Abstract

This article examines everyday choices made by 8,892 adults with intellectual and developmental disabilities (IDD) and support-related choices made by 6,179 adults with IDD receiving services from 19 state developmental disabilities program agencies that participated in the 2008–2009 National Core Indicators Project. Controlling for physical and sensory impairment, age, behavioral support, communication, and state, people in residential settings with 16 or more people had less everyday choice than those in other living arrangements. People with mild and moderate IDD had more control over everyday choices when living in their own homes, whereas people with severe and profound IDD had more control when living in agency homes of 3 or fewer residents. For people of all levels of IDD, institutional settings of 16 or more residents offered the lowest levels of everyday choice. Controlling for the same covariates, individuals with all levels of IDD living in their own homes had significantly more support-related choices than those in any other residential arrangement. Controlling for individual and residential setting characteristics, the state in which sample members lived was notably predictive of support-related choice. Overall, the tested variables accounted for 44% of the variability in everyday choice and 31% in support-related choice.

In June 2009, U.S. President Barack Obama inaugurated the Year of Community Living in the United States. In doing so, he committed his administration to enforcing the right of Americans with disabilities to receive needed services and supports in the most integrated setting feasible and to “maximizing the choices and opportunities for people who receive long-term services and supports” (White House, 2009). The freedom to exercise personal control and choice in the decisions that shape everyday life is a foundational concept in American and international perspectives on human rights. It is recognized as a right for people with intellectual and developmental disabilities (IDD) in the United States in the Developmental Disabilities Assistance and Bill of Rights Act of 2000. That act states as its first congressional finding that “disability is a natural part of the human experience that does not diminish the right of individuals with developmental disabilities to live independently, to exert control and choice over their own lives” (114 Stat. 1678). The fundamental right to choice in the lives of people with disabilities is also recognized internationally in the United Nations Convention on the Rights of Persons with Disabilities. The convention is founded on eight general principles, the first of which is, “Respect for inherent dignity, individual autonomy including the freedom to make one's own choices and independence of persons” (United Nations, 2006).

Although personal choice is a widely valued concept, it is also a broad concept. It includes decisions about relatively minor, routine matters that make up daily life to less frequent but also more consequential decisions that have substantial and often long-term consequences for an individual. We examined everyday choice and support-related choice using scales developed through factor analysis of nine choice-related items from the National Core Indicators Project (Lakin et al., 2008). The everyday choice scale included choices about one's daily schedule, free time, and minor purchases. The support-related choice scale included choice of who to live with, where to live, the staff members scheduled, type of work or day activity, and the identity of the case manager (Lakin et al., 2008).

Everyday choice is more readily available to people with IDD than is support-related choice (Heller, Miller, & Factor, 1999; Lakin et al., 2008; Wehmeyer & Metzler, 1995). The level of IDD and/or adaptive behavior is strongly related to choice availability, with those with milder disability enjoying more choice (Heller et al., 1999; Lakin et al., 2008; Stancliffe, 1997; Stancliffe, Abery, & Smith, 2000; Stancliffe et al., 2011). The relation between choice making and other personal characteristics seems less consistent. For example, Stancliffe et al. (2000) found that less choice was associated with more serious challenging behavior, but other investigators have reported no significant link between challenging behavior and choice (Emerson et al., 2000). Lakin et al. (2008) found—after controlling for level of intellectual disability and medical needs, mobility and sensory impairment—that challenging behavior was associated with everyday choice, but that the association became insignificant after factoring in variables such as place of residence and state.

In other studies, environmental characteristics were found to be related to opportunities for choice (Lakin et al., 2008; Stancliffe et al., 2000). Given the consistent association between individual characteristics and choice and between individual characteristics and service-setting program characteristics, well-designed analyses of choice have controlled for individual characteristics to assure valid comparisons. Using covariance to control for adaptive and challenging behavior, Stancliffe et al. (2000) found that residents of supported-living services exercised more personal choice than residents of an intermediate care facility for the mentally retarded (ICF/MR). Controlling for level of intellectual disability, medical care needs, mobility, behavioral, psychiatric, and autism diagnoses, Lakin et al. (2008) found that living in one's own home was associated with both more everyday choice and more service-related choice (the latter, arguably, as much a cause of living in one's own home as an effect). Other well-controlled studies have reported that community-living settings that serve fewer residents were associated with greater choice (Burchard, Hasazi, Gordon, & Yoe, 1991; Stancliffe, 1997; Stancliffe & Keane, 2000; Stancliffe & Lakin, 1998; Wehmeyer & Bolding, 1999).

Most research on choice has been limited to samples from single states or smaller areas. Those studies, although contributing to the growth of evidence regarding choice, have been constrained by sample size and related limitations in the number of variables that could be controlled. They have also been limited by the inability to measure state differences in the analyses. In a six-state study of choice, after controlling statistically for a wide range of individual characteristics and service setting types, state emerged as a significant factor in accounting for variability in both everyday and support-related choice (p < .001) and accounted for more variance in support-related choice (10%) than any other factor except level of intellectual disability (Lakin et al., 2008). The present study replicates this past work but expanded the number of states from 6 to 19 and further elucidated not only the effect of setting type but also setting size. Unlike in the previous study, individuals living in a family member's home were included. This study also further refined several of the covariates included in the regression analysis. The purpose of the current study is (a) to add to the evidence about the extent and the types of choices individuals with different levels of IDD tend to make depending on the types and size of settings in which they live and (b) to outline potential implications of living in each of the settings with a certain level of IDD.

Methodology

Instrumentation

The National Core Indicators (NCI) Consumer Survey was developed by the National Association of State Directors of Developmental Disabilities Services (NASDDDS) and the Human Services Research Institute (HSRI; Smith & Ashbaugh, 2001). The survey is administered annually in states that participate in the NCI program. The quality and consistency of implementation across states is supported by a training program for interviewers. Train-the-trainer sessions and standard training materials are provided to participating states so that interviewers receive consistent training. The materials include instruction in interviewing people with IDD and a question-by-question review of the survey tool. Within the package are training manuals, presentation slides, training videos, scripts for scheduling interviews, lists of frequently asked questions, and picture response formats. To reduce coding errors, states received standard data-entry materials, including codebooks and databases with controlled data-entry forms.

The NCI survey gathers four types of data: descriptive—demographics, functioning, diagnoses, and behavioral health; service use—setting size and service funding source; service and lifestyle—friendships, community participation, family involvement, participation in a self-advocacy group, choices, measures of respect, and employment; and well-being—individual satisfaction/liking of one's life and services. Demographic variables are completed using data from individual records and setting administrators. Choice variables are completed by the person with a disability whenever possible. When the person is unable to answer independently, a response is sought from a knowledgeable proxy (family, advocate, and staff), except on the well-being items that require subjective responses (e.g., liking where one lives, feeling staff are nice and polite). For items allowing proxy responses the source of the information (consumer or proxy) is recorded.

The NCI survey was developed with extensive involvement of a program advisory committee of state officials and other advisors to assure that it validly represented the established national goals of services for people with IDD. A focus group of individuals with intellectual disability participated in a pretest of the face validity of the questions. Focus group participants highlighted problematic questions, identified words that needed further definition, and suggested alternative ways of phrasing questions. These modifications were incorporated into the final survey. Field-testing occurred in three states with interrater reliability tests yielding agreement of 92%–93%, and a lone test–retest of reliability resulting in 80% agreement (Smith & Ashbaugh, 2001).

Sample

In 2008 and 2009, 11,569 adults with IDD from 20 states participated in the Consumer Survey component of the NCI project. The 20 states were Alabama, Arizona, Connecticut, Delaware, Georgia, Illinois, Indiana, Kentucky, Louisiana, Maine, Missouri, North Carolina, New Jersey, New York, Ohio, Oklahoma, Pennsylvania, South Carolina, Texas, and Wyoming. Each state randomly sampled its population of adults (aged 18-plus) with IDD receiving institutional, community, or home-based services, or some subset of these (e.g., some states restricted their sample to recipients of home and community-based services). Sample sizes in participating states ranged from 193 (Delaware) to 1,502 (New York) and averaged 578.

Final Sample

Our final sample included a subset of the people surveyed in 2008 and 2009. The full 2008–2009 NCI Consumer Survey dataset included 11,569 people residing in 20 states. Individuals from one state were omitted from this study because they did not collect survey data on key variables used in the analyses (e.g., level of ID, vision or hearing impairment, and autism), reducing the sample to 11,230. Another 163 sample members were eliminated because interviewers judged their responses unlikely to be reliable and valid. Finally, we excluded 2,171 individuals for whom residence type and size were not reported, yielding a final sample size of 8,892. Sample sizes varied across analyses because of the different patterns of missing individual data. The sample for analyses regarding support-related choice was further reduced to 6,179 people because of additional missing data.

Dependent and Independent Variables

Dependent Variables

Two choices scales were developed based on a factor analysis of the nine choice-related questions in the NCI Consumer Survey, and a factor loading cutoff of .50 or above (Lakin et al., 2008). The Everyday Choice scale included three items on choice of (a) daily schedule, factor loading  =  .70; (b) how to spend free time, factor loading  =  .72; and (c) what to buy with spending money, factor loading  = .60. Internal consistency was .80 (Cronbach's alpha). The support-related choice scale included six items about choice of (a) who they live with, factor loading  =  .51; (b) where they live, factor loading  =  .50; (c) staff who helps at home, factor loading  =  .71; (d) work or day activity, factor loading  =  .51; (e) staff who helps at work, factor loading  =  .67; and (f) case manager/service coordinator, factor loading  =  .56. Internal consistency was also high: alpha  =  .81. For a scale score to be computed, at least 60% of the scale questions must have been answered. Scale scores were calculated by averaging individual item scores.

Independent Variables

We examined the variability in choice accounted for by level of ID, other individual characteristics, self-report response pattern, and service characteristics.

Level of intellectual disability

Level of ID data came from a single NCI item. Individuals reported to have no ID were combined with those listed as having mild ID resulting in values ranging 0 (mild/no ID) to 3 (profound ID).

Individual characteristics

Eight independent covariates were examined in the multiple regression analyses. Table 1 provides detailed information about each independent variable.

Table 1

Coding for Independent Variables and Covariates in Regression Analyses

Coding for Independent Variables and Covariates in Regression Analyses
Coding for Independent Variables and Covariates in Regression Analyses

Self-report response patterns

The self-report variable indicated the proportion (0% to 100%) of the choice items answered by the person rather than by a proxy respondent.

Residence characteristics

We used NCI size and type of residence variables to create a single categorical variable (residence type) with seven levels. The NCI residence variable included four categories of congregate care (group homes, agency owned/operated apartments, specialized institutions for people with IDD including ICFs/MR, and nursing facilities), and three types of individualized housing (parent/relative's homes, host/foster homes, and homes owned or leased by the person). We combined the four categories of congregate care into a single type (agency). Individuals whose residence type was listed as other or unknown were excluded. We also recoded the continuous NCI residence size variable into five categories: Agency 1–3 (1–3 residents with IDD), Agency 4–6 (4–6 residents with IDD), Agency 7–15 (7–15 residents with IDD), Agency 16+ (16 or more residents with IDD), and unknown. The size categories were created with the purpose to capture a greater nuance beyond distinguishing between agency and non-agency settings. This coding allowed for comparisons between agency-operated settings of four different sizes. We subsequently used both of the new variables to create a new residence type variable that had eight categories: (a) own home, 1–3 people with IDD; (b) family home; (c) host family home, 1–3 people with IDD; (d) agency-operated home, 1–3 people with IDD; (e) agency-operated home, 4–6 people with IDD; (f) agency-operated home, 7–15 residents with IDD; (g) agency-operated institution, 16 or more people with IDD; and (h) unknown. The final residence-type variable included the following adjustments: Own home and host family home settings with four or more residents with IDD (fewer than 20 cases) were recoded as agency-operated settings of the reported size (i.e., Agency 4–6, 7–15). Because of the rarity of places indicated to be own home, family home, or host family settings having more than three individuals—when size was not reported for own home—family home or host home settings were coded as having one to three residents with IDD. Individuals in agency-operated settings of unknown size were excluded.

State

The 19 states were dummy-coded with the overall variability accounted for by the states participating in the study. Consistent with the participation agreements, states were not specifically identified in the results.

Results

Descriptive Analyses

Table 2 describes the sample of 8,892 individuals in the Everyday Choice scale analyses; even though the sample was smaller for the Service-Related Choice scale analyses, the sample characteristics were not practically or statistically different. Table 2 summarizes individual characteristics of the sample grouped by residence type. There were statistically significant differences between residence types for each demographic item as indicated by chi-square results. For example, people living in their own homes were most likely to have a mild ID (68.6%), whereas people living in the institutional settings with 16 or more residents were most likely to have a profound ID (47.8%). The proportion of participants with mobility limitations ranged from 16.3% in the host family group to 47.7% in the institution (16 or more residents) group. Overall, 12.4% of the sample reported having a vision or hearing impairment, ranging from 8.5% of people living in their own home to 24.4% of those in the larger institutions. About 44% of the sample was between 18 and 39 years of age, of which 68.4% lived with family members and 22.4% in institutions. Conversely, people aged 55 years and older made up 7.2% of those living with family members and 33.7% of those in institutions. The percentage of people requiring extensive supports to respond to challenging behavior was fairly consistent across the agency-operated settings of different sizes (from 19.0% to 25.5%) and considerably higher than among the non-agency settings.

Table 2

Selected Characteristics of Adults with Intellectual Disability Grouped by Residence Type in 19 States (Everyday Choice, N  =  8892)

Selected Characteristics of Adults with Intellectual Disability Grouped by Residence Type in 19 States (Everyday Choice, N  =  8892)
Selected Characteristics of Adults with Intellectual Disability Grouped by Residence Type in 19 States (Everyday Choice, N  =  8892)
Table 2

Continued

Continued
Continued

Across residence types, 53.3% of the sample reported having a mental health diagnosis or taking a psychotropic medication. The proportions were fairly consistent across setting types (55.1%–71.1%), except that people living with family members were less likely to have this characteristic (33.4%). Approximately 10% of the sample reported autism diagnosis. The percentage of sample members with autism diagnoses was highest among adults living with family members (12.6%), likely reflecting the relatively younger ages of people living with family members and the relatively recent upswing in the diagnosis of autism among children and young adults (Hewitt et al., 2011). Overall, 48.8% of the sample members answered all of the choice questions by self-reports with another 8.4% self-reporting more than half of the questions. About three quarters (74.9%) of people living in their own homes self-reported on more than half of all choice items as compared with about one quarter (24.6%) of large institution residents. Not surprisingly, self-report is highly correlated with level of ID (r  =  .60 for the entire sample). On an average, both level of ID and self-reporting are highest among people living in their own homes and lowest among people living in institutions. Overall, 76.1% of the sample reported speech being their most common means of expression, with those living in their own homes most likely to do so (89.0%) and those living in the larger institutions least likely to do so (48.4%).

Responses to Individual Choice Items

The pattern of responses to the choice items by placement type is summarized in Table 3. With regard to everyday choices variables, 47.9% of respondents reported deciding on their own when they would get up, eat, and go to bed; 65.2% reported deciding on their own how to spend their free time; and 53.7% reported choosing on their own what to buy with their spending money. For the support-related choices, 32.2% reported having control in choosing the people they live with, 23.1% in choosing the place they live, 20.6% in choosing the people who help them at home, 28.8% in choosing where they worked or spent the day, 11.2% in choosing the people who help them at work, and 10.1% in choosing their case manager or service coordinator.

Table 3

Responses on Everyday (N  =  8892) and Support-Related Choice (N  =  6179) Scales Grouped by Residence Type in 19 States

Responses on Everyday (N  =  8892) and Support-Related Choice (N  =  6179) Scales Grouped by Residence Type in 19 States
Responses on Everyday (N  =  8892) and Support-Related Choice (N  =  6179) Scales Grouped by Residence Type in 19 States
Table 3

Continued

Continued
Continued

Correlations Among Independent Variables

Correlations were calculated prior to entering independent variables into regression analyses to determine whether any of the variables were strongly related to one another. For everyday choice, even though there were high correlation coefficients of .60 and .56 (p < .001) between the proportion of individuals who answered most questions by self-report and the level of ID and between self-report and the primary means of expression variable (verbal or nonverbal responses), respectively, the variance inflation factor assumption was not violated when those variables were entered into the regression analyses. For support-related choice, even though the correlations were lower, the pattern followed. Because self-report and primary means of expression variables belong to the same construct, both variables were entered into the regression in a single block (communication). The correlations for all other independent variables were below r  =  .35, with the exception of the relation between behavior profile mental health (r  =  .44 for both everyday and support-related choice). Mental health was made part of the behavior profile block in regression analyses.

Ordinary Least Square Regression Analyses

Inferential statistics in the form of linear ordinary least-squares regression were computed separately for each of the choice scales (see Tables 4 and 5). Independent variables were entered in blocks. The blocks used in the analyses were structured to move from individual setting to state-level factors. They included five blocks of individual descriptors:

  1. level of ID,

  2. physical/sensory limitation (e.g., mobility, vision and/or hearing impairment),

  3. age,

  4. behavioral supports (e.g., challenging behavior, psychiatric condition and/or autism diagnosis), and

  5. communication (e.g., ability to self-report in the survey, primary means of communication).

Table 4

Standardized Regression Coefficients (Beta) from a Linear Ordinary Least-Squares Regression of Selected Variables on Everyday Choice Subscale Scores (19 States, N  =  8892)

Standardized Regression Coefficients (Beta) from a Linear Ordinary Least-Squares Regression of Selected Variables on Everyday Choice Subscale Scores (19 States, N  =  8892)
Standardized Regression Coefficients (Beta) from a Linear Ordinary Least-Squares Regression of Selected Variables on Everyday Choice Subscale Scores (19 States, N  =  8892)
Table 5

Standardized Regression Coefficients (Beta) from a Linear Ordinary Least-Squares Regression of Selected Variables on Support-Related Choice Subscale Scores (19 States, N  =  6179)

Standardized Regression Coefficients (Beta) from a Linear Ordinary Least-Squares Regression of Selected Variables on Support-Related Choice Subscale Scores (19 States, N  =  6179)
Standardized Regression Coefficients (Beta) from a Linear Ordinary Least-Squares Regression of Selected Variables on Support-Related Choice Subscale Scores (19 States, N  =  6179)

Residence type was dummy-coded with agency-operated homes of providers for three or fewer residents as the referent. Significance levels for the other categories of residence type show whether choice for people in these types of residence were significantly different than choice for people living in the agency-operated home of three or fewer residents. The final block was state, for which individual states were dummy-coded. A sampled state was used as the referent for the state variable.

Everyday choice

Table 4 shows that the independent variables included in the regression analysis accounted for 44% of the variability in everyday choice. In the final equation, all of the tested variables except vision or hearing impairments and autism diagnosis contributed a statistically significant proportion of variability in everyday choice. Level of ID, self-report, and primary means of expression accounted for the largest proportions of variability explained. Level of ID accounted for 31% of the variability in everyday choice when tested alone. Together, Blocks 2 through 4 increased the variability explained modestly from 31% to 33%. The communication block accounted for an additional 7% of variability explained. Once personal characteristics were accounted for, type of residence was a significant predictor of everyday choice. In the final equation, accounting for differences in both personal characteristics and state, people in the own home group had significantly more choice than those in the Agency 1–3 group, while those in family home (living with parents or other relatives), Agency 7–15, and Agency 16+ (i.e., institutions with 16 or more residents) reported significantly less everyday choice than those in the Agency 1–3 group. There were not significant differences in choice between the Agency 1–3 group, the host family group, and the Agency 4–6 groups. Finally, the state in which the person lived accounted for a statistically significant additional 2% of variability explained in everyday choice after all other variables were included in the regression.

Support-related choice

Table 5 shows factors associated with support-related choice. Overall, the variables included in the regression accounted for 31% of the variability in reported support-related choice. Mobility, vision, or hearing impairment; age; challenging behavior; mental health; and autism diagnosis did not account for a statistically significant proportion of variability in support-related choice in the final equation. Level of ID accounted for 9% of the variability, with the variables in Blocks 2 through 5 (physical/sensory impairments, age, behavior-related conditions) together adding a total of 3% of variability explained. The residence type variable explained an additional 8% of variability explained, and state added 11% of variability explained. Once the other factors were accounted for, people in their own home group reported significantly more choice than those in the Agency 1–3 group. To a lesser degree, people living in host family arrangements also had more support-related choice than those in the Agency 1–3 group. Conversely, people in the Agency 1–3 group made significantly more choices than those in Agency 4–6, Agency 7–15, and Agency 16+ groups.

Follow-up Analyses of Impact of Level of ID and Residence Type on Choice

Having completed testing of our initial hypotheses, we conducted a series of post hoc follow-up tests to explore the pattern of results. For these analyses, we examined the mean scale score of choice in each level of ID by residence cell, adjusting the means for the significant covariates from the final regression model for each type of choice. Univariate between-subjects fixed-effects analyses of variance (ANOVA) with four levels of ID and seven residence types were computed on means adjusted for the other regression predictors. Analyses for both everyday choice and support-related choice revealed a significant interaction between level of ID and residence type. Figure 1 shows the pattern of differences in adjusted means by level of ID and residence type for everyday choice; Table 6 provides the sample size and adjusted means for each combination of groups. Results for support-related choice are summarized in Table 7 and Figure 2. For each choice scale, a series of level of ID by residence type, Tukey's b, post hoc pair-wise comparisons were computed. The number of pair-wise comparisons that were statistically significant was reported along with the number of comparisons in which a particular level of ID by residence type was associated with greater levels of choice (see Tables 6 and 7).

Figure 1.

Adjusted Means on the Everyday Choice Scale (N  =  8,892).

Figure 1.

Adjusted Means on the Everyday Choice Scale (N  =  8,892).

Figure 2.

Adjusted Means on the Support-Related Scale (N  =  6,179).

Figure 2.

Adjusted Means on the Support-Related Scale (N  =  6,179).

Table 6

Adjusted Means for Everyday Choice Subscale by Level of ID and Residence Type (19 States)

Adjusted Means for Everyday Choice Subscale by Level of ID and Residence Type (19 States)
Adjusted Means for Everyday Choice Subscale by Level of ID and Residence Type (19 States)
Table 7

Adjusted Means for Support-Related Choice Subscale by Level of ID and Residence Type (19 States)

Adjusted Means for Support-Related Choice Subscale by Level of ID and Residence Type (19 States)
Adjusted Means for Support-Related Choice Subscale by Level of ID and Residence Type (19 States)

Everyday choice

As expected based on the regression analyses, the amount of everyday choice reported differed by level of ID, F(3,8891)  =  1585.19, p < .001, η2  =  .27, and residence type, F(6,8891)  =  31.14, p < .001, η2  =  .01. The interaction was also marginally significant at F(18, 8891)  =  1.65, p < .05, η2  =  .00. Within each residence type, all total contrasts for levels of ID were significant using the Tukey's b, post hoc procedure (p < .05). In every case, people with no or mild ID reported the most everyday choice followed by people with moderate ID, people with severe ID, and people with profound ID (see Figure 1). What varied was the relative amount of difference between people with different levels of ID in different placement types.

Among people with mild ID (Figure 1, solid line), those in own home, Agency 1–3, and host family groups had the most choice (adjusted means  =  1.69, 1.65, and 1.64, respectively), whereas people with mild ID in the Agency 16+ group reported the least amount of everyday choice (adjusted χ  =  1.49) with the other setting types in between (see Table 6).

People with moderate ID reported less everyday choice than those with mild or no ID. Similarly to people with mild ID, people with moderate ID living in own home and agency-operated settings of three or fewer residents reported having the most everyday choice (adjusted χ  =  1.57 and 1.53, respectively), whereas people with moderate ID living in institutions (Agency 16+) had the least amount of everyday choice (adjusted χ  =  1.38).

People with severe ID reported less everyday choice than those with mild/no or moderate ID, regardless of setting type. For people with severe ID, those in the Agency 1–3 group reported having the most everyday choice (adjusted χ  =  1.25), whereas those living in institutions (Agency 16+) reported having the least everyday choice (adjusted χ  =  1.09). People with severe ID living in host family residences reported notably less choice (adjusted χ  =  1.14) than those in their own home, Agency 7–15, family home, or Agency 4–6 groups (adjusted χ  =  1.21 to 1.19, respectively).

Finally people with profound ID reported less everyday choice than those with mild/no, moderate, or severe ID, regardless of the type of setting in which they lived. People with profound ID in the Agency 1–3 and host family groups reported having the most everyday choice (adjusted χ  =  1.01 and 1.00), and those in the Agency 16+ and Agency 7–15 groups had the least everyday choice (adjusted χ  =  .83 and .91).

Of the four different placement types supporting three or fewer individuals with ID, the own home settings were most supportive of everyday choice for people with mild and moderate ID, whereas agency-operated homes of three or fewer residents were most supportive of everyday choice for people at severe and profound ID levels.

Support-related choice

As expected, based on the regression analyses, the amount of support-related choice reported differed by level of ID, F(3,6178)  =  232.75, p < .001, η2  =  .10, and type of residence, F(6, 6,178)  =  13.03, p < .001, η2  =  .01. The interaction was small but statistically significant at F(18, 6,178)  =  1.89, p < .05, η2  =  .00. Within each residence type, all total contrasts for levels of ID were significant using the Tukey's b, post hoc procedure (p < .05). In every case, people with no or mild ID reported the most support-related choice followed by people with moderate ID, people with severe ID, and people with profound ID (see Figure 2 and Table 7). There were, however, differences in which residence type was associated with the most support-related choice for individuals with different levels of ID.

Among individuals with each level of ID, those in the own home group reported the most support-related choice (adjusted χ  =  .91, .85, .69, and .62, respectively) followed by those living in family homes (adjusted χ  =  .85, .78, .65, and .56, respectively). Among individuals with mild and moderate ID, those living in institutions (Agency 16+) reported the least amount of choice (adjusted χ  =  .77 and .70). People with severe ID living in host families, agency-operated homes of 4–6 residents or institutions (Agency 16+) had the least support-related choice. For people with profound ID, those living in agency operated settings of 4–6 residents were decidedly lower in reporting support-related choice than those living in other setting types.

Discussion

This study examined the effect of residence size and residence type on everyday choice and support-related choice of 2008–2009 participants in the National Core Indicators Consumer Survey, controlling for other predictors of choice. The adults in the sample lived in seven different residential types (agency-operated settings with 3 or fewer, 4–6, 7–15, or 16 or more residents; homes owned or rented by the service recipient [own home]; host family settings; or homes shared with family members).

A previous analysis of factors associated with choice using a multi-state sample of NCI interviewees found that setting size interacted with level of ID in influencing everyday and support-related choice (Lakin et al., 2008). In that study's multiple regression analysis, 50% of the variability in everyday choice and 47% of support-related choice were accounted for by level of ID, mobility, age, self-reported versus proxy response, setting type (agency versus host family or own home and ICF/MR versus home and community-based services [HCBS] waiver funding), and state. Here, we have built on that study with a much larger sample (19 states rather than 6 states; 8,892 people for everyday choice, and 6,179 people for support-related choice rather than 2,398 people) and extended the analysis by classifying residence in a more nuanced way by size and type and by including individuals who live in a family member's home. Unlike the earlier study, ICF/MR and HCBS were not included as predictor variables because they typically varied with agency size. In the current version of the NCI survey, individuals with ID living with family were also asked about choice of where and with whom to live and thus were included in this analysis. Like the previous study, the present study found that everyday choice was associated with level of ID, mobility, self-report, residence type, and state. Unlike the previous study, this study found the presence of challenging behavior, mental health diagnosis, and primary means of expression (speech or other) were also significant predictors of choice. The present study accounted for slightly less variability in everyday choice than the previous study (44% versus 50%).

The previous study categorized types of residence but without distinctions within group settings for the size of the groups with which people lived. In making such distinctions, we found important differences in the amount of choice by residential arrangement. For people with mild and moderate ID, living in one's own home clearly provided opportunities for more choice in everyday decisions (personal schedule, use of free time, use of one's own resources). Host families, small agency-operated settings (three or fewer residents), and living with family members were also generally predictive of relatively greater control over everyday choices. In contrast, people with severe and profound ID living in agency-operated settings of three or fewer residents tended to have the highest everyday choice scale scores, with people with severe ID living in their own homes, with family members and larger group homes having relatively high group average scale scores as did people with profound ID living in host family settings. Agency-operated settings of 16 or more residents were consistently least supportive of everyday choice across all levels of ID.

The present study accounted for 31% of the variability in support-related choice. The aforementioned previous six-state study accounted for 46% of the variability. Like the previous study, level of ID, self-report, residence type, and state explained significant proportions of variability in support-related choice. Unlike in the earlier study, challenging behavior and mental health diagnosis were not significant predictors of support-related choice. The present study also found that using speech as the primary means of expression was associated with more support-related choice.

Both the earlier six-state study and the present study showed that living in one's own home was associated with more support-related choice when compared to living in any agency-operated setting. People living in their own homes always reported more support-related choice than people in other settings regardless of their level of ID, followed by people living with family members. However, agency-operated homes of 4–6 and 7–15 residents provided nearly as much support-related choice for people with mild and moderate ID. For people with severe and profound ID, living in a host family was associated with relatively more support-related choice.

Both studies reported that state residences contributed to variability explained over and above that explained by personal and service characteristics. The amount of variability associated with state was particularly impressive in the area of support-related choice (10% in the earlier study and 11% in the present study). The consistency of these findings confirmed the conclusion in the previous study that although personal and environmental factors were substantially associated with opportunities to exercise choice, the culture, expectations, resources, and available accommodation options established within individual state, developmental disabilities service systems have significant effect on the extent to which people are actually afforded significant say in the decisions that affect their lives.

Areas for Further Study

Differential outcomes among the four types of settings with three or fewer people with ID

The findings that people with all levels of ID reported greater everyday choice if they lived in their own homes or in the smallest agency settings than if they lived in family-based arrangements (with their own family or a host family) warrants additional consideration. In general, these smaller residential options are by far the fastest growing part of the service system (Lakin et al., 2010). People living with family members made up about 58% of all people receiving support under the auspices of state developmental disabilities services systems, and the combined total of people with ID in family and nonfamily settings of three or fewer people made up 77.5% of all service recipients in the United States (Lakin et al., 2010). The context of family living and the support provided to members with disabilities in the family may well be inherently less oriented to individual choice and more to family-group activity than the supports provided to people in their own homes and in small, homelike, but staffed agency homes. If family and host family homes are less supportive of everyday decision making for people with ID, is this because of the inherent nature of the environment, or might it be it that in the context of family care caregivers have not had the training, expectations, or opportunities to learn about the importance of decision making to individual development, autonomy, and respect? Notably, people living in host family homes were the least likely to self-report on everyday choice questions (i.e., most likely to have proxy respondents), with the one exception of people living in institutional settings of 16 or more residents. People living in the smallest setting (1–3 residents with disability) host home were reported to be least likely to have someone else make decisions without their input about all three of the everyday choices as well as about five of the six support-related choices.

Proxy responding and primary means of communication

Both regressions show significant effects for self-report and primary means of expression. Controlling for other variables in the regression analyses, these effects suggested a clear positive relation between the degree of self-reporting and verbal communication and a higher level of reported choice. This pattern was the clearest in everyday choice, where the communication block accounted for an additional 7% of variance explained. Of particular concern in this association was the extent to which people with limited communication abilities were reported to not participate in decision making. Although, clearly, cognitive limitations would be expected to be associated with independent decision making, it would seem very desirable that higher proportions of individuals, especially those with severe and profound levels of ID and communication limitation, were at minimum reported to participate in or have input in the decisions affecting their lives. One factor that may have contributed to the apparent non-involvement in many decisions by individuals with communication limitation was the extremely low rate of availability of augmentative and alternative communication aids to service users (Stancliffe et al., 2010).

Difference between everyday choice and support-related choice

This study, like the previous NCI study of choice (Lakin et al., 2008), found that people with ID receiving supports from the IDD service system reported having much less support-related choice (adjusted overall M  =  .78) than everyday choice (adjusted overall M  =  1.42) on a scale from 0 (none) to 2 (complete). Like the previous study, this difference existed at each level of intellectual disability. We have shown elsewhere that people who chose where and with whom they live experienced a range of more positive lifestyle outcomes (Stancliffe, Lakin, Taub, Chiri, & Byun, 2009). In addition to having little support-related choice, the absence of key choices about where and with whom to live may have consequences such as increased loneliness, decreased liking of one's home, and decreased reported happiness. Additional research is needed to identify support practices that produce increased choice-making opportunities for people with ID, regardless of their level of ID.

Limitations and Cautions

Although this study is one of the largest to date looking at everyday and support-related choice among American adults with ID who receive supports from the IDD system, several cautions are warranted. For example, the large sample size means that relatively weak effects may be found to be statistically significant. To counteract this, we reported effect sizes to allow readers to judge the practical significance of our findings. While the follow-up analyses of differences in the amount of choice by level of ID and residence type revealed many statistically significant differences, the analyses were conducted as post hoc analyses. Future analyses of both the NCI data set and other data sources using a prior hypothesis testing would help clarify the extent to which these results are due to random error. The interaction component for both everyday choice and support-related choice, although statistically significant, accounted for less than 1% of the variability in those outcomes. Given the inherent instability of interaction effects, further direct investigation of level of ID by setting-type interactions would be warranted to confirm or disconfirm these findings.

The NCI Consumer Survey data provide a wonderful opportunity to examine outcomes for adults with ID receiving supports in the IDD system. It is the only ongoing data collection effort in the United States that provides a random sample of support recipients in many different states. We look forward to continuing to refine our analyses of outcomes to include examination of cross-sectional differences over time and to take advantage of the ongoing program of quality improvement for the NCI data collection effort.

The Human Services Research Institute (HSRI) employs several of the authors. HSRI coordinates the National Core Indicators project and receives a fee from participating states. The authors alone are responsible for the content and writing of the article. Preparation of this article was supported by Grant #H133G080029 from the National Institute on Disability and Rehabilitation, U.S. Department of Education (federal funds for this three-year project totaled $599,998).

References

References
Americans with Disabilities Act of 1990, Pub. L. 101-336, 42 U.S.C. 12101 et seq
.
Burchard
,
S. N
.,
Hasazi
,
J. E
.,
Gordon
,
L. R
.,
&
Yoe
,
J
.
(
1991
) .
An examination of lifestyle and adjustment in three community residential alternatives
.
Research in Developmental Disabilities
,
12
,
127
142
.
Developmental Disabilities Assistance and Bill of Rights Act of 2000, Pub. L. No. 106-402, 42 U.S.C. 15001
.
Emerson
,
E
.,
Robertson
,
J
.,
Gregory
,
N
.,
Hatton
,
C
.,
Kessissoglou
,
S
.,
Hallam
,
A
.,
...
Netten
,
A
.
(
2000
) .
Quality and costs of community based residential supports, village communities, and residential campuses in the United Kingdom
.
American Journal on Mental Retardation
,
105
,
81
102
.
Heller
,
T
.,
Miller
,
A. B
.,
&
Factor
,
A
.
(
1999
) .
Autonomy in residential facilities and community functioning of adults with mental retardation
.
Mental Retardation
,
37
,
449
457
.
Hewitt
,
A. S
.,
Stancliffe
,
R. J
.,
Johnson Sirek
,
A
.,
Hall-Lande
,
J
.,
Taub
,
S
.,
Engler
,
J
.,
Bershedsky
,
J
.,
Fortune
,
J
.,
&
Moseley
,
C. R
.
(
2012
) .
Characteristics of adults with autism spectrum disorder who use adult developmental disability services: Results from 25 US states
.
Research in Autism Spectrum Disorders
,
6
(
2
) ,
741
751
.
Lakin
,
K. C
.,
Doljanac
,
R
.,
Byun
,
S
.,
Stancliffe
,
R
.,
Taub
,
S
.,
&
Chiri
,
G
.
(
2008
) .
Choice making among Medicaid Home and Community-Based Services recipients and ICF/MR residents in six states
.
American Journal on Mental Retardation
,
113
(
5
) ,
325
342
.
Lakin
,
K. C
.,
Larson
,
S
.,
Salmi
,
P
.,
&
Scott
,
N
.
(
2010
) .
Residential services for persons with developmental disabilities: Status and trends through 2009
.
Minneapolis, MN
:
University of Minnesota, Research and Training Center on Community Living
.
Olmstead et al. v. LC et al. 1999, 527 U.S. 581
.
Smith
,
G
.,
&
Ashbaugh
,
J
.
(
2001
) .
National Core Indicators Project: Phase II consumer survey technical report
.
Cambridge, MA
:
Human Services Research Institute
.
Available at http://www.hsri.org
Stancliffe
,
R. J
.
(
1997
) .
Community living-unit size, staff presence and resident's choice-making
.
Mental Retardation
,
35
,
1
9
.
Stancliffe
,
R. J
.,
Abery
,
B. H
.,
&
Smith
,
J
.
(
2000
) .
Personal control and the ecology of community living settings: Beyond living-unit size and type
.
American Journal on Mental Retardation
,
105
(
6
) ,
431
454
.
Stancliffe
,
R. J
.,
&
Keane
,
S
.
(
2000
) .
Outcomes and costs of community living: A matched comparison of group homes and semi-independent living
.
Journal of Intellectual and Developmental Disability
,
25
(
4
) ,
281
305
.
Stancliffe
,
R. J
.,
&
Lakin
,
K. C
.
(
1998
) .
Analysis of expenditures and outcomes of residential alternatives for persons with developmental disabilities
.
American Journal of Mental Retardation
,
103
(
6
) ,
552
568
.
Stancliffe
,
R. J
.,
Lakin
,
K. C
.,
Larson
,
S. A
.,
Engler
,
J
.,
Taub
,
S
.,
&
Fortune
,
J
.
(
2011
) .
Choice of living arrangements
.
Journal of Intellectual Disability Research
,
55
(
8
) ,
746
762
.
Stancliffe
,
R. J
.,
Lakin
,
K. C
.,
Taub
,
S
.,
Chiri
,
G
.,
&
Byun
,
S
.
(
2009
) .
Satisfaction and sense of well-being among Medicaid ICF/MR and HCBS recipients in six states
.
Intellectual and Developmental Disabilities
,
47
(
2
) ,
63
83
.
Stancliffe
,
R. J
.,
Larson
,
S. A
.,
Auerbach
,
K
.,
Engler
,
J
.,
Taub
,
S
.,
&
Lakin
,
K. C
.
(
2010
) .
Individuals with intellectual disabilities and augmentative and alternative communication: Analysis of survey data on uptake of aided AAC, and loneliness experiences
.
Augmentative and Alternative Communication
,
26
(
2
) ,
87
96
.
United Nations
. (
2006
) .
Convention on the rights of persons with disabilities
.
New York, NY
:
Author
.
Wehmeyer
,
M. L
.,
&
Bolding
,
N
.
(
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
.,
&
Metzler
,
C. A
.
(
1995
) .
How self-determined are people with mental retardation? The national consumer survey
.
Mental Retardation
,
33
(
2
) ,
111
119
.

Author notes

Editor-in-Charge: Susan L. Parish

Authors:

Renáta Tichá (e-mail tich0018@umn.edu), University of Minnesota, Research and Training Center on Community Living, 150 Pillsbury Dr., SE, Minneapolis, MN, 55455, USA; K. Charlie Lakin, National Institute on Disability and Rehabilitation Research; Sheryl A. Larson, University of Minnesota, Research and Training Center on Community Living; Roger J. Stancliffe, University of Sydney, Australia; Sarah Taub, Human Services Research Institute; Joshua Engler, Human Services Research Institute; Julie Bershadsky, Human Services Research Institute; and Charles Moseley, National Association of State Directors of Developmental Disabilities Services.