Psychology and Aging 1996, Vol. 11, No. 2, 377-380
Copyright 1996 by the American Psychological Association, Inc. 0882-7974/96/$3.00
Response Biases as a Confound to Expressed Burden Among Spousal Caregivers of Suspected Dementia Patients Norm O'Rourke
Beth E. Haverkamp
Vancouver Hospital and Health Sciences C e n t r e - University of British Columbia Site
University of British Columbia
Sandra Rue
Holly Tuokko
Simon Fraser University
University of Victoria
Sherri Hayden and B. Lynn Beattie Vancouver Hospital and Health Sciences Centre--University of British Columbia Site Burden among family caregivers has yet to be adequately assessed relative to social desirability response sets. This study examined the impact of two different response biases relative to expressed burden among spousal caregivers of suspected dementia patients. In addition to individual social desirability, it was hypothesized that a distinct social desirability construct exists as a function of one's perception of self within relationships. As predicted, a significant inverse association was apparent between expressed burden and this systemic response bias (as measured by the Edmonds Marital Conventionality Scale), whereas this association was not apparent with a measure of individual social desirability (the Marlowe-Crowne Social Desirability Scale). This findingsuggeststhat systemic social desirability may exist as a significantresponse bias.
psychological processes. By failing to consider the possible influence of socially desirable responding, the existing research therefore implies that the degree of distress perceived by informal caregivers is fully assessed by existing self-report burden measures. The role of response biases has received considerable attention relative to other negative psychological states such as depression and hopelessness. Though correlated with these phenomena ( O'Rourke, 1995; Zarit, 1990), burden is distinct from these constructs that pertain exclusively to the individual. Burden, in contrast, is inextricably grounded within the caregiving context and the caregiver's perceptions of his or her relative. This thereby distinguishes expressed burden as a systemic construct. Use of this term is not intended to imply that caregiver burden is necessarily reciprocal (i.e., the common meaning within systems theory), but rather that it is based within a system or a relationship. We hypothesized that this distinction may also delineate a discrete social desirability construct. More precisely, we hypothesized that the perception of self in relationship is distinct from individual self-awareness. Various factors may lead to a unique self-definition in relation to others, which implies that distinct psychological processes might impede full disclosure of one's perception of self in relationship compared with one's singular perception of self. For example, older women have been socialized to define themselves primarily as wives and mothers (Friedan, 1993). Because this cohort was also held largely responsible for the stability of marriage and the family, distinct psychological processes may confound the reporting of women's beliefs and behaviors within these systems. In other words,
As noted by Zarit (1990), family caregiving is among the most widely researched topics in gerontology. To a large degree, this has been led by the study of burden among dementia caregivers. Despite the breadth of study in this area, no work to date appears to have considered the role of response biases relative to expressed burden. This seems to imply that burden among caregivers is immune to self-censorship and other confounding
Norm O'Rourke, Sherri Hayden, and B. Lynn Beat-tie,Clinic for Alzheimer Disease and Related Disorders, VancouverHospital and Health SciencesCentre--University of British Columbia Site, Vancouver,British Columbia, Canada; Beth E. Haverkamp, Department of Counselling Psychology,Universityof British Columbia, Vancouver,British Columbia, Canada; Sandra Rae, Department of Mathematics and Statistics, Simon Fraser University, Vancouver, British Columbia, Canada; Holly Tuokko, Centre on Aging,Universityof Victoria, Victoria, British Columbia, Canada. Data for this study were drawn from the master's thesis of Norm O'Rourke, which was supported, in part, by the Norcen/Superior Propane--Canadian Association on Gerontology Community Researcher Award. We would like to acknowledgethe assistance of the staff at the Clinic for Alzheimer Diseaseand Related Disorders, especiallyDaniel Corrin, who kindly volunteered to interview participants for this project. Most important, however,we wish to express our appreciation to the men and women who graciouslyagreed to take part in this study. Correspondence concerning this article should be addressed to Norm O'Rourke, who is now at the Centre for Health Services and Policy Research, University of British Columbia, #429-2194 Health Sciences Mall, Vancouver,British Columbia, Canada V6T 1Z3. Electronic mail may be sent via Internet to
[email protected]. 377
378
BRIEF REPORTS
Table 1
Descriptive Features of Patient and Caregiver Samples Feature
N
M
SD
Patient age Years married Total family income (Canadian dollars) Duration (in months) of symptoms FRS (severity) total Caregiver age Caregiver BHS total Caregiver BI total Caregiver EMCS total Caregver MC-SDS total
70 70 69 70 70 70 70 69 70 70
69.3 36.7 $38,080 54.2 22.0 66.2 4.2 27.2 6.9 20.4
8.48 14.66 22,100 35.51 5.52 9.75 3.31 15.90 4.07 5.17
Note. FRS = Functional Rating Scale; BHS = Beck Hopelessness Scale; BI = Burden Interview; EMCS = Edmonds Marital Conventionality Scale; MC-SDS = Marlowe-Crowne Social Desirability Scale.
distinct psychological processes may cause these w o m e n to defensively m o n i t o r expressions relative to their relationships, whereas they may experience less reticence w h e n reporting negative self-perceptions. Should this be the case, it is reasonable to speculate t h a t a distinct response bias, here t e r m e d a systemic response bias, may i m p e d e candid responding on relational measures. This rationale f o r m s the basis o f the present study. We hypothesized t h a t a systemic response bias exists that is distinct from individual f o r m s o f socially desirable responding. Furthermore, we hypothesized t h a t this systemic c o n s t r u c t would have a significant inverse relationship with expressed b u r d e n a m o n g spousal caregivers a n d t h a t this inverse relationship would n o t exist between expressed b u r d e n a n d caregivers' individual response biases.
spondents if they would endorse 33 statements regarding uncommon or undesirable behaviors. We selected the Edmonds Marital Conventionality Scale (EMCS; Edmonds, 1967) for this study to assess the hypothesized systemic response bias. As defined by Edmonds (1967), conventionalization refers to an overly positive appraisal of one's relationship as a function of social norms. Spouses are asked to indicate whether 15 statements pertaining to their partner and relationship are true or false (e.g., "If my spouse has any faults, I am not aware of them"; "I have never regretted my marriage, not even for a moment"). As with more traditional measures, higher totals on the EMCS are believed to indicate higher levels of socially desirable responding. It is believed that those responding higher on the EMCS are less likely to respond candidly on other measures pertaining to their spouse and marriage. To obfuscate the instrument's intent, we mixed the items of the EMCS with four items from the Marital Status Inventory (Weiss & Cerreto, 1980) that do not measure social desirability. These items were not included in the EMCS total scores (i.e., the maximum score remained 15 points). The Beck Hopelessness Scale ( BHS; Beck & Steer, 1988) was selected for this study as a distinct measure of pessimism. Other research conducted at this facility suggests that hopeless ideation is strongly related to caregiver burden (O'Rourke, Haverkamp, Tuokko, Hayden, & Beattie, 1995 ). The BHS presents respondents with a series of 20 statements regarding perceptions of self and one's future ( 11 positive and 9 negatively keyed items). Beck and Steer (1988), concluded that the BHS appears to possess adequate reliability and validity among persons 13 to 80 years. The Functional Rating Scale (FRS; Tuokko, Crockett, Beattie, Horton, & Wong, 1986) is a multidimensional scale designed to quantify patients' level of cognitive and functional impairment along a series of eight 5-point scales (Memory, Social/Occupational Functioning, Home and Hobbies, Problem Solving, Personal Care, Affect, Language, and Orientation). This scale provides a rating of full severity while allowing for measurement of uneven rates of decline across various activities of daily living.
Procedure Method
Participants Seventy spousal caregivers (47 women and 23 men), living in the community with their care recipients, were recruited for this study at the Clinic for Alzheimer Disease and Related Disorders, Vancouver Hospital and Health Sciences Centre--University of British Columbia Site. Participants were consecutively interviewed over an 1 l-month period as part of a larger study conducted by N. O'Rourke ( 1995 ). Patients from all regions of British Columbia are referred to this tertiary diagnostic facility by community physicians. As part of a standard 2-day assessment, both patients and their primary caregivers are seen by various professionals within this multidisciplinary clinic, including geriatric medicine, genetics, neuropsychology, speech pathology, psychiatry, and social work. The sample is described in Table 1.
Measures Participants were administered the Burden Interview (BI; Zarit & Zarit, 1990), a measure specifically developed for caregivers of dementia patients. This 22-item instrument was designed to assess the extent to which careglvers view this role as hav!ng an adverse impact on their health, finances, social life, and emotional well-being. As noted by Paulhus and Reid ( 1989 ), the Marlowe-Crowne Social Desirability Scale (MC-SDS; Crowne & Marlowe, 1960) remains the most widely used measure of response biases. This instrument asks re-
The B1 was administered to caregivers by a social worker who was blind to the hypotheses of this study. All other measures were administered by N. O'Rourke or a student volunteer who was also blind to the hypotheses of this study. A series of comparisons analyzing responses derived on all dependent measures by both interviewers revealed no significant differences (O'Rourke, 1995 ), which suggests that interviewer bias did not contaminate the results of this study. Results A two-factor analysis o f variance was p e r f o r m e d to test the p r i m a r y hypotheses o f this study. Participants were grouped w i t h i n t h i r d s on the basis of response distributions on b o t h social desirability measures separately a n d in c o m b i n a t i o n (e.g., the b o t t o m t h i r d o f the d i s t r i b u t i o n on b o t h the M C - S D S a n d the E M C S ) . O n the basis o f the hypothesis t h a t a systemic response bias is separate a n d distinct from individual response biases, we e x a m i n e d social desirability constructs for a potential interaction with i n h i b i t e d responding on the BI. We a s s u m e d t h a t a n additive effect m i g h t emerge between the M C - S D S a n d the EMCS. T h o u g h a significant m a i n effect for the E M C S was evident on the BI, F ( 2 , 59) = 4.79, p < .01, there was n o m a i n effect for the M C - S D S on the BI, F ( 2 , 59) = 0.12, p = .89. Moreover, n o interaction effect between the two social desirability m e a s u r e s emerged, F ( 4 , 59) = 1.26, p = .30.
BRIEF REPORTS
379
Table 2
Correlation Coefficients and Hierarchical Regression Analysis of Relationship Between Social Desirability and Other Variables on Burden Interview Scores Variable
BI (DV)
BHS
BHS FRS Dur Sx MC-SDS
0.36*** 0.32** 0.28* -0.09
-0.20 0.14 -0.12
EMCS
-0.39****
0.25*
FRS
Dur Sx
0.24* -0.01
-0.09
0.04
-0.26*
MC-SDS
0.26*
B
SE B
1.13 0.90 -0.01 0.01
0.60 0.30 0.06 0.33
0.21 0.33*** -0.01 0.00
-1.29
0.40
-0.36***
R 2 = .23 for initial set of variables; AR 2 = . 11 subsequent to entry of EMCS within the regression equation (ps < .005). Multivariate outliers were identified using Mahalanobis distance. With df= 5, a chisquare (x.299)value of 15.1 necessitated the removal of three cases prior to regression analysis. With removal of outliers and three missing BI scores, revised N = 64. BI = Burden Interview; DV = dependent variable; BHS = Beck Hopelessness Scale; FRS = Functional Rating Scale; Dur S~ = duration of dementia symptoms; MC-SDS = Marlowe-Crowne Social Desirability Scale; EMCS = Edmonds Marital Conventionality Scale. *p