Abstract

Consumer choice is a key concept in developmental disability intervention, but relatively little quantitative research has focused on the relationship between choice and quality of life. This study used data from Washington state's Division of Developmental Disabilities 2002 National Core Indicators study (Human Services Research Institute, 2001a, 2001b) to examine the relationship between choice and 3 quality-of-life indicators: community inclusion, rights, and opportunities for relationships. Consumers (N = 224) with mild intellectual disabilities participated in the study. Structural equation modeling was used to assess the influence of type of living arrangement and choice on quality of life. Consumers who lived in the community and made more choices had higher scores on quality-of-life indicators. The findings have implications for disability policy, practice, and future research.

Over the last 30 years, consumer choice has emerged as a key concept in developmental disability (DD) intervention (Bambara, 2004; Mary, 1998). Consumer decision making about services and daily life is considered a central element of self-determination (Wehmeyer, 2001) and empowerment for people with DDs (Renz-Beaulaurier, 1998). Approaches to service delivery, such as person-centered planning, which places the consumer in the role of decision maker about what supports are needed and who will provide them (O'Brien & O'Brien, 2002), are now recognized as best practice.

Despite broad acceptance of choice as a key service principle, little is known about the relationship of choice to quality of life. Although there is a hypothesized positive relationship between choice and quality of life for people with DDs (Brown & Brown, 2005; Cummins, 2005; Halle, 1995), empirical validation remains limited. The purpose of this study was to examine the relationships among choice, living arrangement, and three quality-of-life indicators: community inclusion, rights, and relationships in a population of persons with primarily mild intellectual disabilities (ID).

Quality of Life

Quality of life has been defined in various ways in the developmental disability literature. Schalock et al. (2002, 2004) have identified eight core domains of quality of life for people with DDs: interpersonal relations, social inclusion, personal development, physical well-being, self-determination, material well-being, emotional well-being, and rights (Schalock, 2004; Schalock, Bonham, & Marchand, 2002). However, other researchers (Cummins, 2005) have argued that a comprehensive definition of quality of life is unwieldy, does not lend itself well to theoretical testing, and does not have empirical justification. Campo et al. (1996, 1997) examined the measurement characteristics and correlates of the Quality of Life Index (Keith, Schalock, & Hoffman, 1986). Using exploratory factor analysis, Campo et al. (1996) found three subscales in the Quality of Life Index measuring Environmental Control, Social Relations, and Community Involvement. Building on Campo's model, we used three scales to measure quality of life: Community Inclusion, Rights, and Relationships.

Although some scholars have argued that choice is a component of quality of life (Schalock, 2004; Schalock, Bonham, & Marchand, 2002), Cummins (2005) argued that choice has a causal relationship to quality of life and should, thus, be considered as distinctly different. In addition, Brown and Brown (2005) have argued that individual choice is a fundamental principle guiding interventions designed to improve quality of life. Other scholars have noted how choice may influence particular domains associated with quality of life. For example, DeJong (1983) argued that if consumers are given greater control over their services, they will select the services that best meet their needs, remove barriers to participation in society, and facilitate independent living. Last, consumer choice may increase interpersonal relations and community integration. When given choice in activities, consumers may choose to spend more time with friends and family and have more access to the community (Davis & Faw, 2002; Neely-Barnes, 2005).

Although theoretical reasons for a link between consumer choice and quality of life have been suggested, only a few prior studies have examined the question. Two studies examined the relationship between choice and one or more indicators of quality of life. Kosciulek and Merz (2001) found a positive relationship between consumer direction and community integration. Gardner and Carran (2005) found a modest relationship between choice and freedom from abuse and neglect and a moderate relationship between choice and performance of different social roles. Two additional studies have examined the relationship between choice and quality of life in samples of people with mild ID. In an international study, Lachapelle et al. (2005) found that each of four aspects of self-determination (behavioral autonomy, self-regulation, psychological empowerment, and self-realization) predicted greater quality of life. Wehmeyer and Schwartz (1998) found a significant correlation between self-determination and quality of life but not between life choices and quality of life.

Living Arrangement

The literature suggests that consumer choice and self-determination are moderated by size of living arrangement. Smaller residential units have been associated with greater self-determination (Tossebro, 1995), more personal control (Stancliffe, Abery, & Smith, 2000), and more choice (Wehmeyer & Bolding, 1999). Duvdevany et al. (2002) found that people living with their families had higher levels of self-determination and lifestyle satisfaction than people living in group settings.

In addition, prior research has indicated relationships among living arrangement, community integration, and family contact. Living in an institutional setting has been associated with less integration in the community (Gardner, Carren, & Nudler, 2001; Stancliffe & Lakin, 1998) and less contact with family (Stancliffe & Lakin, 1998, 2006) than living in smaller arrangements. In addition, Gardner, Carren, and Nudler (2001) found that people with severe–profound ID experienced significantly less autonomy and attainment of goals when they lived in large settings. In another study, Gardner and Carran (2005) found that people with mild–moderate ID living independently or with family experienced more personal outcomes (e.g., autonomy, rights, health and wellness) than people living in larger group arrangements. Last, in a qualitative study, Hagner, Snow, and Klein (2006) found that people with DDs who owned their own homes interacted frequently with family and people in the community. We have found no prior studies that have examined the relationship of living arrangement, choice, and quality of life in one model.

Theoretical Model

We proposed and tested two theoretical models in this study: (a) the influence of choice and living arrangement on three quality-of-life indicators (i.e., rights, community inclusion, and relationships) and (b) the influence of choice and living arrangement on a latent quality-of-life variable. Although we proposed important questions that should be asked for the entire developmental disability population, we used data collected on a population of people with primarily mild ID. The following hypotheses were used to build the models:

  • Greater choice will be negatively correlated with living in larger residential settings.

  • Consumer choice will be positively correlated with quality-of-life measures.

  • People living in smaller settings will report better quality of life.

Method

Data Source

Data from Washington state's 2002 National Core Indicators (NCI) Consumer Survey were used in this study (Human Services Research Institute, 2001a, 2001b). NCI conducts annual surveys of consumers, family members, and providers, with the purpose of developing performance and outcome indicators that developmental disability authorities in the state can use to evaluate their service system. The 61 core indicators, tested during Phases I and II of the NCI project, were developed by a task force consisting of staff of developmental disability authorities nationwide. Interrater and test–retest reliabilities were conducted in three pilot studies and resulted in 92%–93% agreement between raters and 80% agreement between test–retest. In addition, questions were used throughout the survey to test for inconsistent response patterns such as acquiescence (the tendency to answer “yes” to every question). Interviewer feedback was used in the early stages of the project to help the staff improve the questions and instructions (Human Services Research Institute, 2001a). The NCI survey items address (a) case management, (b) choice and control, (c) community connections, (d) family outcomes, (e) functioning and disability, (f) informal support, (g) information and planning, (h) satisfaction, and (i) services and supports (Human Services Research Institute, 2003).

Sixteen states and 1 developmental disability authority participated in the 2002 consumer survey (Human Services Research Institute, 2003). We used data only from Washington state in this study. The consumer survey, the only survey used in this study, had three sections. The Background Information section, which included questions about demographics, diagnoses, health, residence, services, features of self-determination, and behavioral support needs, was mailed to case managers. The Consumer Interview Part I, which included questions about work and daytime activities, home, friends and family, rights, and services or supports, was a face-to-face interview completed with the consumer. Consumers responded either verbally or with pictures. Only the consumer was allowed to respond to these questions. The Consumer Interview Part II, which included questions about community inclusion, choice, and access, was also a face-to-face interview. If the consumer was unable to answer the questions in this section, a proxy responded instead. Proxy respondents included advocates, family members, home staff, and day staff.

The Washington state Division of Developmental Disabilities contracted with the Center on Disability Policy and Research (CDPR) of the University of Washington to hire and train interviewers and collect data. The interviewers hired by CDPR were trained using the curriculum provided by the Health Services Research Institute for National Core Indicator states as well as a local curriculum that the CDPR developed for training interviewers. All of the interviewers also had prior work experience providing direct-care services to people with DDs.

Sampling Procedures

We drew a random sample of 2,000 consumers who were receiving services from the Washington state Division of Developmental Disabilities from a sampling pool of approximately 11,680 adults. Respondents or their legal guardians were contacted by mail for consent to participate in the study. After respondents or the legal guardians had replied to the mailing and agreed to participate, the case manager was asked to complete the Background Section of the survey. Last, the consumer (or his/her representative) was contacted to complete the face-to-face interview. Out of the 2,000 consumers randomly selected, only 343 replied to the mailing and participated in the NCI study (Human Services Research Institute, 2001a, 2001b). Out of the 343, nine individuals were unwilling to complete the consumer interview and their data were removed. Another 13 cases were removed because more than 50% of their responses to all survey questions were missing.

Ninety-seven participants had more than 50% missing data from the items in Section I of the Consumer Interview. The survey design excluded proxies from responding to items from this section, which included some of the questions on rights and all relationship questions. Missing data patterns from this section were examined, and it was determined that most of the respondents who had missing data were unable to answer the questions because of cognitive or communication limitations. Because these participants differed substantially in their levels of ID and rates of co-occurring disorders from those who answered the questions in this section, we determined that including their responses would lead to biased estimates. After removing those participants with more than 50% missing in this section, 224 respondents remained in the study.

Sample

The sample (N = 224) ranged in age from 20 years to 84 years (M = 39.4). About half of the respondents were men (n = 111). The majority of the sample was White (n = 189). Other respondents reported their race or ethnicity as American Indian (n = 4), Asian American (n = 3), Black (n = 6), Latino (n = 6), and other (n = 10; see Table 1).

Table 1

Demographic Characteristics and Descriptive Statistics for the Full Sample (N = 224)

Demographic Characteristics and Descriptive Statistics for the Full Sample (N = 224)
Demographic Characteristics and Descriptive Statistics for the Full Sample (N = 224)

Measures

Level of ID

Level of ID was assessed as part of the questionnaire sent to the case manager and had five possible response categories: 0 = no mental retardation (MR), 1 = mild MR, 2 = moderate MR, 3 = severe MR, and 4 = profound MR (the term mental retardation was used in the questionnaire). Case managers were asked to refer to administrative records to determine the appropriate level.

Living arrangement

Case managers were asked three questions about the person's residence. First, they were asked to characterize the type of home situation. Responses to this question were recoded to create a measure of size and consumer control over the residential environment, with four categories: 3 = institutionalized, 2 = community-based small-group living, 1 = family home, and 0 = own home. An open-ended question asked how many people with DDs lived in the person's immediate residence. Responses to this question were recoded into four categories: 1 = one person (or, the study participant only), 2 = two to three people, 3 = four to eight, and 4 = nine or more. A final question asked who owned the study participant's home, with three response categories: 2 = state or agency owned, 1 = family owned, and 0 = individual owned.

Choice

Five indicators of consumer choice were used in this study: (a) chooses schedule, (b) chooses what to do during free time, (c) chooses what to buy, (d) chose home staff, and (e) chose case manager. All questions were coded as follows: 0 = someone else decides, 1 = the person has help deciding, and 2 = the person decides independently. Choice items were selected if there was no skip pattern or if the skip pattern could be meaningfully interpreted as the client deciding or not deciding. Other choice items that were available in the National Core Indicators data set (Human Services Research Institute, 2001a, 2001b) could not be used due to skip patterns. For example, consumers were not asked, “Did you choose where you live?”, if they lived with their family. Consumers were not asked, “Did you choose who helps you at work?”, if they had no job and/or no job staff. The five variables were used as indicators of choice.

Community inclusion

Eight questions, asked of either the consumer or a proxy respondent, examined the consumer's inclusion in the community. Consumers were asked whether they go (a) shopping, (b) out on errands, (c) out for entertainment, (d) out to eat, (e) out to religious services, (f) out to clubs or community meetings, (g) out for exercise or sports, and (h) who the respondent goes with into the community. Responses were coded from 0 to 2 with a low score indicating no inclusion and a high score indicating access to the community. Three questions (i.e., religious services, sports or exercise, and who the respondent goes with) had a possible middle response category indicating that the person participated in a less inclusive setting or with less choice in participation. Responses to these eight questions were combined to form an index of community inclusion with a range of 0 to 16.

Rights

Six variables were used to create a scale of rights. Consumers were asked whether they are given the right to (a) open their mail, (b) use the phone, (c) be alone, and (d) go to self-advocacy groups. In addition, consumers were asked whether staff asked permission to (e) enter their home and (f) their bedroom. Responses were coded as follows: 0 = rights are never respected, 1 = rights are sometimes respected, and 2 = rights are always respected. Rights variables were chosen for this study if there was no skip pattern or if the skip pattern could be meaningfully interpreted as either “always” or “never” responses. Data from these six rights questions were combined to create an index with a range of 0 to 12.

Relationships

Five questions were used to create an index of opportunities for and access to relationships. Respondents were asked whether they (a) have friends, (b) have a best friend, (c) see friends when they want, (d) see family when they want, and (e) whether they ever feel lonely. Responses were coded as follows: 0 = no, 1 = sometimes, and 2 = yes. Responses to “ever feel lonely” were reverse coded. Questions were combined to create a scale with a range of 0 to 10.

Proxy

Proxy respondents were used if the consumer was unable to respond to part or all of the questions in the final section of the survey (questions on choice, community inclusion, and selected questions on rights). Each interview question with a possible proxy respondent had an indicator of whether the consumer or the proxy answered. Data from these indicators were combined to create an index of consumer participation in the survey. The index was coded as follows: 0 = consumer answered no questions, 1 = consumer answered at least one but less than half, 2 = consumer answered more than half but not all, and 3 = consumer answered all.

Prior research has suggested that answers given by consumers and proxies may not be directly comparable (Stancliffe, 1995, 2000). One method of dealing with the influence of proxy is to collect data on whether study participants or proxies responded and to use the information as a control variable (Schalock, Bonham, & Marchand, 2000). This method of controlling for the influence of proxy was used in the present study.

Data Analysis

Data were analyzed using SPSS 13.0 (SPSS, 2004) and MPLUS version 3.1 (Muthén & Muthén, 2004). Analysis of the descriptive statistics was completed using SPSS 13.0. The SPSS Missing Value Analysis (MVA) module was used to examine missing data patterns and impute missing values in a two-step process. SPSS MVA imputes values for continuous variables through a maximum likelihood method based on Little and Rubin's (1987) work on expectation-maximization (EM) algorithms. Variables used in this study ranged from 0% to 11% cases missing, with 25 complete cases. Missing data were not imputed for categorical variables. Cases that had imputed values outside the range of the variable were recoded to either the highest or lowest meaningful value as appropriate. Missing data were also not imputed for consumers who answered less than 50% of the questions in Section I of the consumer interview. We assumed that most of the consumers with missing data from this section failed to respond for nonrandom reasons related to their ability to communicate without the assistance of a proxy. Thus, imputing data for the full sample for this section might lead to biased estimates (Schafer & Graham, 2002).

Structural equation modeling techniques were used to assess the fit of the models shown in Figures 1 and 2. Prior to testing the model, Pearson correlations were examined for continuous variable relationships, and Spearman's rho correlations were examined for ordinal variable relationships and continuous–ordinal variable relationships. Then, a confirmatory factor analysis (CFA) was conducted to test the adequacy of the measurement model of choice, living arrangement, and quality of life.

Figure 1

Influence of choice and living arrangement on community inclusion, relationships, and rights. CM = case manager; MR = mental retardation. Not pictured: No. of residents with developmental disabilities (DDs) ↔ choose staff, −.33***. *p < .05. **p < .01. ***p < .001

Figure 1

Influence of choice and living arrangement on community inclusion, relationships, and rights. CM = case manager; MR = mental retardation. Not pictured: No. of residents with developmental disabilities (DDs) ↔ choose staff, −.33***. *p < .05. **p < .01. ***p < .001

Figure 2

Influence of choice and living arrangement on quality of life. CM = case manager; MR = mental retardation. Not pictured: No. of residents with developmental disabilities (DDs) ↔ choose staff −.33***. *p < .05. **p < .01. ***p < .001

Figure 2

Influence of choice and living arrangement on quality of life. CM = case manager; MR = mental retardation. Not pictured: No. of residents with developmental disabilities (DDs) ↔ choose staff −.33***. *p < .05. **p < .01. ***p < .001

After we obtained a satisfactory measurement model, we tested the structural model. Model 1 assessed the relationship of latent choice and living arrangement variables to each of three measured quality-of-life indicators: community inclusion, rights, and opportunities for relationships. Model 2 assessed the relationship of the latent choice and living arrangement variables to a latent quality-of-life variable. A latent variable is used when a concept is theorized to exist but is not directly observable and instead is measured through a set of indicators (Kline, 1998). The advantage of using latent variables is that one can take into account the imperfect reliability of measures by incorporating error into the model. Because none of the items in the National Core Indicators survey (Human Services Research Institute, 2001a, 2001b) directly measures living arrangement, choice, or quality of life, indicators of these concepts can be used to estimate these variables.

The analysis of each model was completed in a two-step process. Both confirmatory factor analysis and structural equation modeling techniques were conducted using MPLUS version 3.1 (Muthén & Muthén, 2004) with weighted least square parameter estimates. This estimation method is well suited to data that contain both continuous and ordered categorical indicators.

To assess the overall goodness of fit of the measurement model and the structural model, four indexes were used: a chi-square test that was mean and variance adjusted using a full-weight matrix, the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the weighted root mean square residual (WRMR; Hu & Bentler, 1998). The chi-square test uses covariances for the continuous variables and probits regressions for the ordered categorical variables. A full explanation of the chi-square procedure can be found in Muthén (2004). The comparative fit index has values ranging from 0 to 1, with values over .9 indicating a good fit. The RMSEA can be interpreted as follows: values less than .05 indicate a close fit, and values between .05 and .08 indicate an acceptable fit (Kline, 1998). We used the WRMR because it is suitable when sample statistics have a wide range of variances and when sample statistics are on different scales. Values of WRMR below .9 indicate a good fit (Muthén, 2004). Last, the Lagrange multiplier test was used to examine the parameters fixed to zero. This test approximates the amount by which the overall chi square would decrease if a particular parameter were added to the model (Kline, 1998).

Results

Approximately 87% of the respondents had a diagnosis of ID. The most frequently reported other diagnoses were psychiatric disorder (21.7%) and seizure disorder (20.4%). Most respondents lived in their own home (41.1%) or with a family member (34.8%) and were the only person with a developmental disability living in their household (55.4%). Other respondents lived in community-based group homes (18.8%) or in institutional settings (5.4%). See Table 1 for additional descriptive statistics.

In addition to descriptive statistics, we examined the correlations of all observed variables used in the structural models. Pearson correlations for continuous–continuous relationships and Spearman's rho correlations for continuous–ordinal and ordinal–ordinal relationships are reported in Table 2.

Table 2

Correlation Matrix for the Variables Used in the Structural Equation Models (N = 220)

Correlation Matrix for the Variables Used in the Structural Equation Models (N = 220)
Correlation Matrix for the Variables Used in the Structural Equation Models (N = 220)

Influence of Choice and Living Arrangement on Community Inclusion, Rights, and Relationships

Prior to testing the model depicted in Figure 1, the living arrangement and choice measurement model was tested and had a good fit (see Table 3). The model was tested with factor loadings free to vary and all factor variances fixed at 1.0 (Kline, 1998). The errors on Variable 5 (change staff) and Variable 6 (number of people with DD) were allowed to covary in all models, as the Lagrange multiplier test indicated that adding this parameter would improve model fit (see Table 3 for the chi-square statistics). The choice and living arrangement factors had a correlation of −.27 that was significant at p < .01. In other words, greater choice was associated with smaller living arrangements (see Table 4 for the standardized factor loadings).

Table 3

Fit Indexes: Confirmatory Factor Analyses (CFAs) and Structural Equation Models (SEMs)

Fit Indexes: Confirmatory Factor Analyses (CFAs) and Structural Equation Models (SEMs)
Fit Indexes: Confirmatory Factor Analyses (CFAs) and Structural Equation Models (SEMs)
Table 4

Confirmatory Factor Analysis of Choice, Living Arrangement, and Quality of Life

Confirmatory Factor Analysis of Choice, Living Arrangement, and Quality of Life
Confirmatory Factor Analysis of Choice, Living Arrangement, and Quality of Life

The structural model depicted in Figure 1 had a good fit (see Table 3). The relationship between choice and living arrangement was nonsignificant, yet it was retained because it improved the model fit. The path from proxy to choice was significant. Respondents who answered questions through the assistance of a proxy experienced less choice. Significant paths were also present from level of ID to choice and living arrangement. People with more severe ID experienced less choice and lived in larger group environments. Significant paths were found from choice to rights and community inclusion. Respondents who experienced more choice also experienced greater respect of rights and increased community inclusion. The path from living arrangement to rights was also significant. Respondents who lived in smaller individualized or family environments experienced greater respect for their rights.

The paths from living arrangement to opportunities for relationships and choice to opportunities for relationships were not significant but were retained because they improved the model fit. Although not significant, these paths were in the expected direction. Larger living arrangements had a weak association with greater opportunity for relationships, and greater choice had a weak association with greater opportunity for relationships.

The model explained 45% of the variance in choice, 14% of the variance in living arrangement, 7% of the variance in community inclusion, 5% of the variance in opportunities for relationships, and 23% of the variance in rights. In other words, use of proxy respondent and level of ID contributed a great deal to explaining variation in choice. In addition, choice and living arrangement contributed a great deal to explaining the consumer's experience of respect for rights but little to explaining community inclusion or relationships. The path coefficients and explained variance are reported in Figure 1.

Influence of Choice and Living Arrangement on Quality of Life

Figure 2 presents a test of the structural model of the relationships among choice, living arrangement, and a latent quality-of-life variable. Prior to testing the structural model, another confirmatory factor analysis model was estimated. All factor loadings were significant except for chose case manager; the results are shown in Table 4. Again, factor loadings were free to vary and factor variances were fixed at 1.0. The measurement model yielded an adequate fit (see Table 3). As in the two-factor model, the correlation between the choice factor and the living arrangement factor was −.27 (p < .01). The correlation between the choice and quality-of-life factors was .62 (p < .001), and the correlation between living arrangement and quality of life was −.43 (p < .001). In other words, smaller living arrangements were associated with greater choice and greater quality of life, and greater choice was associated with greater quality of life. The Lagrange multiplier test did not indicate significant loadings for choice items on the quality of life factor. Thus, we decided that the model confirmed the conceptualization of choice as a separate factor from quality of life.

The structural model had a good fit (see Table 3). Again, proxy respondent had a significant relationship to choice, and level of ID was significantly related to choice and living arrangement. Respondents with more severe ID experienced less choice and lived in larger group settings. The correlation between choice and living arrangement was not significant but was retained in the model because it improved the model fit. Although not significant, the relationship between choice and living arrangement was in the expected direction: Greater choice was weakly associated with smaller living arrangements. Both paths from choice and living arrangement to quality of life were significant. In other words, respondents who made more choices and who lived in smaller or more individualized living arrangements experienced greater quality of life.

The model explained 45% of the variance in choice (R2 = .45), 14% of the variance in living arrangement (R2 = .14), and 45% of the variance in quality of life (R2 = .45). In other words, choice and living arrangement explained a great deal of the variation in respondents' quality of life. In addition, proxy respondent and level of mental retardation explained a great deal of the variation in choice.

Discussion and Implications

We hypothesized that both greater choice and smaller living arrangement would be significantly related to increased rights, relationships, and community inclusion in Model 1 and to increased quality of life in Model 2. Although not all the anticipated relationships were established, important benefits to smaller living arrangements and greater choice were noted from this study. In the first model, smaller living arrangements were associated with experiencing more protection of rights, and greater choice was associated with experiencing more protection of rights and greater access to the community. In the second model, both greater choice and smaller living arrangements were associated with increased quality of life. These findings have important implications for developmental disability policy and practice. First, the findings support the current movement toward closing institutions and creating more independent living arrangements (Lakin et al., 1999) and emphasize the need to expand opportunities for consumers to live in smaller settings. Second, this study supports current practice models that give consumers more choice and control over their daily lives (O'Brien & O'Brien, 2002) and emphasizes the need to continue to increase the choice that people with ID experience.

In both models, level of ID was associated with living arrangement and choice. People with less severe ID experienced greater choice and lived in smaller settings. These findings are consistent with prior research that has indicated that people with more severe ID experience less choice (Wehmeyer & Garner, 2003) and overrepresentation in ID/DD institutions (Lakin, Anderson, Prouty, & Polister, 1999).

In this study, we did not find an association between choice and living arrangement. This finding is inconsistent with prior research that has found a negative association between large-group living arrangements and consumer choice or consumer self-determination (Stancliffe, Abery, & Smith, 2000; Tossebro, 1995; Wehmeyer & Bolding, 1999). It is possible that the limited measures of both choice and living arrangement contributed to this nonsignificant finding. If we had been able to use a more comprehensive measure of choice or measured a broader concept such as self-determination, we might have had significant findings. Alternatively, this finding might be related to the characteristics of the study population. People with severe–profound ID are more likely to live in large-group situations than people with mild–moderate ID (Lakin et. al., 1999). People with severe–profound ID also experience more barriers to choice (Reid & Green, 2002). If more people with severe– profound ID had been able to participate in the study, a significant relationship between choice and living arrangement might have been found.

The results of this study suggest that three measures—community inclusion, rights, and opportunities for relationship—can be used to model quality of life. This finding supports the work of Campo et al. (1996, 1997), who found that quality of life has three indicators. Our findings also support prior theoretical work that argues that there is an underlying quality-of-life construct for people with DDs (Schalock, 2004; Schalock et al., 2002). However, there is a need to follow up this study with additional research using validated and more comprehensive measures of quality of life.

Last, this study had important findings and implications for the use of proxies in research. A strong relationship was present in both models between proxy respondent and choice. Participants in the study who had a proxy respond to their choice questions experienced less choice than those who answered without a proxy. Prior research has suggested that responses from individuals with DDs and from proxies may not be directly comparable (Stancliffe 1995, 2000). Thus, it may be that this finding can be accounted for by differences in the proxy's and consumer's perception of the ability to make choices. As an alternative, it may be that respondents, who were not able to answer without the assistance of a proxy, also have more difficulty making and expressing their choices. Additional research is needed to determine why this relationship existed. However, this study raises an important question about whether people with limited verbal abilities have adequate opportunities to make and express choices and underscores the need for practitioners and researchers to continue to look for ways for nonverbal individuals to express choice.

Also concerning the use of proxies, only a subset of the participants in the NCI study (Human Services Research Institute, 2001a, 2001b) could participate in the present study. Because the survey instrument relied solely on consumer responses for some items, the individuals with the most severe IDs were missing key information and consequently unable to participate in this study. Although some studies have questioned the reliability of proxies (Reid & Green, 2002; Umb-Carlsson & Sonnander, 2005), the failure to use the proxy option across all items resulted in missing data for a key segment of the original sample. It is recommended that NCI consider ways to address this loss of data including adding the proxy response option to more items or using observational methods.

Limitations and Future Directions

This study has a number of limitations. First, it is important to note that the model of the influence of choice and size of living arrangement on quality of life is theoretical. Although survey questions were worded to suggest a directional relationship, the direction of the relationships in the model cannot be confirmed because all variables were measured at the same time. Second, a limited number of variables were used in this model. Other variables (e.g., measures of work–day-program integration, collaboration with case manager and support staff, and additional quality-of-life indicators) might have contributed to the model if they had been adequately measured without the skip patterns present in some of the NCI (Human Services Research Institute, 2001a, 2001b) items. In addition, there are more comprehensive measures of choice and quality of life that could have been used. Future studies could examine whether these relationships among choice, living arrangement, and quality of life exist when more comprehensive measures of the constructs are used.

In addition, it is important to note that these data from Washington state may not be generalizable to the U.S. developmental disability population because definitions of developmental disability vary among states. The generalizability of the study may also be compromised because of the low percentage of the sampling pool (11%) that participated in the study. In addition, it is important to note that people with mild and moderate ID were overrepresented in the study. Thus, the results of this study cannot be generalized to the population of people with severe and profound IDs. Future studies that use proxy respondents or observational methods could examine the relationship among choice, living arrangement, and quality of life for people with more severe IDs.

Last, we were not able to model the relationship among choice, living arrangement, and quality of life without correlating the measurement error on two relationships between items: between the “chose staff” item and the “number of people with DD living with consumer” item and between the “chose staff” item and the “chose case manager” item. Measurement error reflects either random chance or systematic variation in the way measurement occurs (i.e., use of self-reports vs. use of staff reports). When error terms are correlated, the items may measure something in common that is not represented in the model (Kline, 1998). In the case of the relationship between chose staff and number of people with DD, it is not typically the case that people in large-group settings have the opportunity to choose their own staff. Thus, the nature of the staffing arrangements in large versus small living arrangements may account for this finding. The other relationship between the error terms on chose home staff and chose case manager may represent some understanding about the ability to choose paid support persons. When individuals understand that they can choose one type of paid staff, they may then understand that they can choose others.

Despite these limitations, this study makes an important contribution to our knowledge of the association among living arrangement, choice, and quality of life for the population of individuals with mild and moderate ID. It suggests that smaller living arrangements and increased choice are having a positive impact on people's lives. The results support current DD policy and practice movements that seek to increase opportunities for choice and individualized living arrangements and suggest that we should continue in this direction.

References

References
Bambara
,
L. M.
2004
.
Fostering choice-making skills: We've come a long way but still have a long way to go.
Research and Practice for Persons with Severe Disabilities
29
:
169
171
.
Brown
,
R. I.
and
I.
Brown
.
2005
.
The application of quality of life.
Journal of Intellectual Disability Research
49
:
718
727
.
Campo
,
S. F.
,
W. R.
Sharpton
,
B.
Thompson
, and
D.
Sexton
.
1996
.
Measurement characteristics of the quality of life index when used with adults who have severe mental retardation.
American Journal on Mental Retardation
100
:
546
550
.
Campo
,
S. F.
,
W. R.
Sharpton
,
B.
Thompson
, and
D.
Sexton
.
1997
.
Correlates of the quality of life of adults with severe or profound mental retardation.
Mental Retardation
35
:
329
337
.
Cummins
,
R. A.
2005
.
Moving from the quality of life concept to a theory.
Journal of Intellectual Disability Research
49
:
699
706
.
Davis
,
P.
and
G.
Faw
.
2002
.
Residential preferences in person-centered planning.
In S. Holburn & P. M. Vietze (Eds.), Person-centered planning: Research, practice and future directions (pp. 203– 221). Baltimore: Brookes
.
DeJong
,
G.
1983
.
Defining and implementing the independent living concept.
In N. M. Crewe & I. K. Zola (Eds.), Independent living for physically disabled people: Developing, implementing, and evaluating self-help rehabilitation programs (pp. 4–27). San Francisco: Jossey-Bass
.
Duvdevany
,
I.
,
H.
Ben-Zur
, and
A.
Ambar
.
2002
.
Self-determination and mental retardation: Is there an association with living arrangement and lifestyle satisfaction?
Mental Retardation
40
:
379
389
.
Gardner
,
J. F.
and
D. T.
Carran
.
2005
.
Attainment of personal outcomes by people with developmental disabilities.
Mental Retardation
43
:
157
174
.
Gardner
,
J. F.
,
D. T.
Carran
, and
S.
Nudler
.
2001
.
Measuring quality of life and quality of services through personal outcome measures: Implications for public policy.
In L. M. Glidden (Ed.), International review of research in mental retardation (Vol. 24, pp. 7–99). San Diego, CA: Academic Press
.
Hagner
,
D.
,
J.
Snow
, and
J.
Klein
.
2006
.
Meaning of homeownership for individuals with developmental disabilities: A qualitative study.
Mental Retardation
44
:
295
303
.
Halle
,
J. W.
1995
.
Innovations in choice-making research: An editorial introduction.
Journal of the Association of Persons With Severe Handicaps
20
:
173
174
.
Hu
,
L.
and
P. M.
Bentler
.
1998
.
Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification.
Psychological Methods
3
:
424
453
.
Human Services Research Institute.
2001a
.
Consumer survey: Phase II technical report.
Cambridge, MA: Author
.
Human Services Research Institute.
2001b
.
Core indicators project: Consumer survey.
Cambridge, MA: Author
.
Human Services Research Institute.
2003, March
.
Consumer survey: Final report 2001–2002.
Cambridge, MA: Author
.
Keith
,
K. D.
,
R. L.
Schalock
, and
K.
Hoffman
.
1986
.
Quality of life: Measurement and programmatic implications. Lincoln, NE: Region V Mental Retardation Services.
Kline
,
R. B.
1998
.
Principles and practices of structural equation modeling.
New York: Guilford Press
.
Kosciulek
,
J. F.
and
M.
Merz
.
2001
.
Structural analysis of the consumer-directed theory of empowerment.
Rehabilitation Counseling Bulletin
44
:
209
216
.
Lachapelle
,
Y.
,
M. L.
Wehmeyer
,
M. C.
Haelewyck
,
Y.
Courbois
,
K. D.
Keith
,
R.
Schalock
,
M. A.
Verdugo
, and
P. N.
Walsh
.
2005
.
The relationship between quality of life and self-determination: An international study.
Journal of Intellectual Disability Research
49
:
740
744
.
Lakin
,
K. C.
,
L.
Anderson
,
R.
Prouty
, and
B.
Polister
.
1999
.
State institution populations less than one third of 1977, residents older with more impairments.
Mental Retardation
37
:
85
86
.
Little
,
R. J. A.
and
D. B.
Rubin
.
1987
.
Statistical analysis with missing data.
New York: Wiley
.
Mary
,
N. L.
1998
.
Social work and the support model of services for people with developmental disabilities.
Journal of Social Work Education
34
:
247
260
.
Muthén
,
B. O.
2004
.
Mplus: Statistical analysis with latent variables technical appendices.
Los Angeles: Muthén & Muthén
.
Muthén
,
L. K.
and
B. O.
Muthén
.
2004
.
Mplus: Statistical analysis with latent variables user's guide.
Los Angeles: Muthén & Muthén
.
Neely-Barnes
,
S. L.
2005
.
Consumer choice in developmental disability services: Assessing the impact on quality of life indicators [Doctoral dissertation, University of Washington, Seattle].
Dissertation Abstracts International
66
:
2725
.
O'Brien
,
C. L.
and
J.
O'Brien
.
2002
.
The origins of person-centered planning.
In S. Holburn & P. M. Vietze (Eds.), Person-centered planning: Research, practice and future directions (pp. 3– 27). Baltimore: Brookes
.
Reid
,
D. H.
and
C. W.
Green
.
2002
.
Person-centered planning with people who have severe multiple disabilities: Validated practices and misapplications.
In S. Holburn & P. M. Vietze (Eds.), Person-centered planning: Research, practice, and future directions (pp. 183–202). Baltimore: Brookes
.
Renz-Beaulaurier
,
R.
1998
.
Empowering people with disabilities: The role of choice.
In L. M. Gutierrez, R. J. Parsons, & E. O. Cox (Eds.), Empowerment in social work practice: A sourcebook (pp. 73–84). Pacific Grove, CA: Brooks/ Cole
.
Schafer
,
J. L.
and
J. W.
Graham
.
2002
.
Missing data: Our view of the state of the art.
Psychological Methods
7
:
147
177
.
Schalock
,
R. L.
2004
.
The concept of quality of life: What we know and do not know.
Journal of Intellectual Disability Research
48
:
203
216
.
Schalock
,
R.
,
I.
Brown
,
R.
Brown
,
R. A.
Cummins
,
D.
Felce
,
L.
Matikka
,
K. D.
Keith
, and
T.
Parmenter
.
2002
.
Conceptualization, measurement, and application of quality of life for persons with intellectual disabilities: Report of an international panel of experts.
Mental Retardation
40
:
457
470
.
Schalock
,
R. L.
,
G. S.
Bonham
, and
C. B.
Marchand
.
2000
.
Consumer-based quality of life assessment: A path model of perceived satisfaction.
Evaluation and Program Planning
23
:
77
87
.
SPSS, Inc.
2004
.
SPSS 13.0 [computer software].
Chicago: Author
.
Stancliffe
,
R. J.
1995
.
Assessing opportunities for choice making: A comparison of self and staff reports.
American Journal on Mental Retardation
99
:
418
429
.
Stancliffe
,
R. J.
2000
.
Proxy respondents and quality of life.
Evaluation and Program Planning
23
:
89
93
.
Stancliffe
,
R. J.
,
B. H.
Abery
, and
J.
Smith
.
2000
.
Personal control and the ecology of community living settings: Beyond living-unit size and type.
American Journal on Mental Retardation, 100(431–454)
.
Stancliffe
,
R. J.
and
K. C.
Lakin
.
1998
.
Analysis of expenditures and outcomes of residential alternatives for persons with developmental disabilities.
American Journal on Mental Retardation
102
:
552
568
.
Stancliffe
,
R. J.
and
K. C.
Lakin
.
2006
.
Longitudinal frequency and stability of family contact in institutional and community living.
Mental Retardation
44
:
418
429
.
Tossebro
,
J.
1995
.
Impact of size revisited: Relation of number of residents to self-determination and deprivatization.
American Journal on Mental Retardation
100
:
59
67
.
Umb-Carlsson
,
O.
and
K.
Sonnander
.
2005
.
Comparison of reports by relatives and staff on living conditions of adults with intellectual disabilities.
Mental Retardation
44
:
120
127
.
Wehmeyer
,
M. L.
2001
.
Self-determination and mental retardation: Assembling the puzzle pieces.
In H. N. Switzky (Ed.), Personality and motivational differences in persons with mental retardation (pp. 147–198). Mahwah, NJ: Erlbaum
.
Wehmeyer
,
M. L.
and
N.
Bolding
.
1999
.
Self-determination across living and working environments: A matched samples study of adults with mental retardation.
Mental Retardation
37
:
353
363
.
Wehmeyer
,
M. L.
and
N. W.
Garner
.
2003
.
The impact of personal characteristics of people with intellectual and developmental disability on self-determination and autonomous functioning.
Journal of Applied Research in Intellectual Disabilities
16
:
255
265
.
Wehmeyer
,
M.
and
M.
Schwartz
.
1998
.
The relationship between self-determination and quality of life for adults with mental retardation.
Education and Training in Mental Retardation and Developmental Disabilities
33
:
3
12
.

Author notes

This research was supported by a grant from the Arc of Washington Trust Fund. Earlier versions of the findings were presented at the Alliance for Full Participation Summit, September 2005, Washington, DC, and at the Society for Social Work and Research Conference, January 2007, Washington, DC. We thank the Washington state Division of Developmental Disabilities and the University of Washington Center for Disability Policy and Research for their assistance in gathering data. In addition, we thank Monica Oxford for her assistance in data analysis and Gunnar Almgren and Jim Whittaker for their assistance in reviewing drafts.

Authors:

Susan Neely-Barnes, PhD (sneely2@utk.edu), Assistant Professor, College of Social Work, University of Tennessee, 711 Jefferson Ave., Room 607W, Memphis, TN 38163. Maureen Marcenko, PhD, Associate Professor, School of Social Work, University of Washington, Seattle, WA 98105. Lisa Weber, PhD, Senior Research and Data Manager, Washington State Division of Developmental Disabilities, Olympia, WA 98504-5310