218 GENDER SOCIETY/April 2003 Level of financial assistance was measured by the amount of money caregivers spent on food,clothing,medicine,doctor's visits,home care,and hospitalization for one's relative.Respondents used a five-point response set ranging from the elder pays it all to we pay it all to answer these questions.The Cronbach's alpha for financial assistance was.98. Independent variables.Three sets of independent variables were included in the regression analysis:caregiver/care recipient attributes,the cultural aspects of xiao. and the structural constraints of family resources.The first set of variables included the gender of the caregiver and four control variables representing characteristics of the care recipient:the gender and marital status of the care recipient and two mea- sures of the elder's functional level.The elder's need for assistance with ADL was measured with four items using a five-point response set ranging from 1 =no help needed to 5=cannot perform the activity.The items included using the toilet,dress- ing,moving in and out of the bed and chairs,and eating.The mean score was 2.96, and the internal reliability alpha for ADL was.94.The elder's need for assistance with IADL was measured with eight items:bathing,doing light housework,doing heavy housework,laundry,shopping,mobility outside,transportation,and money management.The mean score was 4.19,and the Cronbach's alpha level was.96. Two measures were constructed to capture the cultural attitudes associated with xiao.These measures of patrilocal norms and social pressure were constructed from items adapted from measures developed by Montgomery(1996),Gallois et al. (1996),and Choi (1993)and were adjusted to correspond to the Chinese context. Caregivers were asked to respond by using a five-point scale ranging from com- pletely disagree to completely agree.The three items included in the measure of patrilocal norms reflect a traditional cultural norm for elder care under the patri- local family structure,which may differ from the contemporary urban practice of elder care."The three items included in the measure of social pressure reflect a per- son's psychological or emotional acknowledgment of peer pressure as the major impetus for provision of care.The Cronbach's alphas for the measures of patrilocal norms and social pressure were.73 and.70,respectively. The third set of independent variables was intended to capture the structural con- straints placed on the family resources.These included measures of the caregiver's monthly income and two dichotomous variables indicating the employment status of the caregiver and whether the elder received a pension.A measure reflecting the elder's receipt of medical benefits was originally included in the study but was omit- ted from this analysis due to its high correlation with receipt of a pension(r=.72). Family size was measured as the number of family members available to be involved with care for the dependent elder.Other than assistance from a dependent elder's living spouse,family assistance virtually always refers to help from the caregivers'siblings.There were twoexceptions in this sample:One woman assisted briefly with the care of her daughter's mother-in-law,and another granddaughter provided most of the financial assistance for the care of her grandmother. Downloaded from gas.sagepub.com at UCSF LIBRARY CKM on January 27.2015Level of financial assistance was measured by the amount of money caregivers spent on food, clothing, medicine, doctor’s visits, home care, and hospitalization for one’s relative. Respondents used a five-point response set ranging from the elder pays it all to we pay it all to answer these questions. The Cronbach’s alpha for financial assistance was .98. Independent variables. Three sets of independent variables were included in the regression analysis: caregiver/care recipient attributes, the cultural aspects of xiao, and the structural constraints of family resources. The first set of variables included the gender of the caregiver and four control variables representing characteristics of the care recipient: the gender and marital status of the care recipient and two mea￾sures of the elder’s functional level. The elder’s need for assistance with ADL was measured with four items using a five-point response set ranging from 1 = no help needed to 5 = cannot perform the activity. The items included using the toilet, dress￾ing, moving in and out of the bed and chairs, and eating. The mean score was 2.96, and the internal reliability alpha for ADL was .94. The elder’s need for assistance with IADL was measured with eight items: bathing,1 doing light housework, doing heavy housework, laundry, shopping, mobility outside, transportation, and money management. The mean score was 4.19, and the Cronbach’s alpha level was .96. Two measures were constructed to capture the cultural attitudes associated with xiao. These measures of patrilocal norms and social pressure were constructed from items adapted from measures developed by Montgomery (1996), Gallois et al. (1996), and Choi (1993) and were adjusted to correspond to the Chinese context. Caregivers were asked to respond by using a five-point scale ranging from com￾pletely disagree to completely agree. The three items included in the measure of patrilocal norms reflect a traditional cultural norm for elder care under the patri￾local family structure, which may differ from the contemporary urban practice of elder care.2 The three items included in the measure of social pressure reflect a per￾son’s psychological or emotional acknowledgment of peer pressure as the major impetus for provision of care.3 The Cronbach’s alphas for the measures of patrilocal norms and social pressure were .73 and .70, respectively. The third set of independent variables was intended to capture the structural con￾straints placed on the family resources. These included measures of the caregiver’s monthly income and two dichotomous variables indicating the employment status of the caregiver and whether the elder received a pension. A measure reflecting the elder’s receipt of medical benefits was originally included in the study but was omit￾ted from this analysis due to its high correlation with receipt of a pension (r = .72). Family size was measured as the number of family members available to be involved with care for the dependent elder. Other than assistance from a dependent elder’s living spouse, family assistance virtually always refers to help from the caregivers’siblings. There were two exceptions in this sample: One woman assisted briefly with the care of her daughter’s mother-in-law, and another granddaughter provided most of the financial assistance for the care of her grandmother. 218 GENDER & SOCIETY / April 2003 Downloaded from gas.sagepub.com at UCSF LIBRARY & CKM on January 27, 2015