: SPSS Inc.

45.

Straub

Boudreau

M. C.

Gefen

(2004). Validation guidelines for IS positivist research. Communications of Association for Information Systems, 13(24), 380–427. http://aisel.aisnet.org/cais

46.

Thompson

(2004). Exploratory and confirmatory factor analysis: Understanding concepts and applications. American Psychological Association. https://doi.org/10.1037/10694-000

47.

Wakui

Cheng

S. T.

(2016). Filial responsibility. In Pachana

(Eds.), Encyclopedia of geropsychology (pp. 1–8). Springer.

48.

Weems

G. H.

Onwuegbuzie

A. J.

(2001). The impact of midpoint responses and reverse coding on survey data. Measurement and Evaluation in Counseling and Development, 34(3), 166–176.

49.

Werner

Goldstein

Buchbinder

(2010). Subjective experience of family stigma as report by children of Alzheimer’s disease patients. Qualitative Health Research, 20(2), 159–169. https://doi.org/10.1177/1049732309358330

50.

Willis

G. B.

(2005). Cognitive interviewing: A tool for improving questionnaire design. SAGE.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.17 MB