PRELIMINARY VALIDATION OF AN ATTITUDES AND EXPECTATIONS ABOUT FUTURE CAREGIVING ASSESSMENT MEASURE
by
KENDALL L. WEBER
B.S., Northern Arizona University, 2018
A thesis submitted to the Graduate Faculty of the University of Colorado Colorado Springs
in partial fulfillment of the requirements for the degree of
Master of Arts Department of Psychology
This thesis for the Master of Arts degree by Kendall L. Weber
has been approved for the Department of Psychology
by
Sara Qualls, Chair
Daniel Segal
Rachel Thayer
Anticipatory Caregiving Scale: Development and Preliminary Validation of An Attitudes and Expectations about Future Caregiving Assessment Measure
Thesis directed by Professor Sara Qualls
ABSTRACT
As public awareness of family caregiving has grown, adults likely anticipate the role they may play as a caregiver for an aging loved one. Although anticipatory planning for caregiving has been studied, no measure of multiple dimensions of the anticipated caregiving experience exists. The purpose of the present study is to develop and validate the Anticipatory Caregiving Scale (ACS), an assessment of adult children’s attitudes toward and level of expectation surrounding their potential role as an informal caregiver to a parent or parent-in-law. A sample of 540 adults aged 18 and over recruited online completed the ACS, along with scales to assess convergent, discriminant, and concurrent validity, including the Preparedness for Caregiving Scale, Positive and Negative Affect Schedule, Cultural Justification for Caregiving Scale, Cost of Care Index, and Scope of Care Scale. The ACS consists of six subscales that assess affect surrounding future caregiving, anticipated lifestyle interference, self-efficacy surrounding future caregiving, anticipated caregiving resources, endorsement of norms of family care which influence anticipated caregiving, and the relationship quality, current and anticipated, with the potential care recipient. The whole scale and subscales were found to be internally consistent as measured with Cronbach’s alpha, and demonstrated evidence for convergent, concurrent, and discriminant validity.
Key words: caregiving, caregiving assessment, future care, anticipatory caregiving, adult child caregiving
CHAPTER
I. INTRODUCTION ...1
Informal Caregiving ...2
Anticipatory Caregiving...3
Affect Surrounding Future Caregiving ...5
Self-Efficacy Surrounding Future Caregiving ...6
Anticipated Lifestyle Interference ...7
Anticipated Resources ...8
Social & Familial Normal, Values & Beliefs ...9
Relationship with Potential Care Recipient ...11
Summary of Domains of Anticipatory Caregiving ...13
Scale Development ...14
Question Construction & Response Options ...14
Reliability ...15
Validity ...17
Filling the Gap ...19
Present Study ...20
II. METHOD ...24
Sample...24
Measures ...25
Demographics ...25
Anticipatory Caregiving Scale ...26
Preparedness for Caregiving Scale ...28
Positive and Negative Affect Schedule ...28
The Cultural Justification for Caregiving Scale ...29
Cost of Care Index ...29
Cost of Care Scale ...30
III. RESULTS ...31
Descriptives...31
Internal Consistency (H1) ...31
Variation in Responses by Demographics ...34
Gender ...34
Race...34
Age ...38
ACS Whole-Scale and Subscale Correlation matrix (H2) ...39
Convergent Validity (H3) ...40 Discriminant Validity (H4) ...42 Concurrent Validity (H5) ...42 IV. DISCUSSION ...44 Reliability ...44 Validity ...46 Limitations ...47
Conclusion ...47
REFERENCES ...49
APPENDIX A: CONSENT TO BE A RESEARCH SUBJECT ...57
APPENDIX B: STUDY SCREENER SURVEY ...66
APPENDIX C: ANTICIPATORY CAREGIVER SCALE ...67
APPENDIX D: PREPAREDNESS FOR CAREGIVING SCALE ...70
APPENDIX E: THE POSITIVE AND NEGATIVE AFFECT SCHEDULE ...71
APPENDIX F: CULTURAL JUSTIFICATION FOR CAREGIVER SCALE ...72
APPENDIX G: COST OF CARE INDIX ...73
APPENDIX H: SCOPE OF CARE SCALE ...75
APPENDIX I: DEMOGRAPHICS ...76
APPENDIX J: DEBRIEFING ...79
LIST OF TABLES TABLE
1. Demographic Descriptives ...32
2. Means, Standard Deviations, and Cronbach’s Alpha for Scales and Subscales ...34
3. ACS Inter-item correlations ...35
4. ACS Item Descriptives & Item total Correlations ...37
5. ACS Scale and Subscale Differences Between Males and Females ...38
6. ACS Scale and Subscale Differences Between White and Non-White Participants ...39
7. ACS Scale, Subscales, and Age Correlations ...39
INTRODUCTION
As public awareness of family caregiving has grown, with an estimated 34.2 million people in the United States providing care for an older adult family member (National Alliance for Family Caregiving, 2015), adults likely anticipate the role they may play as a caregiver for an aging loved one. Yet, there is currently no measure of anticipation of the caregiving experience. Strides have been taken to understand family caregivers through measures assessing their experiences (e.g., Caregiver Burden Interview, CBI, Novak & Guest, 1989; Caregiver Reaction Scale, CRS, O’Malley & Qualls, 2020; Zarit Burden Interview, ZBI, Zarit, Orr, & Zarit, 1985). However, few assess how individuals perceive their potential role as caregivers, including their thoughts, feelings, and intent surrounding the idea. Research indicates increased
preparedness, planning, and expectation for caregiving may help promote a more positive caregiving experience for an individual, so evaluating expectations of future caregiving is increasingly important (Schumacher, Stewart, & Archbold, 2007; Wallhagen &
Yamamoto-Mitani, 2006). The purpose of the present study is to develop and validate the Anticipatory Caregiving Scale (ACS), an assessment of adult children’s attitudes towards and level of expectation surrounding their potential role as an informal caregiver to a parent. The measure will assess multiple factors that may influence their willingness to take on a caregiving role and their preparation for it. Those factors include affect and self-efficacy regarding the potential caregiving role, the anticipated interference
caregiving will have on their lifestyle, their anticipated resources, the relationship quality with the anticipated care recipient, and their endorsement of norms of family care which may influence their decision to provide care.
Informal Caregiving
An informal caregiver has been defined as an adult who has a significant relationship with and provides wide-ranging, unpaid assistance to a loved one with a debilitating illness (National Alliance for Caregiving & AARP Public Policy Institute, 2015). Typically, care provided includes assisting care recipients with activities of daily living (i.e., dressing, bathing, eating), instrumental activities of daily living (i.e., grocery shopping, housework, using the telephone), coordination of care, and acting as their care recipient’s advocate (Levine, Reinhard, Feinberg, Albert, & Hart, 2003; National
Alliance for Caregiving & AARP Public Policy Institute, 2015). However, caregivers also frequently provide care recipients with emotional support and care. In fact, research suggests that for many caregivers, some of the most difficult and time-consuming care provided to care recipients is emotional support (Bakas, Lewis, & Parsons, 2001). A noteworthy distinction between informal caregiving and care that family members may inherently provide simply as part of a relationship, is that in informal caregiving the care recipient now requires assistance in one or more of these domains, due to the debilitating illness (Qualls, 2016).
Just as the landscape of caregiving looks different from person to person, it also differs depending on the type of caregiving relationship (i.e., adult child, spouse). More caregivers are adult children caring for an aging parent than any other type of relationship (e.g., friend, other relative), with spousal caregiving the second most common caregiving
relationship (National Alliance for Family Caregiving, 2015). Although previous research has not yet met consensus on who (adult children vs. spouses) experiences more
caregiver burden or stress related to the responsibilities and demands associated with caregiving, clear differences do exist between their experiences (Pinquart & Sörensen, 2011). These types of caregivers may differ in hours spent caregiving, reported care recipient behavioral problems, and living situations. More specifically, spousal caregivers are more likely to live with their care recipients and provide more direct hours of care to their care recipient than adult child caregivers (Pinquart & Sörensen, 2011). On the other hand, adult child caregivers who are less likely to live with their care recipients may still complete the same number of caregiving tasks in a shorter period of time as spousal caregivers (Pinquart & Sörensen, 2011). Spouses tend to become the primary attachment figures for adults, and thus may experience more relationship strain than adult child caregivers. However, adult child caregivers often encounter more role conflict than spousal caregivers, as they maintain their own spousal relationships and may be caring for their children as well (Pinquart & Sörensen, 2011). Given these and other differences between adult child and spousal caregivers, it is important that adult child and spousal caregivers be examined separately when it comes to caregiving, anticipatory or current. In the present study, the focus is on developing a measure to assess key dimensions of the anticipatory caregiving experience of adult children.
Anticipatory Caregiving
Anticipatory caregiving is defined as the awareness about the potential need for future care for oneself or a loved one (Sörensen, 1998). Sörensen conceptualizes anticipation of caregiving as the first step in preparation for caregiving, followed by
decision making, concrete planning, and role socialization. Not only does anticipation influence who ultimately decides to enter into the caregiving role, but may even influence the experience of caregiving and caregiver outcomes. Increased preparedness for
caregiving has been found to be associated with overall caregiver well-being, including acting as a protective factor against caregiver burden, negatively correlating with fatigue, strain, and mood disturbance (Schumacher, Stewart, Archbold, Caparro, Mutale, & Agrawal, 2008; Schumacher, Stewart, & Archbold, 2007). As anticipation of a future caregiving role is the first step to preparedness, evaluation of such views is also an important first step for understanding how to promote family awareness and planning, along with positive caregiver experiences and outcomes, including future health and well-being.
The anticipation of other roles has similarly been found to influence the decision to take on that role and the outcomes in the role. For instance, anticipating retirement has been found to increase preparedness and the likelihood of making the decision to retire (Reitzes, Mutran, & Fernandez, 1998). Expectations surrounding parenthood also influence individuals’ decision-making and preparation. Bass (2015) found that, when compared to men, women tended to shift their career or educational goals in preparation for future parenthood because they anticipated the event and the type of care they would provide more than men. Longitudinal data suggest that adolescents’ values, such as civic engagement and personal responsibility, predict their behaviors, such as alcohol use and engaging in politics, and role attainment in adulthood. Those who endorsed family responsibility as a personal value in adolescence were more likely to marry and have children in adulthood (Finlay, Wray-Lake, Warren, & Maggs, 2015). Therefore,
identifying expectations and attitudes towards a potential future role, such as caregiving, may impact the decision to enter and experience in that role.
Anticipation of caregiving, as defined in the current study, has been associated in the literature with multiple dimensions: anticipated feelings, losses and gains, resources, and other expected personal experiences related to future caregiving. The rationale for including each of these domains of anticipatory caregiving is that each may influence the decision to provide care, the experience of being a caregiver, and caregiver outcomes. The literature that justifies the potential importance of each domain is reviewed in the sections below. Two types of evidence are used to support inclusion of the domain in the scale: 1) evidence that the domain differentiates caregivers from non-caregivers and thus might have influenced the choice to engage in the role, and 2) that the domain predicts outcomes in caregivers.
Affect Surrounding Future Caregiving. Affective experiences or feelings may influence the decision to engage in certain behaviors (Charpentier, De Neve, Li, Roiser, & Sharot, 2016; Mischel, 2004), and thus may influence the decision to deliver
caregiving supports and services. Emotions involved in anticipating caregiving could include happiness, nervousness, irritability, fear, determinism, and inspiration, and have all been associated with more general decision-making (Lench et al., 2019; Mellers, Schwartz, & Ritov, 1999), the strength of which can influence one’s decision to engage in a role or event (Mellers & McGraw, 2001; Morewedge & Buechel, 2013). However, no research has described the presence or prevalence of those emotions in adults specific to their anticipation of possible caregiving role for a particular person. Current caregivers experience more negative affect and symptoms of depression and less positive affect
when compared to non-caregivers (Grossman & Gruenewald, 2017). Additionally, longitudinal data has shown that positive affect in caregivers at one point in time has been associated with more social support and perceived gains from caregiving up to 12 months later (Grigorovich et al., 2017). Rae (1998) identified the emotion work involved in caregiving, including management of not only the emotions of the care recipient but also of the caregivers. Rae proposed the importance of teaching skills in emotion regulation and management to caregivers, starting with identifying emotions. Verstaen, Moskowitz, Snowberg, Merrilees, and Dowling, (2018) are currently conducting randomized control trials of a protocol to do just this, increase positive affect in caregivers and assess subsequent improvements in mental health, physical health, and other caregiver outcomes. Preliminary data from this intervention suggest that increasing positive affect has lowered stress and improved health outcomes in caregivers, suggesting that identifying and managing emotions in caregivers is important to their wellbeing. Certainly, the presence of positive and negative emotions has been established in caregivers, and those emotions have predicted outcomes. Yet the role of affect in anticipatory caregiving is unknown in its presence, intensity, and ability to predict engagement in caregiving. Logic suggests that the positivity of feelings a person has about caregiving, and caregiving for a particular person, will influence the likelihood of providing care and perhaps will influence the experience of caregiving and outcomes. Thus, this dimension is deemed important to include in a measure of anticipatory caregiving.
Self-Efficacy Surrounding Future Caregiving. Caregiving self-efficacy, or the belief in one’s ability to manage the challenges and complete the tasks associated with
caregiving, may influence one’s attitude towards providing care to an aging family member. Because self-efficacy is one’s own appraisal of potential or current ability to cope with these arising demands, it may also influence a chosen course of action
(Bandura, 1982). For instance, the belief that one is able or unable to handle the demands of caregiving seems likely to influence the choice to engage or not. Certainly,
self-efficacy has been demonstrated to correlate with outcomes in caregivers. Cross-sectional data show that self-efficacy is negatively correlated with depressive symptoms of
dementia caregivers, caregiver burden and stress, and relates to caregivers’ perception of positive aspects of the caregiving experience, in that better self-efficacy is related to more positive perceptions of caregiving (Gilliam & Steffen, 2006; Semiatin & O'Connor, 2012). Enhanced self-efficacy promotes positive caregiver experiences as evidenced in an intervention study. Hendrix et al (2016) provided informal caregivers of cancer patients with enhanced informal caregiver training protocol, and found that it increased self-efficacy, preparedness, and reduced caregiver stress. In sum, anticipatory self-efficacy may influence the decision to provide care and impact caregiver outcomes, making its measurement necessary in a scale of anticipatory caregiving.
Anticipated Lifestyle Interference. The extent to which the caregiving role is anticipated to interfere in one’s lifestyle may impact expectations and willingness to provide care. The demands of caregiving may affect lifestyles in multiple domains, such as health, employment, social activity, pleasurable activities, household responsibilities, etc. In a cross-sectional study examining the relationship between anticipated gains and losses associated with caregiving (e.g., increased self-efficacy or loss of autonomy) and self-reported willingness to provide future care, Rohr and Lang (2016) found that those
who reported a willingness to provide care anticipated fewer losses than those who were unwilling to provide care. Adult children who believe caregiving will negatively impact their lifestyle may be less likely to anticipate or to actually provide care to their aging parent.
Anticipation of interference is interesting to study because lifestyle interference has been demonstrated to impact current caregivers’ experiences and outcomes. In a study examining family caregivers of terminal cancer patients, Cameron, Franche, Cheung, & Stewart (2002) found that lifestyle interference mediated the relationship between level of care provided and emotional distress, such that more lifestyle interference led to more distress regardless of the amount or level of care provided. Cameron et al. (2002) based their study on Devins's Illness Intrusiveness Model, which posits that a care recipient’s illness impacts the caregiver’s mental health only to the extent that they cannot engage in valued activities (Devins, 1994). If losses in this domain occur, caregivers are negatively impacted. It would logically hold true that those
anticipating the role of caregiver may foresee this impact as well. If they believe they will be unable to engage in valued activities, they may anticipate a more negative caregiving experience or be less likely to anticipate providing care.
Anticipated Resources. Anticipated caregiver resources, including family support, community resources, and personal resources (i.e., time, money), may influence decisions to provide care and the anticipated experience of caregiving. Research indicates that more personal resources, such as time not already dedicated to other responsibilities and money, may influence one’s decision to provide care (Pillemer & Suitor, 2013). Factors that leave an adult child with less time and increased likelihood for role conflict,
such as being married, working, and having children have been found to decrease the likelihood of them providing care to parents (Bucx, Wel, & Knijn, 2012; Dautzenberg, Diederiks, Philipsen, Stevens, Tan, & Vernooij-Dassen, 2000; Moen, Robison, & Dempster-McClain, 1995; Wolf, Freedman, & Soldo, 1997). Caregivers with more perceived social support, such as from family and friends, have been found to have less caregiver burden than those with less perceived support (Shiba, Kondo, & Kondo, 2016; del-Pino-Casado, Friás-Osuna, Palomino-Moral, Ruzafa-Martínez, & Ramos-Morcillo, 2018). The same appears to hold true for those who have more actual and perceived formal support from healthcare providers, such that those with more support report less caregiver burden than those with less reported formal support (Chambers et al., 2004; Verbakel, Metzelthin, & Gertrudis, 2018). Therefore, those who may have or anticipate a future with fewer of these resources may not choose to provide care to their aging parent, or if they do, may experience more role conflict and subsequent negative outcomes. Anticipation of resources is an important domain to include in a measure of anticipatory caregiving because it may influence the decision to provide caregiving as well as the experience and outcomes of caregiving.
Social & Familial Norms, Values & Beliefs. Social norms, the typical or appropriate expected behaviors, beliefs, or values in a given group, influence expectations about future caregiving. In caregiving research, the social and familial norms, or norms learned within the family, with the most importance in shaping the caregiving experience are a sense of duty or obligation, and reciprocity (Silverstein, Conroy, Giarrusso, & Bengtson, 2002; Speirs & Konnert, 2017; Wallhagen &
familism or filial piety, is one value or norm that may influence the decision to provide care to them. Although norms of obligation are powerful in Euro-American aging families (Giarrusso, & Bengtson, 2002), these norms are particularly associated with Asian cultures (Speirs & Konnert, 2017; Wallhagen & Yamamoto-Mitani, 2006). In a cross-sectional study comparing cultural difference among young adults’ expectations about future care of family members, filial piety was the strongest predictor of
anticipation about providing care (Speirs & Konnert, 2017). While comparing Japanese and American daughters and daughters-in-law caregivers across six months, Wallhagen and Yamamoto-Mitani (2006) similarly found that differing social norms or caregiving exist, including filial piety and duty. While they found that both American and Japanese indicated that caregiving was a moral obligation, something they felt that they had to do in line with their self-image and beliefs, they also found that Japanese caregivers see caregiving as a more normative, anticipated life event, whereas American caregivers reported it as an unexpected and jarring experience. Wallhagen & Yamamoto-Mitani (2006) posit that these results indicate the need for anticipatory socialization for those whose cultural norms may lead them to view caregiving as an unexpected life detour to allow for education and preparation. In other words, cultural norms and values may influence anticipation about providing care, although the weight of the impact may vary across cultural groups. In the development of their Sociocultural Stress and Coping Model of caregiving, Knight and Sayegh (2010) found that familism and cultural values like individualism and collectivism were not associated with caregiver outcomes such as caregiver burden, or physical and mental health. Instead, they found that more nuanced cultural values at a local level, as well as cultural responses to stress and coping, were a
better indicator of caregiver outcomes and appraisals of burden. This suggests that, while familism and certain cultural values may lead to the decision to provide or anticipate providing care, they alone may not impact caregiver outcomes and well-being once caregiving. Instead, other cultural differences, such as in coping styles, may have a larger impact.
Past research also indicates that the belief in reciprocity may lead adult children to feel obligation to care for their aging parents because their parents have provided them with care in the past. Silverstein, Conroy, Giarrusso, & Bengtson (2002) found that reciprocity predicts future care, such that more support and resources provided to children from parents within a family predicted future amount of support and care children provided to parents. Leopold, Raab, and Engelhardt (2014) similarly found that reciprocity was a driving factor amongst siblings’ decisions about who provides care to a parent in need. Although reciprocity may be a driving factor behind caregiving, it should be noted that, even if past emotional, financial, or other support was absent, adult
children still provided increasing care to their parents over time, suggesting other
motivating factors, such as familism, may be at play regardless of reciprocity. Therefore, a mixture of cultural norms, beliefs, and values may influence decisions to provide care, expectations about what that care may look like, and outcomes for both caregivers and their care recipients. One’s own felt pressure or obligation to provide care may lead to differences in determination of who in a family provides care, preparation and type and amount of care provided.
Relationship with Potential Care Recipient. Consideration of the quality of, and satisfaction with, the adult child-parent relationship may also influence how an adult
child anticipates future caregiving. For instance, an adult child may consider the quality and history of the relationship with the parent when making decisions about whether or not to provide care. Prior research indicates that the better the quality of the adult child-parent relationship, the more likely that child will provide care to their child-parent when needed (Stuifbergen, Van Delden, & Dykstra, 2008). Rohr and Lang (2016) found that those who were willing to provide care were more likely to be satisfied with their relationship with the potential care recipient than those undecided or unwilling about becoming caregivers, indicating relationship satisfaction may lead to anticipating care. Relationship quality between caregivers and care recipients, both prior to and during caregiving may also predict caregiver outcomes. Relationship quality has been shown to be a predictor of depressive symptoms and caregiver wellbeing amongst caregivers, such that a stronger pre-caregiving and current caregiving relationship leads to higher
caregiver well-being and fewer depressive symptoms (Adams, McClendon, & Smyth, 2008; Quinn, Clare, & Woods, 2012).
Anticipation of how caregiving may affect the relationship with the parent or the parent’s preference for future care may also influence an adult child’s decision to provide care. For instance, a strengthened relationship between a caregiver and care recipient may be perceived as a gain associated with caregiving, while a weakened or strained
relationship may be an anticipated loss (Cohen, Colantonio, & Vernich, 2002; Rohr & Lang, 2016). Similarity between an adult child and a parent, including shared attitudes and values, is additionally important to creating and maintaining strong relationships (Pillemer & Suitor, 2014; Suitor, Sechrist, Gilligan, & Pillemer, 2011). In two studies examining mother’s preferences for a future caregiver, results indicated a preference for
naming children with similar attitudes and social characteristics to themselves (Pillemer & Suitor, 2006; Pillemer & Suitor, 2014). Additionally, children may be aware of the personal expectation to provide care for particular family members. In a longitudinal study examining mother’s preferences regarding which adult child will provide care, mother’s preferences at Time 1 predicted who would actually provide care at Time 2, seven years later (Pillemer & Suitor, 2014; Suitor, Gilligan, & Pillemer, 2013). When mothers’ expectations about who would provide care were not met, the mothers reported increased depression at Time 2 than those whose expectations were met. Therefore, not only does relationship quality impact both the adult child and parent’s choice of future caregiver, but may impact the child’s attitudes towards caregiving.
Summary of Domains of Anticipatory Caregiving. Current literature on anticipatory caregiving, or the awareness of the potential need for future care for oneself or a loved one, suggests that the domains of affect, self-efficacy, lifestyle interference, resources, norms, and relationship between caregiver and care recipient are important in the context of future caregiving. Positive affect and increased self-efficacy surrounding future care has been linked to increased likelihood of providing care and better caregiver outcomes once providing care. Anticipated changes in lifestyle and available resources influence attitudes toward future caregiving, such that positive lifestyle changes and more available resources may increase people’s willingness to provide care. The social norms and expectations about providing care (i.e., duty, reciprocity), relationship quality of between the adult child and parent, and expectations of the future care recipient influence decisions to provide care as well as their expectations of the experience. Specific aspects
of each of these domains need to be represented in the items in the proposed scale, and subjected to evaluation of the structure, reliability, and validity for the scale.
Scale Development
Development of a new scale involves careful review of existing literature,
followed by item generation and decisions regarding item response style, and assessment of the scale’s reliability and validity along with its underlying factor structure. Scale development begins with the generation of items that capture the breadth and depth of the construct being measured, in this case, anticipatory caregiving (Boateng, Neilands, Frongillo, Melgar-Quiñonez, & Young, 2018). Items should be grounded in the current empirical or theoretical literature, so items generated for the present scale will be taken from the six domains previously described as impacting future or current caregiving outcomes. In order to ensure accuracy and consistency of construct measurement,
reliability and validity are assessed. Additionally, it should be noted that the present study attempts to provide preliminary evidence for reliability and validity of the ACS, knowing that additional aspects of scale validation will require administration across differing samples, settings, contexts, etc.
Question Construction & Response Options. Scale items should be written as simple, straightforward statements, with response options that capture variability in responses. A commonly used approach is to have respondents indicate level of agreement on a Likert-type scale, such as the option used in the present study: 1 (strongly disagree), 2 (disagree), 3 (somewhat disagree), 4 (somewhat agree), 5 (agree), 6 (strongly agree). Note that in this example, a neutral response option was excluded. Although research indicates participants may prefer to have such a response option (Luskin & Bullock,
2001), other research indicates that a neutral response option may discourage participants from putting in the work required to evaluate their opinion (Krosnick et al., 2002).
Krosnick and Presser (2009) also established that 5 to 7-point Likert scales tend to be more reliable than scales with fewer response options, with benefits levelling off past 7 points, making 5 to 7 response options most desirable.
Reliability. Measurement reliability is the consistency with which a measure can assess a particular construct. Consistency should be present across items and time, and where relevant, across judges’ ratings of a behavior/observation, reflecting the three types of reliability: internal consistency, temporal stability, and interrater reliability.
Interrater Reliability. A measure demonstrates interrater reliability when two or more judges or observers demonstrate a high degree of agreement or consensus among their scoring. However, interrater reliability is most relevant when scales require observational ratings from trained judges or coders. For example, if a set of raters were trained to count how many words related to caregiving an adult child used during an interview and they each counted the same number of words, there would be interrater reliability. As the anticipatory caregiving scale is a self-report measure, it does not require raters’ observation and ratings, and therefore, interrater reliability will not be measured in the present validation.
Internal Consistency. A measure demonstrates internal consistency when the individual items that make up the measure are interrelated (Crano, Brewer & Lac, 2015). The hope is that, with internal consistency, the items will all be measuring the same construct, such as anticipatory caregiving. A commonly used measurement of a scale’s internal reliability is Cronbach’s alpha (Cronbach, 1951). Cronbach’s alpha is obtained
by taking the average correlations between all possible pairs of “split-half” correlations. Split-half correlations are obtained by devising one “whole” scale, dividing those items into two equal sets, and correlating their mean or total scores. Cronbach’s alpha ranges from 0.00 to 1.00, with 0.00 representing poor internal consistency and 1.00 representing excellent internal consistency. Acceptable internal consistency reliability is typically considered at Cronbach’s alpha of .60 or higher (Crano et al., 2015). A measurement tool can be conceptualized as unidimensional, generating a single score, or multi-dimensional and generate multiple subscale scores. The present study aims to measure Cronbach’s alpha for both the whole-test and subscales, understanding that subscales will likely ‘hang together,’ or correlate more strongly with one another than items taken altogether (Nunnally, 1967).
Additionally, both inter-item and item-total correlations between both subscale and scale items will be reviewed along with Cronbach to measure internal consistency. Inter-item correlations measure the extent to which one item is related to other items on a scale or subscale. Item-total correlations represent the relationship between an item on a scale or subscale, and the total score of that scale or subscale without that item. The purpose of an item-total correlation is to determine whether an item is consistent with the average of the other item scores on a test. A higher value indicates that the item is related to the other items on the scale and measures the same construct, while lower values indicate that the item may not be related to the other items on the scale.
Temporal Stability. A measure demonstrates temporal stability if the scores or observations from one test administration are similar to those obtained at a second test administration, with the same items and respondents (Crano et al., 2015). Test-retest
reliability tests temporal stability by comparing scores on the same test, with the same respondents, at two different points in time. A large, positive correlation between the two measure scores would denote good test-retest reliability. Establishing temporal stability through test-retest is out of the scope of this study, but should be evaluated in future studies.
Validity. Measurement validity is the ability of a scale to accurately measure its intended underlying construct. Construct validity is typically made up of face validity, content validity, criterion validity, convergent validity, and discriminant validity.
Face Validity. At a glance, the items that comprise a scale should capture the construct which it claims to be measuring (Crano et al., 2015). Face validity is often not considered a true measurement of validity because it is not based on research or other data. Additionally, the argument can be made that face validity may actually be purposefully avoided in some instances, to avoid things such as social desirability or malingering. Items in the present study were written with the general aim of appearing to capture the construct of anticipatory caregiving, as judged by students and faculty in a laboratory that specializes in family caregiving research, drawing on language and constructs used in the research literature as described above and is therefore expected to have face validity.
Content Validity. Content validity is the degree to which a measure represents the complete scope of the construct. It is typically achieved by generating a set of items which covers the full range of the construct. These items may be generated by reviewing previous literature or theories, capturing the conceptual definition of the construct, or consulting experts or focus groups. In the present study, items were generated based off
of the current literature as well as related scales and constructs (e.g., preparedness for caregiving). Additionally, items were generated in an attempt to cover all of the features, or factors, that may logically make up anticipatory caregiving. The literature reviewed above provides the empirical basis for the proposed subscales within the new measure.
Criterion Validity. Criterion validity is the degree to which a measure is
statistically related to an outcome (Crano et al., 2015). Criterion validity can be divided into concurrent validity and predictive validity. Concurrent validity is established with cross-sectional data, by correlating scores on the measure with the targeted outcome. The measure would have acceptable evidence for concurrent validity if a moderate to high correlation is found between the measure and the targeted outcome. For instance, anticipating caregiving should lead to increased preparedness for caregiving (Sörensen, 1998). Predictive validity is established with longitudinal data, correlating scores on the measure at Time 1 with the targeted outcome at Time 2. The measure would have acceptable predictive validity if a moderate to high correlation is found. Measuring predictive validity is beyond the scope of this study, but should be evaluated in future studies, for example, by establishing statistical relatedness between scores on the
Anticipatory Caregiving Scale at Time 1, and attitudes about current caregiving or simply caregiving status at Time 2.
Convergent Validity. Convergent validity is the degree that measures of different constructs that should be related conceptually, are empirically (statistically) related (Crano et al., 2015). Convergent validity is established by comparing a new scale or subscale to established measures of constructs that should theoretically be related to the new scale. For example, a norms-of-family-care subscale theoretically should be related
to a cultural justification for caregiving measure. Correlations between the new measure and established measures of cultural values that promote caregiving should ideally reach a moderate, but not high, correlation, to establish that the scales are related but distinct.
Discriminant Validity. Discriminant validity is the degree that measures of different constructs that should be unrelated, are statistically unrelated (Crano et al., 2015). Like convergent validity, discriminant validity is established by comparing the new scale or subscale to established measures of constructs which should theoretically be unrelated to the new scale. For example, a measure of state affect, such as the Positive and Negative Affect Schedule, should be relatively unrelated to aspects of anticipatory caregiving like anticipated resources (Watson, Clark, & Tellegen, 1988). Correlations between the new measure and established measures should ideally have a zero to low correlation.
Filling the Gap
Existing scales that attempt to measure an anticipatory aspect of caregiving typically only assess one domain of attitudes towards or expectations surrounding caregiving. The Preparedness for Caregiving Scale (Archbold, Stewart, Greenlick, & Harvath, 1990) only assesses people’s perceived ability to meet various needs of care recipients. The Cost of Care Index (Kosberg & Cairl, 1986) assesses perceived potential or actual consequences of providing care to an aging family member, including personal and social restrictions, physical and emotional health, value, care recipient as
provocateur, and economic costs. However, each of these scales neglect to capture other important domains of anticipatory care as cited in the literature previously. The
measurement of beliefs about, attitudes towards, and expectations about one’s potential caregiving experience with a particular family member. It fills a gap in current measures by presenting a multidimensional assessment of expectation surrounding future
caregiving, specific to the unique relationship and experience of the adult child-parent dyad.
Present Study
The purpose of the present study is to develop and provide preliminary validation of a scale to evaluate attitudes and expectations about future caregiving, specifically to evaluate adult children’s attitudes and expectations about providing future care for a parent. The scale is expected to provide both a global score indicating strength of anticipation of providing future care, as well as subscale scores that assess expectations, including affect surrounding future caregiving, anticipated lifestyle interference, self-efficacy surrounding future caregiving, anticipated caregiving resources, endorsement of norms of family care which influence anticipated caregiving, and the relationship quality between the potential care recipient and care giver.
This study tested the reliability and validity of a set of items created by set of researchers using a sample of 540 adults ages 18 and older who were recruited online through the University of Colorado Colorado Springs’ Sona System and Amazon’s Mechanical Turk. Hypotheses were as follows:
1) The ACS whole-scale and subscales will each demonstrate internal consistency through the use of Cronbach’s alpha, item total reliability, and inter-item
2) Multiple dimensions of anticipatory caregiving will be identified. Specifically, six subscales (Affect, Self-efficacy, Lifestyle Interference, Resources, Norms of Family Care, and Relationship Quality) will be identified using a correlation matrix. It is expected that the subscales will be correlated with each other, but not with excessive overlap. Each subscale will additionally have a strong, positive correlation with the ACS whole-scale.
3) Subscales will demonstrate evidence for convergent validity as follows: a) The ACS Affect and Self-efficacy subscales will positively correlate with the Positive and Negative Affect Schedule’s Positive Scale, and negatively correlate with the Positive and Negative Affect Schedule’s Negative Scale (PANAS; Watson, Clark, & Tellegen, 1988). Research suggests a link between state affect, affect forecasting, and cognition, including how judgments are made (Clore & Huntsinger, 2007; Marroquín & Nolen-Hoeksema, 2015). As this literature suggests that current mood is correlated with judgments about future events and ratings of future events and how the future will feel, state affect, measured by the PANAS, should be related to affect surrounding future caregiving. Research also suggests a link between state affect and self-efficacy, such that more positive affect correlates with increased self-efficacy, and that more negative affect correlates with decreased self-efficacy, in a variety of domains (Hoeppner, Kahler, & Gwaltney, 2014; Joseph, Royse, Benitez, & Pekmezi, 2013; Kavanagh & Bower, 1985). It is expected that an extension of this would
be observed when comparing state affect to anticipated self-efficacy about caregiving.
b) The ACS Lifestyle Interference, Relationship Quality, and Resources subscales will correlate negatively with the Cost of Care Index (CCI; Kosberg & Cairl, 1986). The Cost of Care Index assesses dimensions of cost associated with caregiving, including personal and social restrictions, physical and emotional health, value, care recipient as provocateur, and economic costs. Therefore, the ACS subscales which also assess these dimensions should be statistically related.
c) The ACS Norms of Family Care subscale will correlate positively with Cultural Justification for Caregiving Scale (CJCS; Dilworth-Anderson, Goodwin, & Williams, 2004). Both the CJCS and ACS Norms of Family Care subscale measure cultural expectations of one’s decision to provide care such as duty, obligation, religion, and expectation, and thus should be statistically related.
4) Subscales will demonstrate evidence for discriminant validity as follows: The ACS Norms of Family Care and Resources subscales will not correlate with the PANAS Negative or Positive Scale. The existing literature does not suggest a link between state affect and appraisals of future resources or norms of family care in general, or more specifically in the context of anticipatory caregiving. Thus, it is expected that these subscales will be unrelated to the PANAS.
5) The ACS will demonstrate evidence for concurrent validity as follows: The overall ACS will be positively correlated with the Preparedness for Caregiving
Scale (PCS; Archbold, Stewart, Greenlick, & Harvath, 1990) and the Scope of Care Scale (SCS). Sörensen (1998) posited that anticipating caregiving should lead to increased preparedness for caregiving, thus, scores on the ACS should predict and correlate with scores on a scale tapping caregiver preparedness such as the PCS. Sörensen (1998) also delineated anticipatory caregiving into different task types (ADLs/IADLs). It would logically hold true that a strong anticipation of the caregiver role would predict willingness to provide a broad scope of care. Thus, the ACS should predict scores on a scale measuring willingness to take on different care tasks (ADLs/IADLs) such as the SCS.
METHOD Sample
The final sample consisted of 540 participants recruited online using the following inclusion criteria: English-speaking, live within the United States, aged 18 years or older, and able to identify one living parent or parent-in-law for whom they have not previously provided informal caregiving services (defined as having provided
required, wide-ranging, unpaid assistance due to a debilitating illness) but who may need it in the future. Participant age ranged from 18 – 73 (M = 36.25, SD = 15.34). Participants were predominantly female (58%), White (76%), and anticipated being the primary caregiver in the future (83%). Most participants completed the scale with their mother in mind (57%), rated the potential care recipient’s current health as ‘fair’ (33%) or ‘good’ (30%), and their own health as ‘very good’ (39%) or ‘excellent’ (31%) Seventy-two percent of the sample reported never having provided care to someone else. See Table 1 for sample characteristics.
Procedure
Participants were recruited though the University of Colorado Colorado Springs’ (UCCS) Sona System (n = 202), a web-based software that allows UCCS students to serve as participants in research studies for extra credit, and Amazon’s Mechnical Turk (MTurk; n = 388), an online data collection service. Samples recruited through MTurk have been found to be comparable to community, in-person, and other online samples in
terms of race, gender, and quality of responses (Buhrmester, Kwang, & Gosling, 2011). Participants consented to participate in the study (Appendix A), after which they were screened to determine whether they currently have a living parent or in-law for whom they have not previously acted as an informal caregiver but who may need it in the future (Appendix B). Those who met this criterion continued to the survey, while those who do not did not complete the survey. In total, 590 participants completed the screening, 31 were not granted access to the following survey due to not meeting the criterion, and another 19 were excluded because, although they completed the screener and met criteria for inclusion, they chose not to continue on to the study questionnaires. Thus, the final sample consisted of 540 participants who were granted access to the study and first completed the Anticipatory Caregiving Scale (ACS; Appendix C). The presentation of the following scales were then counterbalanced after the ACS to avoid order effects: Preparedness for Caregiving Scale (PCS; Appendix D), Positive and Negative Affect Schedule (PANAS; Appendix E), The Cultural Justification for Caregiving Scale (CJCS; Appendix F), The Cost of Care Index (CCI; Appendix G), Scope of Care Scale (SCS; Appendix H), and demographics (Appendix I). Once completed, participants were provided with a brief debriefing (Appendix J). Participants recruited via Sona were compensated with extra credit for a chosen course while those recruited through MTurk were compensated with $3.00.
Measures
Demographics. Participants reported their age, gender, race, ethnicity, education level, employment status, and income. They also reported the relationship identity of the person about whom they have answered the scale (e.g., father, mother, father-in-law,
mother-in-law), whether they have identified as an informal caregiver to someone else, and if so, to whom.
Anticipatory Caregiving Scale. The ACS is a 36-item self-report measure designed to assess adult children’s attitudes and expectations about providing future care for a parent. Items such as “When I think about caregiving for this person, I feel
nervous,” are measured on a 6-point scale ranging from 1 (strongly disagree) to 6
(strongly agree). The full-scale score is computed by averaging all items, creating a range from 1 to 6, with higher scores indicating more positive attitudes towards or increased expectations for providing future care. The scale includes six subscales: Affect, Self-efficacy, Lifestyle Interference, Resources, Norms of Family Care, and Relationship Quality.
Affect. The Affect subscale is comprised of six items measuring the respondent’s feelings towards future caregiving, and includes items such as “When I think about caregiving for this person, I feel happy.” The items will be averaged after reverse-scoring items two, three, and four, creating a range from 1 to 6, with higher subscale scores indicating more positive feelings towards the future caregiving role.
Self-efficacy. The Self-efficacy subscale consists of six items measuring the respondent’s expectations about their ability to adapt to the caregiver role and perform well, and includes items such as “I will not have the necessary abilities to be a caregiver for this person.” The items will be averaged after reverse-scoring items 10, 11, and 12, creating a range from 1 to 6, with higher subscale scores indicating greater anticipation of the ability to adapt to and perform well in the caregiver role.
Lifestyle Interference. The Lifestyle Interference subscale consists of six items measuring if and how the respondent anticipates caregiving to impact their lifestyle, and includes items such as “I will be able to balance caregiving for this person with other aspects of my life.” The items will be averaged after reverse-scoring items 14, 16, and 18, creating a range from 1 to 6, with higher subscale scores indicating less anticipated lifestyle interference in the context of future caregiving.
Resources. The Resources subscale is comprised of six items measuring the respondent’s beliefs about available resources should they take on the caregiving role, and includes items such as “I will not have adequate time to be a caregiver for this person.” The items will be averaged after reverse-scoring items 20, 22, and 23, creating a range from 1 to 6, with higher subscale scores indicating more anticipated resources.
Norms of Family Care. The Norms of Family Care subscale is comprised of six items measuring the respondent’s endorsement of norms that may drive them towards caregiving, and includes items such as “It is my duty to provide care to this person.” The items will be averaged after reverse-scoring items 25, 27, and 30, creating a range from 1 to 6, with higher subscale scores indicating greater endorsement of norms or beliefs that drive them to provide care.
Relationship Quality. The Relationship subscale is comprised of six items measuring the respondent’s current and anticipated relationship with the potential care recipient, and includes items such as “Caregiving would strain my relationship with this person.” The items will be averaged after reverse-scoring items 32, 34, and 36, creating a range from 1 to 6, with higher subscale scores indicating better current and anticipated quality and satisfaction of the adult child-parent relationship.
Preparedness for Caregiving Scale. The PCS (Archbold et al., 1990) is an 8-item, self-report measure designed to assess people’s perceived ability to meet various needs of care recipients. Items such as “How well prepared do you think you are to take care of your family member’s physical needs?” are measured on a 5-point Likert scale ranging from 0 (not at all prepared) to 4 (very well prepared). The scale score is
computed by taking the mean of all 8 questions, creating a score range from 0 to 4, with higher cores indicating more perceived preparedness for caregiving. Validity and
reliability for the measure is strong, for example the Cronbach’s α ranging from .88 to .93 with (Carter et al., 1998; Hudson & Hayman-White, 2006). The original scale provides respondents with an opportunity to write in specific areas they wish through the inclusion of an open-ended ninth question which is not included in calculating the global score. The present study excluded this final question.
Positive and Negative Affect Schedule. The PANAS (Watson, Clark, &
Tellegen, 1988), is a 20-item, self-report measure that consists of two scales, the positive affect scale and the negative affect scale, designed to respectively assess current positive and negative emotions. Participants are presented with a list of words such as “excited” and “distressed,” and are asked to rate to what extent they are feeling that way at that moment. On a 5-point scale ranging from 1 (very slightly or not at all) to 5 (extremely). The positive affect scale score is computed by summing the 10 positive affect items, creating a range from 10 to 50, with higher scores indicating more positive affect, while the negative affect scale score is computed by summing the 10 negative affect items, creating a range from 10 to 50 with higher scores indicating more negative affect.
Validity and reliability for the measure is strong, with Cronbach’s α ranging from .84 to .87 (Vera-Villarroel et al., 2019; Watson, Clark, & Tellegen, 1988).
The Cultural Justification for Caregiving Scale. The CJCS
(Dilworth-Anderson et al., 2004) is a 10-item, self-report measure designed to evaluate caregivers’ cultural reasons for providing care to an aging loved one. Items such as “I give care because it is my duty to provide care to elderly dependent family members” are measured on a 4-point scale ranging from 1 (strongly disagree) to 4 (strongly agree). The scale score is computed by summing all items, creating a range from 10 to 40, with higher scores indicating stronger cultural reasons for providing care. Validity and reliability for the measure is strong, with Cronbach’s α ranging from .84 to .86 (Dilworth-Anderson et al., 2005; Dilworth-Anderson et al., 2004). For purposes of the present study, directions will be modified from “I give care because” to “I would give care because” because participants in the present study may not be current caregivers.
Cost of Care Index. The CCI (Kosberg & Cairl, 1986) is a 20-item, self-report measure designed as a case management tool to assess perceived potential or actual consequences of providing care to an aging family member. The scale assesses dimensions of cost, including personal and social restrictions, physical and emotional health, value, care recipient as provocateur, and economic costs. Items such as “I feel that my elderly relative will be an overly demanding person to care for” are measured on a 4-point scale ranging from 1 (strongly disagree) to 4 (strongly agree). The scale score is computed by summing all items, creating a range from 20 to 100, with higher scores indicating a higher perceived cost of providing care. Validity and reliability for the measure is strong, with Kosberg & Cairl (1986) finding Cronbach’s α of .91.
Scope of Care Scale. The SCS is a 13-item, self-report measure designed by the
author of the present study to capture respondent’s willingness to assist with individual ADLs or IADLs. ADLs include basic self-care tasks such as bathing and feeding, and IADLs include more complex self-care tasks such as managing finances and medication. Items such as “If needed, I would bathe this person” are measured on a 6-point scale ranging from 1 (strongly disagree) to 6 (strongly agree). The scale score is computed by averaging all items, creating a range from 1 to 6, with higher scores indicating a greater willingness to provide a broad scope of care. Items for the scale were adapted from the Physical Self-Maintenance Scale for ADLs and the Personal Self-Maintenance Scale for IADLs (Lawton & Brody, 1969).
RESULTS Descriptives
Descriptive statistics for participant demographics (see Table 1) and all scales and subscales were calculated (see Table 2). Mean scores and standard deviations for the ACS and its subscales are as follows: ACS total score (M = 4.06, SD = .70), Affect subscale (M = 3.41, SD = .97), Self-Efficacy subscale (M = 4.41, SD = 1.00), Lifestyle Interference subscale (M = 3.89, SD = .90), Resources subscale (M = 4.08, SD = .91), Norms of Family Care subscale (M = 4.04, SD = .74), and Relationship Quality subscale (M = 4.52, SD = .96). Scores on the total scale and subscales were relatively normally distributed, with skewness indices ranging from -.59 to .25.
Internal Consistency (H1)
Internal consistency reliability as estimated by Cronbach’s alpha for the ACS total score was excellent (α = .92), whereas internal consistency estimates for the subscales ranged from poor to good (Affect subscale α = .78, Self-Efficacy subscale α = .86, Lifestyle Interference subscale α = .74, Resources subscale α = .76, Norms of Family Care subscale α = .37, and Relationship Quality subscale α = .80). Generally, items demonstrated good item-total and inter-item correlations (see Table 3 and 4), with the exception of items 25 and 28 on the Norms of Family Care subscale, which both demonstrated poor item-total and inter-item correlations. Thus, items 25 and 28 were removed from the scale, changing slightly the overall means, standard deviations, and
Table 1 Demographic Descriptives Variable N (%) M SD Age 36.25 15.34 Number of Siblings 1.98 1.53 Gender Male 40.9 Female 57.8 Other 1.3 Race White 75.7 Black/African American 5.8 American Indian 1.1 Asian 12.5 Multiracial 3.4 Other 1.5 Ethnicity
Not Hispanic or Latino 84.2 Hispanic or Latino 15.8
Education
High school or less 7.9
Some college 43.6 Undergraduate degree 31.4 Graduate degree 17.2 Employment Employed 71.1 Unemployed 5.4 Retired 4.5 Student 17.0 Unable to work/disability 2.1 Income $0-15,000 29.9 $15,001-45,000 35.3 $45,001-75,000 20.1 $75,001+ 14.7 Relation to Participant
Table 1 (continued) Variable N (%) M SD Mother 57.1 Father 25.0 Mother-in-law 12.6 Father-in-law 5.2
Will you be primary CG?
Yes 82.5
No 17.5
When will you provide care?
5 years 30.5
6-10 years 24.9
11-15 years 20.9
20+ years 23.7
This person’s health
Excellent 9.1 Very good 19.2 Good 30.4 Fair 33.2 Poor 8.0 Your health Excellent 31.0 Very good 39.0 Good 23.5 Fair 5.6 Poor 0.9
Have you provided care to someone else?
Yes 27.6
No 72.4
Cronbach’s alpha for the ACS total score (M = 4.08, SD = .74, α = .93) and Norms of Family Care subscale (M = 4.12, SD = .92, α = .43). All analyses performed hereafter reflect the updated scale and subscales, omitting items 25 and 28.
Table 2
Means, Standard Deviations, and Cronbach’s alpha for Scales and Subscales
Variable M SD Cronbach’s α ACS 4.06 .70 .92 Affect 3.41 .97 .78 Self-efficacy 4.41 1.00 .86 Lifestyle Interference 3.89 .90 .74 Resources 4.08 .91 .76
Norms of Family Care 4.04 .74 .37 Relationship Quality 4.52 .96 .80 PCS 2.26 0.94 .93 PANAS Positive 29.51 9.83 .93 PANAS Negative 19.74 9.38 .94 CJCS 31.32 5.93 .87 CCI 44.13 13.39 .95 SCS 5.26 .71 .94
Variation in Responses by Demographics
Gender. An independent sample t-test was conducted to compare scores on the ACS whole-scale and subscales for male and female participants. Statistical assumptions for independent sample t-test were evaluated. Scores on the scales were relatively
normally distributed and homogeneity of variances was met utilizing Levene’s test for equality of variances. Statistically significant differences between males and females were found on the ACS whole-scale and Self-Efficacy, Lifestyle Interference, Resources, and Relationship Quality subscales, such that women reported higher mean scores than men (see Table 5). However, the effect sizes were small for each.
Race. An independent sample t-test was conducted to compare scores on the ACS whole-scale and subscales for White and Non-White participants because the sample size was too small to compare scores between specific racial/ethnic groups. Statistical
Table 4
ACS Item Descriptives & Item-Total Correlations
Variable M SD Item-total correlation Item 1 3.78 1.41 .46 Item 2 2.84 1.42 .38 Item 3 2.95 1.49 .43 Item 4 3.04 1.48 .56 Item 5 4.50 1.24 .55 Item 6 3.44 1.40 .45 Item 7 4.59 1.17 .63 Item 8 4.76 1.15 .57 Item 9 5.06 1.02 .59 Item 10 4.03 1.47 .70 Item 11 4.18 1.48 .67 Item 12 3.98 1.50 .60 Item 13 4.45 1.13 .32 Item 14 3.51 1.50 .67 Item 15 4.09 1.31 .52 Item 16 3.87 1.45 .57 Item 17 4.11 1.25 .57 Item 18 3.47 1.49 .45 Item 19 4.15 1.33 .49 Item 20 3.68 1.47 .44 Item 21 4.61 1.06 .42 Item 22 3.76 1.46 .59 Item 23 3.85 1.39 .61 Item 24 4.46 1.35 .49 Item 25 3.81 1.50 -.11 Item 26 4.71 1.14 .42 Item 27 4.06 1.58 .42 Item 28 3.65 1.36 .10 Item 29 4.08 1.60 .22 Item 30 3.97 1.77 .28 Item 31 4.58 1.28 .46 Item 32 5.18 1.11 .44 Item 33 4.75 1.49 .52 Item 34 4.22 1.43 .50 Item 35 4.51 1.27 .44 Item 36 4.04 1.48 .64
Table 5
ACS Scale and Subscale Differences Between Males and Females
Variable Female Male t p ƞ2
M (SD) M (SD) ACS 4.16 (.72) 3.99 (.76) 2.60 .01* .01 Affect 3.41 (.92) 3.44 (1.03) -.37 .71 <.001 Self-efficacy 4.56 (.95) 4.22 (1.03) 3.94 <.001* .03 Lifestyle Interference 3.98 (.86) 3.79 (.93) 2.36 .02* .01 Resources 4.16 (.89) 3.99 (.91) 2.18 .03* .01
Norms of Family Care 4.25 (.90) 4.11 (.94) 1.73 .08 .01 Relationship Quality 4.61 (.96) 4.42 (.94) 2.28 .02* .01 *p < .05
relatively normally distributed and homogeneity of variances was met utilizing Levene’s test for equality of variances for all but analyses between race and the Lifestyle
Interference and Resources subscales. Values for equal-variances-not-assumed were used in these two instances. Statistically significant differences between White and Non-White participants were found on the Affect, Self-Efficacy, and Relationship subscales, such that Non-White participants score higher than White participants on the Affect subscale and White participants score higher than Non-White participants on the Self-Efficacy and Relationship subscales (see Table 6). However, the effect sizes were small for each.
Age. A two-tailed Pearson correlation was performed to assess the relationship between participant age and scores on the ACS whole-scale and subscales. The statistical assumptions for Pearson correlation were evaluated, with scores for age and the ACS relatively normally distributed. Age was significantly negatively correlated with scores on the ACS whole-scale, and the Affect, Self-Efficacy, Lifestyle Interference, and
Table 6
ACS Scale and Subscale Differences Between White and Non-White Participants
Variable White Non-White t p ƞ2
M (SD) M (SD) ACS 4.09 (.76) 4.06 (.71) .29 .77 <.001 Affect 3.35 (.96) 3.63 (.94) -2.98 .003* .02 Self-efficacy 4.47 (1.00) 4.23 (.97) 2.40 .02* .01 Lifestyle Interference 3.88 (.92) 3.94 (.82) -.66 .53 <.001 Resources 4.08 (.96) 4.08 (.77) .02 .99 <.001
Norms of Family Care 4.20 (.93) 4.16 (.91) .40 .69 <.001 Relationship Quality 4.57 (.97) 4.38 (.93) 2.03 .04* .01 *p < .05
Relationship subscales, such that as age increased, scores on the scale and subscales decreased (see Table 7). However, the effect sizes were small for each.
Table 7
ACS Scale, Subscales, and Age Correlations Variable Age ACS -.14** Affect -.10* Self-efficacy -.09* Lifestyle Interference -.16** Resources -.07 Norms of Family Care -.07
Relationship Quality -.15** *p < .05, p < .01 **
ACS Whole-Scale and Subscale Correlation Matrix (H2)
A series of two-tailed Pearson correlations were performed to create a correlation matrix between the six hypothesized ACS subscales and the ACS whole-scale scores (see Table 8). The statistical assumptions for Pearson correlation were evaluated. As expected, there were strong, positive correlations between the ACS and each of the ACS subscales (Affect subscale r = .75, p < .001, Self-Efficacy subscale r = .86, p < .001, Lifestyle Interference subscale r = .87, p < .001, Resources subscale r = .80, p < .001, Norms of
Family Care subscale r = .61, p < .001, and Relationship Quality subscale r = .78, p < .001). Additionally, each subscale was significantly, positively correlated with the other subscales, ranging from r = .30 (Norms of Family Care subscale and Affect subscale) to r = .71 (Lifestyle Interference subscale and Self-efficacy subscale). These data indicate that the subscales share common variance ranging from 9% to 50%.
Convergent Validity (H3)
A series of two-tailed Pearson correlations were performed to assess the relationships among the ACS subscales and other measures (PCS, PANAS Positive & Negative, CJCS, CCI, SCS; see Table 8). The statistical assumptions for Pearson correlation were evaluated. Scores on the scales were relatively normally distributed, with skewness indices ranging from -1.24 to 1.05. As hypothesized, the ACS Affect subscale and PANAS Positive Scale formed a moderate, positive relationship (r = .48, p < .001). The ACS Affect subscale formed a small, negative correlation with the PANAS Negative Scale (r = -.25, p < .001). Results also found a moderate, negative relationship between the ACS Self-efficacy subscale and the PANAS Negative Scale (r = -.44, p < .001) as hypothesized, and a small, positive relationship between the ACS Self-efficacy subscale and the PANAS Positive Scale (r = .25, p < .001). Additionally, as
hypothesized, the ACS Lifestyle Interference (r = -.69, p < . 001), Relationship Quality (r = -.69, p < . 001), and Resources (r = -.62, p < .001) subscales each formed strong, negative correlations with the CCI. Finally, as hypothesized, results found a strong, positive relationship between the ACS Norms of Family Care subscale and the CJCS (r = .58, p < .001).
Discriminant Validity (H4)
To assess discriminant validity, a series of two-tailed Pearson correlations were performed to assess the relationship between the ACS Norms of Family Care subscale and the PANAS Negative Scale, the ACS Norms of Family Care subscale and the PANAS Positive Scale, the ACS Resources subscale and the PANAS Positive scale (see Table 8). The statistical assumptions for Pearson correlation were evaluated. The Norms of Family Care, Resources, and PANAS Positive and Negative scales were all relatively normally distributed with skewness indices ranging from -.36 to 1.05. In contrast to hypotheses, the PANAS Positive and Negative subscales correlated with ACS subscales. Results found a significant, small, negative correlation between the Norms of Family Care and PANAS Negative scales (r = -.22, p = <.001), and a significant, small, positive correlation between the Norms of Family Care and PANAS Positive scales (r = .14, p < .001). Results also found a significant, medium, negative correlation between the Resources and PANAS Negative scales (r = -.42, p = <.001), and a significant, small, positive correlation between the Resources and PANAS Positive scales (r = .27, p = <.001).
Concurrent Validity (H5)
To assess concurrent validity, a two-tailed Pearson correlation was performed to assess the relationship between the ACS and the PCS, as well as between the ACS and the SCS (see Table 8). The statistical assumptions for Pearson correlation were evaluated. The ACS, PCS, and SCS were relatively normally distributed with skewness indices of -.21, -.39, and -1.24 respectively. As hypothesized, results showed strong, positive relationships between the ACS and the PCS (r = .58, p = <.001), and the ACS and the
SCS (r = .49, p = <.001) demonstrating good concurrent validity. Additionally, an independent samples t-test was conducted to compare scores on the ACS whole-scale for participants who reported anticipating being the primary caregiver in the future (M = 4.15, SD = .73) versus those who did not (M = 3.77, SD = .76). Participants who reported anticipating being the primary caregiver in the future scored higher on the ACS than those who did not anticipate becoming the primary caregiver, t(534) = 4.59, p < .001.
