| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Address correspondence to Perry Edelman, PhD, Director of Outcomes Research, Mather LifeWays Institute on Aging, 1603 Orrington Avenue, Suite 1800, Evanston, IL 60201. E-mail: pedelman{at}matherlifeways.com
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Assisted living • Resident satisfaction • Family member satisfaction • Surveys • Confirmatory factor analysis
The term assisted living refers to a number of different types of residential care facilities (adult day care homes, adult foster care homes, personal care homes, etc.) as well as to a philosophy of care. Consumers need clear information regarding what services assisted living provides, as well as the impact or benefits of those services. Only by assessing assisted living outcomes and making that information public can consumers make informed decisions regarding the appropriateness of assisted living generally, and the appropriateness of specific assisted living facilities in meeting their needs.
Although the philosophy of assisted living encourages providers to try to develop a more consumer-oriented approach to residential care for older people, there is little empirical data regarding the degree to which assisted living facilities live up to aspirations of increasing resident autonomy, providing a more homelike experience, and enabling residents to age in place. Similarly, the impact that these elements have on resident satisfaction with assisted living is largely unknown. The dearth of information not only presents a formidable obstacle for potential consumers, but it also denies policy makers and providers the opportunity to improve services and promote the highest quality of life possible for assisted living residents.
In this study, we surveyed residents and family members to explore their experiences with, and perspectives regarding, critical features of assisted living. In addition, we examined the circumstances of residents' lives prior to the move to assisted living, as well as the impact of the move itself. We used the findings from these surveys to develop measures of resident and family member satisfaction with assisted living.
Literature Review
Assisted living is a total living environment that includes housing and supportive services as well as opportunities for socialization. Therefore, it is reasonable to expect that a variety of factors determine satisfaction with assisted living; those factors include both subjective (e.g., feelings of autonomy, health perceptions, life satisfaction, self-esteem, and sense of control) and objective elements (e.g., physical/cognitive health and functioning, economic circumstances, and environmental factors). This article focuses on subjective aspects of quality of life from the viewpoint of assisted living residents and their family members.
Researchers have studied and debated consumer satisfaction with long-term care. The literature on consumer satisfaction with long-term-care settings varies widely, with some articles focusing on defining the relevant aspects of consumer satisfaction (e.g., security, comfort, and meaningful activity [R. A. Kane, 2001]) and others focusing on aspects of satisfaction within specific long-term-care settings (e.g., nursing homes [Glass, 1991; Harrington et al., 1999] and continuing care retirement communities [Moran, White, Eales, Fast, & Keating, 2002]). In their review of the literature, R. L. Kane and Kane (2001) found interpersonal qualities of caregivers, private accommodations, and control/choice of aspects of their daily lives to be important factors associated with resident satisfaction. Early findings from R. A. Kane's research (2001) proposed that relationships, privacy, and autonomy also were associated with good quality of life for consumers.
Sikorska (1999) examined the relationships between organizational factors and resident satisfaction with assisted living. She found that, although some facility characteristics were important predictors of satisfaction with assisted living (e.g., size, physical amenities, socio-recreational programming), the resident's functional ability and his or her participation in the decision to move to assisted living also were important. Mitchell and Kemp (2000) found that resident satisfaction was predicted by high cohesion (the extent to which staff members were helpful and supportive of residents and the extent to which residents were supportive of each other), low conflict (extent to which residents expressed anger and were critical of each other and of the facility and staff), and fewer chronic health conditions of the resident.
Applebaum, Straker, and Geron (2000) suggested that attention to consumer choice in health and long-term care has led to increased attention to customer satisfaction. Some researchers have been successful in identifying specific correlates of satisfaction. Buelow and Fee (2000) assessed the mealtime experience, nursing assistants, and recreational activities. They found that preferred qualities of nursing assistants identified by both residents and family members were genuine concern, kindness, respect, and consistent attentiveness. Residents identified a pleasant disposition as an essential quality of nursing assistants, and family members identified knowledge regarding aging, gentle assertiveness, and commitment to staying on the job as essential qualities of nursing assistants. Gesell (2001) found that the assistance and care provided, staff cooperation with one another and with residents, confidence in staff, and evidence of competent management were strongly associated with family member satisfaction.
In summary, a variety of factors influence resident satisfaction with assisted living. These factors include participation in decision making regarding the move to assisted living, resident knowledge about and experience with assisted living services, and the quality of staff and resident interactions, as well as the quality of assisted living operations and management. The literature on satisfaction also identifies the strong emphasis on independence, privacy, autonomy, and normalization (i.e., homelike) espoused in definitions and philosophy of assisted living (Hawes, Phillips, & Rose, 2000). We intend for Figure 1 to serve as a framework for conceptualizing psychosocial predictors of resident satisfaction with assisted living. We have loosely based this framework on the expansion of the Andersen model of health care utilization to include psychosocial factors, such as attitudes, knowledge, and perceived control (Bradley et al., 2002).
|
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Revised drafts underwent an intensive pilot at two assisted living facilities, one a rural, low-income facility in Indiana and the other a moderate- to high-income facility in a Chicago suburb. The pilot-testing included a debriefing with participants in which they identified and discussed items that they did not understand or had difficulty answering. Pilot-testing confirmed that respondents had little difficulty completing the survey in 20 to 30 minutes. We made minor revisions in wording to some items prior to administering the survey to the larger sample.
Resident dimensions included: (a) Safety/Peace of Mind—residents' ability to get assistance with health problems and the peace of mind that knowledge afforded them; (b) Personal Attention—the belief that residents received sufficient staff time and attention to meet their needs; (c) General Satisfaction—satisfaction with living in the facility and the extent to which residents would recommend the facility to others; (d) Staff—staff behavior vis-à-vis residents; (e) Residents—opinions of other residents and the how their behavior impacted the respondent; (f) Knowledge—understanding of resident and facility responsibilities; (g) Autonomy—ability to be independent and make choices; (h) Aides—relationship and satisfaction with aides; (i) Socialization With Family—satisfaction with family interaction; (j) Transportation—satisfaction with transportation service; (k) Privacy—assessment of ability to maintain privacy vis-à-vis staff and other residents; and (l) Activities—satisfaction with activities.
Five family member dimensions were similar to resident dimensions: Staff Responsiveness, Safety, Transportation, Activities, and Resident Responsibilities (similar to the resident Knowledge dimension). The sixth dimension—Family Member Impact—focused on the effect on family members of the resident's move to the assisted living facility. Additional data collected relevant to resident and family member characteristics included gender, age, ethnicity, marital status, living arrangement, education, and resident physical and mental health (Ware, Snow, Kosinski, & Gandek, 1993).
Eleven rural, suburban, and urban assisted living facilities in Illinois and Indiana participated in this study. Of the 365 surveys mailed to residents, 241 returned usable data (a 66% return rate). Because they were recruited by facility staff, the return rate for family members is unknown. Following initial item analysis and elimination of specific cases due to missing data, the final sample sizes for the analyses were 204 residents and 232 family members. The median number of residents and family members that participated across the 11 sites was 21 and 18, respectively.
Respondent Characteristics
Table 1 presents demographic information for residents and family members. Residents were White (100.0%), widowed (72.1%), and female (77.2%), with a mean age of 85.6 (SD = 7.1 years); nearly two thirds had no more than a high school education (62.4%). Family members were similar to residents in that most were White (99.1%) and female (61.0%); however, most family members were married (75.3%), had at least some college education (71.5%), and had a mean age of 60.3 (SD = 12.8 years).
|
We used four different measures to assess model fit: the root mean square error of approximation (RMSEA)(Steiger, 1990), the goodness-of-fit index (GFI)(Joreskog & Sorbom, 1996), the comparative fit index (CFI)(Bentler, 1990), and the Tucker–Lewis coefficient (TLC)(Tucker & Lewis, 1973). RMSEA reflects the absolute size of the residuals that result when using the model to predict the data, adjusting for model complexity, with smaller values indicating better fit. According to Browne and Cudeck (1993), RMSEA 
.05 represents close fit; RMSEA.05–.08 represents reasonably close fit; and RMSEA >.10 represents an unacceptable model. Analogous to R2 in multiple regression, GFI reflects the proportion of available variance–covariance information in the data that the given model explains, with larger GFI values representing better model fit. CFI and TLC, in contrast, each adjust for model complexity using a different formula to gauge how much better the given model fits the data relative to a null model, in which sampling error alone is presumed to explain the covariation among observed measures. Bentler and Bonett (1980) suggested that formal measurement models have a GFI, CFI, and TLC greater than.90.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
2(674, n = 204) = 1290.47, RMSEA =.066, GFI =.76, CFI =.82, TLC =.79. Although the model's RMSEA suggested reasonably close fit and all 12 intended subscales showed acceptable internal consistency reliability (median coefficient alpha =.74), measures of absolute and relative fit were all below the.90 threshold of acceptability.
A Multifaceted Unidimensional Model of Resident Satisfaction
Inspection of the CFA solution revealed that the 12 satisfaction factors were strongly correlated (median factor intercorrelation =.60), and an exploratory principal factor analysis of the 40 satisfaction items revealed a single dominant factor. We therefore recast the original 12-factor model in the form of a single overarching global satisfaction factor reflected in 12 facets of resident satisfaction. Each lower-order facet consisted of a small, reliable subset of items, which we used as measured indicators of a single high-order latent variable labeled resident satisfaction. Although we intended for this multifaceted 1-factor model to encompass the various content areas originally assessed, we reconceptualized these various domains as reflecting a single, higher-order construct.
Investigators have termed this type of structural approach a partial aggregation model (Bagozzi & Edwards, 1998; Bagozzi & Heatherton, 1994). In a partially aggregated measurement model, researchers combine (i.e., average) responses to multiple items measuring a specific subscale in order to form a composite measure. Researchers then use the summary scores for multiple composite measures as indicators in the measurement model and analyze these indicators together to examine their underlying hierarchical structure.
Constructing Reliable Facet Subscales
Based on the earlier exploratory factor analysis, we omitted the Transportation subscale, given its relative independence in relation to the other subscales. Because we now conceptualized the measurement model as reflecting a single, global assessment of resident satisfaction consisting of multiple facets, we also eliminated the General Satisfaction subscale. This left a total of 10 facets for the unidimensional measurement model.
In constructing reliable facet subscales, we focused on two objectives. First, we wanted to minimize the total number of questionnaire items in order to make the instrument easier for respondents to complete. Rather than limiting the number of facets contained in the model, we sought instead to minimize the number of constituent items contained in each facet subscale by searching for the most highly correlated pair of items within each facet subscale. Second, because subscales with reliabilities less than.60 are less likely than subscales with higher reliabilities to be useful as indicators in structural models (Kline, 1998), we wanted each 2-item facet subscale to have an internal consistency reliability of at least.60. For 9 of the 10 facets examined, we were able to identify a single pair of items that had a Cronbach's alpha reliability coefficient 
.60. The only facet for which we were unable to find a sufficiently reliable item-pair was Residents, which initially had only 2 items (r =.41, 
=.56). To create measured indicators for CFA modeling, we constructed mean scores for each facet subscale by averaging the 2 constituent items for each subscale. Table 2 presents the final set of 9 facet subscales, along with their Cronbach's alphas and subscale intercorrelations.
|
2(27, n = 204) = 55.42, RMSEA =.081, GFI =.94, CFI =.97, TLC =.96, and explained between 16% and 70% of the variance in the individual subscales (median R2 =.51). Table 2 also presents the completely standardized factor loadings and squared multiple correlations for the 9 subscales constituting the 1-factor model. All factor loadings were strong and statistically different from zero (all ps <.00001). To assess second-order factor reliability, we constructed a global total score by summing responses to the 9 subscale scores constituting the 1-factor model. Cronbach's alpha for the 9-subscale total score was.87. Thus, the 18 self-report items (scored as 9 facet subscales) appeared to provide a good-fitting, reliable 1-factor model for assessing resident satisfaction.
Assessing Convergent Validity
If the facet subscales were truly indicators of a single latent variable reflecting global resident satisfaction, we would have expected them to be significantly correlated with a general measure related to resident satisfaction. Although the 9 facet subscales had not been designed strictly to measure general satisfaction, they did reflect resident general satisfaction to a certain extent. Therefore, as an initial test of convergent validity, we assessed the degree of association between total score on the 1-factor model of 9 facet subscales and the original 4-item subscale designed to tap General Satisfaction. We constructed a 1-factor total score by summing the 9 facet subscale scores for each respondent. Supporting the convergent validity of the multifaceted unidimensional model, total score was positively correlated with the 4-item General Satisfaction subscale, r =.65, p <.0001.
Providing further support for convergent validity, each of the individual facet subscales also showed a positive correlation with the General Satisfaction subscale: Safety/Peace of Mind (r =.55, p <.0001), Personal Attention (r =.52, p <.0001), Staff (r =.53, p <.0001), Knowledge (r =.52, p <.0001), Aides (r =.41, p <.0001), Privacy (r =.60, p <.0001), Activities (r =.47, p <.0001), Autonomy (r =.36, p <.0001), and Socialization With Family (r =.26, p <.0001). As a set, the 9 facet subscales explained nearly half of the variance in General Satisfaction, R2 =.45. Further evidence of construct validity was provided by the low correlations of General Satisfaction with Autonomy and Socialization With Family. We expected Socialization With Family items to have only an indirect relationship to general satisfaction with the assisted living facility, and Autonomy items to be related to resident expectations and responsibilities, not satisfaction. Thus the 9 facet subscales composed the ALRSS.
Assisted Living Family Member Satisfaction Scale
A Priori Model
The a priori measurement model consisted of 6 correlated factors of family member satisfaction: Staff Responsiveness (3 items), Safety (5 items), Transportation (3 items), Activities (5 items), Family Member Impact (5 items), and Resident Responsibilities (4 items). This 6-factor structure provided only a marginal goodness of fit for the initial set of 25 items, 2(260, n = 232) = 498.91, RMSEA =.061, GFI =.86, CFI =.87, TLC =.85. Although the model's RMSEA suggested reasonably close fit and each of the 6 intended subscales showed acceptable internal consistency reliability, GFI, CFI, and TLC were all below the.90 threshold of acceptability (median fit index =.86), signifying room for improvement in model fit. Before refining the 6-factor model, we first tested whether family member satisfaction was better conceptualized as being multidimensional or unidimensional. In particular, we contrasted the goodness of fit of the a priori 6-factor model with that of a 1-factor model, which assumed family member satisfaction was unidimensional. If satisfaction were truly multidimensional, then the 6-factor model should have fit the data better than a 1-factor model. Supporting multidimensionality, the 1-factor model fit the 25-item dataset poorly, 2(275, n = 232) = 1081.01, RMSEA =.125, GFI =.70, CFI =.56, TLC =.52, and fit significantly worse than did the intended 6-factor model, 
2(15, n = 232) = 582.1, p <.0001.
Refining the A Priori Measurement Model
In the second stage of the analysis, we used CFA as a model generating tool (Joreskog, 1993) to refine the a priori 6-factor structure in order to develop an acceptable measurement model of family member satisfaction. The goal was to develop a model that not only fit the data well and had reliable subscales, but also had a clear and meaningful substantive interpretation in relation to our a priori conceptual framework. To refine the a priori model, we used CFA (a) to find particular dimensions of family member satisfaction that could be combined or eliminated to create a smaller, more parsimonious model; and (b) to pinpoint items that could be dropped in order to improve model fit, based on the proportion of variance that the underlying factor explained in responses to the particular item (i.e., the item's squared multiple correlation, R2).
Because (a) we did not need to minimize the number of questions family members completed and (b) the minimum number of available items per factor for family members was 3, we increased the minimum number of indicators per factor to 3 for the family member instrument (compared with the 2 items per factor used for the resident instrument). In deciding on the maximum number of indicators per factor, we chose to retain no more than 5 indicators. As Bagozzi and Heatherton (1994) noted, measurement models containing more than 4 to 6 indicators per factor are unlikely to fit the data satisfactorily.
In refining the initial CFA model to achieve acceptable fit, we followed an iterative modeling procedure (Brockway, Carlson, Jones, & Bryant, 2002) in which we combined highly intercorrelated factors (r >.90), omitted from each factor the items with the lowest squared multiple correlations (R2s), and assessed the fit of this respecified model repeatedly until the modified model achieved acceptable goodness of fit and each of the subscales was reasonably reliable ( .60).
After several iterations of model respecification, a 5-factor model emerged that provided an acceptable measurement model for the subset of 18 satisfaction items, 2(125, n = 232) = 207.26, RMSEA =.053, GFI =.91, CFI =.93, TLC =.92. This modified 5-factor model not only fit the data reasonably well and was more parsimonious than the initial a priori model, but also maintained the reliabilities of the individual subscales (see Table 3) and largely preserved the conceptual content of the original model.
|
2(10, n = 232) = 510.39, p <.0001, supporting the multidimensionality of family member satisfaction.
Table 4 displays the factor intercorrelations for the 5-factor model. Several results are noteworthy: (a) The strongest factor correlation was between the Staff Responsiveness factor and the Transportation factor (
=.57, p <.0001; 32.5% shared variance); (b) Resident Responsibilities was relatively independent of the other four satisfaction subscales (s =.01–.16; ps >.087); (c) Family Member Impact showed equivalent correlations with the Staff Responsiveness ( =.36), Transportation ( =.38), and Activities factors ( =.36), 2(2, n = 232) = 0.06, p >.99; and (d) Activities was more closely related to Staff Responsiveness ( =.38, or 14.4% shared variance) than to Transportation ( =.20, or 4% shared variance), 2(1, n = 232) = 4.69, p <.031. This pattern of relationships generally supported the discriminant validity of the 5-factor model. Thus the 5 subscales compose the Assisted Living Family Member Satisfaction Scale.
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
The Assisted Living Family Member Satisfaction Scale comprises an especially useful component of this study. Recent research found that, compared with family members of dementia residents living in nursing homes, family members of dementia residents living in residential care or assisted living settings felt more involved and burdened (Port et al., 2005). Although the current study did not focus on assisted living residents with dementia, as one of the primary consumers, family members' satisfaction with choice of setting is crucial. Also, three of the family member scales corresponded to the resident scales: Staff Responsiveness corresponded to Staff, Resident Responsibilities corresponded to Autonomy, and Activities corresponded to Activities. Using comparable scales for residents and family members provides the opportunity to obtain the perspectives of two critical consumers regarding the same components of assisted living. Differences between these two perspectives could direct service providers to alter their services and programs to better serve one type of consumer or the other. Success of assisted living must ultimately be judged by the ability of this residential option to meet the needs of both types of consumers. In addition, the family member measure included scales that assessed two salient issues to family—transportation (a service that many relatives may have provided prior to the resident's move to assisted living) and the overall impact of the move on family members. Thus, in addition to the capacity of the ALRSS to provide unique information related to resident autonomy, privacy, and knowledge, the ability to obtain the perspectives of both residents and family members using similar subscales that assess staff responsiveness, resident autonomy, and activities is a major contribution of the these two new measures. Used in tandem, these measures will provide a more thorough assessment of the success of assisted living to meet the needs of consumers.
Limitations
We should note a number of limitations regarding this study. Site staff determined selection of family member respondents; thus, non-random selection may limit the generalizability of the findings. Limited demographic information on family member respondents (as well as residents) made it difficult to estimate the representativeness of the sample. A new study that includes respondents from a known sampling frame is needed to estimate the generalizability of findings.
Similarly, one cannot assume that the findings from this study are representative of assisted living residents and their family members in Illinois. We undertook the work described in this article in preparation for an evaluation of a 3-year demonstration project to provide assisted living services to moderate- and low-income individuals in Illinois. The demonstration blended the state's home- and community-based Medicaid-waiver program (Community Care Program) with several different congregate senior living buildings. At the time we conducted this study, Illinois did not regulate or license assisted living facilities. This lack of licensure enabled providers to advertise their facilities as assisted living without regard to standardized criteria. In practice, some facilities operated in a gray zone by providing services for individuals with little to no impairment. Given the available data in this study, the extent to which study sites are representative of sites in Illinois is unknown.
Although this study intentionally focused on subjective elements of satisfaction, future studies should include objective elements (such as chronic health conditions, ability to conduct activities of daily living, mental health, and economic circumstances) to fully understand resident and family member satisfaction and the relationship of objective and subjective factors to each other and to satisfaction. In addition, the construct validity of the resident and family member scales described in this paper could be further enhanced by a new study in which the current scales are assessed against multiple measures expected to provide both high (convergent validity) and low correlations (discriminant validity).
In terms of using measures of resident satisfaction to evaluate assisted living providers, the measurement of consumer satisfaction with any type of community-based long-term care is still in the earliest form of development (Geron, 1998). From a conceptual viewpoint, resident satisfaction is a necessary, but not sufficient, component for potential consumers and policy makers to use for assessing assisted living providers. For instance, some of the strongest predictors of resident satisfaction may be beyond the control of the assisted living facility (e.g., who made the decision to move to the facility, whether the resident can still drive). At a minimum, researchers must use resident characteristics (e.g., health, functional, and cognitive status) to adjust for case-mix differences. Also, it is not unreasonable to expect that the relative importance of factors that influence satisfaction could change over time (e.g., as residents experience functional and/or cognitive declines). Thus, the field needs research across the natural life span of assisted living residents to understand which factors are most influential at different points in time. From our viewpoint, providers of assisted living should use these instruments to begin to assess resident and family member satisfaction and to help them understand what is working and what isn't working.
Application of the ALRSS and the Assisted Living Family Member Satisfaction Scale
Assisted living providers may start with the assumption that most consumers are capable of making choices about whether assisted living is the right option for them. However, our experience has been that providers vary with respect to their ability to provide potential residents and their family members with information needed to make informed choices. This decision can be especially problematic for novice assisted living users. In many cases, materials provided to potential residents and their family members have been developed for marketing rather than educational purposes. Findings from satisfaction surveys such as those described in this article can be used by administrators or external agencies to educate potential consumers regarding the estimated impact of specific assisted living facilities from residents' and family members' perspectives.
The satisfaction surveys can be useful to providers for developing an understanding of the expectations of residents and family members. Customer satisfaction can be interpreted as a proxy for the degree to which residents and family members understand the benefits and limitations of residing in an assisted living facility. As suggested by Lowe, Lucas, Castle, Robinson, and Crystal (2003), satisfaction data are critical to supporting a comprehensive information system for providers to identify and address consumer concerns in relation to operational outcomes. Satisfaction data also can help providers identify misconceptions that residents and family members may have regarding the degree to which facilities actually permit aging in place (Kissam, Gifford, Mor, & Patry, 2003). This information can support targeted education efforts for these customer groups both pre-admission and while residents live in assisted living, and may reduce unnecessary (and expensive) resident turnover. This is consistent with Carder and Hernandez's (2004) recommendation that providers develop and implement tools for applying a more consumer-friendly model.
In assisted living there is a clear need for accountability by focusing on resident outcomes. "Performance outcomes must constitute the indicator of quality, as opposed to compliance with minimum standards that serve as the focus of regulatory monitoring" (Zimmerman, Sloane, & Eckert, 2001, p. 132). The measures described in this article can support a process leading to comprehensive quality-of-care evaluations, which are strongly recommended within the assisted living industry (Sheehan & Oakes, 2003). Such provider-initiated measures serve the purpose of stimulating proactive initiatives to address resident and family member concerns and can promote self-monitoring as an important and routine activity. Self-monitoring provider behavior can potentially reduce the need for regulatory oversight and intervention.
|
Footnotes |
|---|
1 Mather LifeWays Institute on Aging, Evanston, IL.
2 Midwest Center for Health Services and Policy Research, Edward Hines, Jr., Veterans Administration Hospital, Hines, IL.
3 Department of Psychology, Loyola University Chicago, IL.
4 School of Nursing, Northern Illinois University, DeKalb.
Decision Editor: Linda S. Noelker, PhD
Received for publication September 22, 2005. Accepted for publication June 20, 2006.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. Wallace Williams, T. Desai, J. T. Rurka, and E. J. Mutran Predictors of Satisfaction for African-American and White Family Caregivers of Adult Care Home Residents Journal of Applied Gerontology, November 1, 2008; 27(5): 568 - 587. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
