Version classiqueVersion mobile
OpenEdition Books

A Quarter-Century of Normalization and Social Role Valorization

 | 
Robert J. Flynn
, 
Raymond Lemay

Part 3: Critical Perspectives on Normalization and Social Role Valorization

9. Are Normalization and Social Role Valorization limited by competence?1

Laird W. Heal

Texte intégral

1 INTRODUCTION

1.1 NORMALIZATION & SOCIAL ROLE VALORIZATION

  • 1 The present report was supported in part by the Office of Special Education and Rehabilitative Serv (...)

1We are all familiar with the defining characteristics of Nirje’s Normalization. Its superseding principle is attributed by Nirje to Bank-Mikkelsen: “To let the mentally retarded obtain an existence as close to the normal as possible” (Nirje, 1969, p. 181). Corollaries to this principle include normal rhythms of the day, week, month, year, and lifetime; integration of people with disabilities with their age peers into the settings that characterize those used by their age peers; the opportunity to exercise their choices, wishes, and desires; and economic support commensurate with that of any other social security recipient (Nirje, 1992).

2While embracing the principle of Normalization, Wolfensberger has been impressed by a construct that might be seen by some to have made Normalization necessary in the first place—valuation. Wolfensberger, through Social Role Valorization (SRV), which he proposes as a social science theory to replace Normalization, urges human service professionals to pursue whatever avenues they might find to increase the valuation of people whose physical or mental deviance might engender devaluation. Especially important in Wolfensberger’s view is to associate people who are disabled with positive and valued imagery.

3Having common roots, these two constructs have much in common. Both of these approaches are ideologically based and provide guidelines for human service providers. Social Role Valorization, and to some extent Normalization, especially as they have been quantified by PASS (Wolfensberger & Glenn, 1975) and PASSING (Wolfensberger & Thomas, 1983), are prescriptive. Both assume that the best strategy for individuals with disabilities is to pursue cultural norms. Nirje and Wolfensberger distinguish between cultural norms and statistical norms. They would not urge people with disabilities to pursue the normal daily seven hours watching television, normal rates of divorce, or other statistically prominent activities of their age peers. Normalization, and especially Social Role Valorization, reserves for disabled individuals the pursuit of culturally ideal, not culturally probabilistic, roles and expectations.

1.2 QUALITY OF LIFE (QOL)

4Most would apee that providing people with disabilities a life as close as possible to cultural norms and further assigning them roles and activities designed to increase their value has been an inestimable benefit to people with disabilities. Unfortunately, very little direct research has been carried out to ascertain the extent to which these constructs can have a positive impact upon the lives of disabled persons. For instance, does the implementation of SRV increase the likelihood that disabled persons will be afforded the “good things in life,” which, according to Wolfensberger (1992), is the ultimate goal of SRV? Wolfensberger (1980) suggests that one may find at least indirect evidence and sometimes direct evidence in related research, even though Normalization and SRV might not even be referred to by the authors.

5Quality-of-life research is one possible area that one may argue is related to SRV and Normalization and that may hint at some of the effects and limitations of their implementation. Wolfensberger (1994), though highly critical of the quality-of-life construct, nonetheless argues that SRV is its equivalent and even suggests that SRV replaces quality of life.

6Quality of life is a simple concept that has generated much discussion although some have suggested that it has turned out to be intractable to investigation, both ideologically and empirically (Goode, 1994; Heal, Borthwick-Duffy & Saunders, 1996; Jamieson, 1993; Parent, 1993; Parmenter, 1992; and Schalock, 1990, 1996). Ideally all approaches to the assessment of quality of life—that is, through cultural norms as embodied by Normalization and Social Role Valorization, through statistical norms, through choices that individual people with disabilities make, through well-being as seen by intimate acquaintances, through subjective well-being as reported by the individual clients themselves, through professional judgment, or through objective indicators of quality of life—should indicate similar levels when they are compared to one another. Unfortunately, the results of assessment using these different approaches have not always been parallel.

7The many approaches to the assessment of quality of life were summarized quite nicely in a recent paper by Hughes, Huang, Kim, Eisenman & Killian (1995). A summary of the classification developed inductively by Hughes et al. appears in Table 9.1. Hughes’ review is very revealing in the present context. First, one is immediately struck by the variation in the instruments that have been used to assess quality of life. These measurements range from psychological well-being and personal satisfaction to the existence of and extent of support services from family and community. “Well-being” heads Hughes’s list in popularity, with 31 references being cited. Unfortunately, this count mixes both interviews of clients themselves and interviews of informants describing their perceptions of clients. We shall see later that clients and informants do not necessarily agree. Second, Social Role Valorization is not mentioned as a category, and Normalization is measured more indirectly than directly. It is surprising that the ideology that has dominated the revolution of human services during the past 25 years would not be seen as a larger component of quality of life. Two other large categories are conspicuous by their absence. The first is the goodness of fit between one’s needs and one’s service supports. This conceptualization of quality of life has been offered in two variations by Schalock and Jensen (1986) and Saunders and Spradlin (1991), and is implied by the new definition of mental retardation by the American Association on Mental Retardation (AAMR, 1992), which stresses the supports necessary to minimize the effect of one’s disability on one’s lifestyle. The second is spiritual fulfillment, which is for many individuals the only QOL pursuit.

8The pages that follow report two recent investigations that bear on the dimensionality of quality-of-life assessment. The first considers three conceptually different dimensions of quality of life—esteem, which Hughes et al. (1995) would presumably list under Category I, “well-being,” of Table 9.1; competence, which would be listed under Category IV, “Self-Determination and Supporting Skills,” of Table 9.1; and support, which would appear under Category VI of Table 9.1. The second investigation compares the self-reports of satisfaction given by clients with mental retardation and the reports regarding quality of life by informants who know the clients well. Both clients’ and informants’ assessments were classified by Hughes under the first major heading in Table 9.1, “Well-being.”

2 QUALITY-OF-LIFE DIMENSIONS

9The first investigation was inspired by the existence of a unique data set from the National Longitudinal Transition Study (Valdes, 1989; Wagner et al., 1991), which featured thousands of variables describing a national sample of current and past special-needs high-school students. This dataset offered an opportunity to evaluate the geoeconomic, educational, family, and personal characteristics that might influence the quality of life of the graduates of special education programs.

10For this study Wagner et al. (1991) contacted a stratified probability sample of all students with disabilities attending U.S. high schools in 1985 and gathered in-school and out-of-school information about them in 1987 from their school records, school personnel, and parents. Three quality-of-life composites—esteem, independence, and support—were constructed from 17 of these variables, using conventional item analysis procedures. These were related to 28 geoeconomic, demographic, cognitive, disability, and school program variables using a canonical correlation.

2.1 DATA SOURCES

11The National Longitudinal Transition Study had three major sources of data: (a) the Parent Survey, (b) School Records Abstracts, and (c) the Survey of Secondary Special Education Programs. In the Parent Survey, parents or de facto guardians of the selected youths were asked by telephone in 1987 to provide information on youths’ family background, characteristics, special services, educational attainment (including postsecondary education), employment experiences, social integration, and their expectations for the youths in the future. School Record Abstracts consisted of information abstracted from the school records of sampled youths for their most recent year in high school (either the 1985-1986 or the 1986-1987 school year). Information abstracted from school records related to disability description and classification, courses taken, grades achieved (if appropriate), school placement, related services received from the school, status at the end of the year, attendance, and other records.

12

TABLE 9.1. CLASSIFICATION OF QUALITY OF LIFEa

TABLE 9.1. CLASSIFICATION OF QUALITY OF LIFEa

a Adapted from Hughes et al. (1995). b The number of references in which Hughes et al. (1995) reported an application

13The Survey of Secondary Special Education Programs was completed by school personnel in schools attended by sample youths in the 1986-1987 school year. School personnel provided schoolwide information on student enrollment, staffing, and programs, and related services offered secondary special education students, policies affecting special education programs and students, and community resources (paraphrased from Wagner, 1989, appendix).

2.2 SAMPLING

14Sampling was completed in two major stages, school sampling and student sampling. In sampling schools, 712 school districts that served students with disabilities at the seventh grade level or higher were selected to represent as completely as possible the 96 strata of four regions of the country, six size (student enrollment) categories, and four levels of district poverty. Student rosters were provided by 325 (45.6%) of the 712 school districts that were approached. From these rosters, 12,833 students or their families were approached, and 65.5% agreed to participate. The sampling is carefully documented by Javits and Wagner (1987). The present study was limited to 713 (21.2%) of the 3,357 students who were out of school when data were gathered in 1987, and had complete data on all 54 variables chosen for the present analyses

2.3 DEPENDENT VARIABLE: QUALITY OF LIFE

15The composite scale of quality of life should ideally assess all of the categories noted by Hughes et al. (1995) and listed in Table 9.1. However, the variables in the National Longitudinal Transition Study data set permitted the construction of composite scales for only three domains of quality of life—self-esteem, independence, and security and support. The variables used in the construction of each of these domains are presented in Table 9.2.

16For each individual in the sample, the composite scale for each domain of quality of life was constructed as follows. First, the cumulative percentage of each variable was computed. The response of each individual was then represented by the cumulative percentages of the 713 cases scoring at his (or her) response level or below. For example, suppose that the valid responses of variable x were 1, 2, 3, and 4, and the corresponding cumulative percentages for the study sample of 713 individuals were 20, 50, 70, and 100 respectively. The responses of individuals were then transformed to their cumulative percentage as follows: 20 for a response of 1; 50 for 2; 70 for 3; and 100 for 4. Thus, if an individual’s level was the highest in the sample, it was represented by the cumulative percentage of 100. This transformation provided a simple metric for combining or comparing ordinal variables. The average of the cumulative percentage scores over items was taken as the quality-of-life level of each domain for each individual in the sample. Thus, the individuals in the sample could be differentially valued on their overall quality of life.

2.4 PREDICTOR VARIABLES

17The composite scales for self-esteem, independence, and security and support reflect three components of quality of life. In an attempt to uncover the meaningful predictors of each of these domains, the potential predictors were drawn from a list of those that had been used in two previous studies, one predicting home independence (Heal & Rusch, 1994) and the other predicting employment (Heal & Rusch, 1995). Predictors consisted of community characteristics, parent characteristics, student characteristics, and school program characteristics.

2.5 INFERENTIAL STATISTICS

18A multiple regression analysis was used to identify the significant predictors of each of the three composites: esteem, independence, and security and support. For this, the composite scale for each of the three domains of quality of life was regressed separately onto the predictors in Table 9.3. Parallel to Heal and Rusch (1994, 1995), the 28 potential predictors were entered in eight ordered blocks into each regression analysis. Their order of entry was specified in advance. The strategy used was to enter geoeconomic characteristics first, followed by family characteristics, three blocks of students’ personal characteristics, then disability categories, next school philosophy, and finally a block of individual students’ school programs. The rationale for such a blockwise regression analysis is that earlier blocks of variables function as the control variables for later blocks (Heal & Rusch, 1994). In most cases, control variables are characteristics of the individual and his or her environment that are relatively permanent, that is, not amenable to intervention. By entering these into a regression equation before one enters programmable features of individuals and/or “their” environment, one can estimate the extent to which one’s schooling can improve one’s future quality of life through programmatic intervention. A canonical correlation analysis was used because the composite scales for the three dimensions of the quality of life were correlated. (See Table 9.4.) The canonical correlation analysis facilitates the study of the relationship between a set of correlated criterion variables and a set of correlated predictor variables, with the objective of predicting one from the other. This approach identifies a canonical variate (linear combination) for the predictor variables that has the highest correlation with the canonical variate (linear combination) of the criterion variables.

TABLE 9.2. COMPOSITION OF EACH DOMAIN OF THE QUALITY-OF-LIFE SCALE: SELF-ESTEEM, INDEPENDENCE, AND SECURITY AND SUPPORT

TABLE 9.2. COMPOSITION OF EACH DOMAIN OF THE QUALITY-OF-LIFE SCALE: SELF-ESTEEM, INDEPENDENCE, AND SECURITY AND SUPPORT

a Item required reverse scoring to make low values reflect low quality of life and high values reflect high quality of life

TABLE 9.3. PREDICTORS OF QUALITY OF LIFE GROUPED BY CHARACTERISTICS

TABLE 9.3. PREDICTORS OF QUALITY OF LIFE GROUPED BY CHARACTERISTICS

TABLE 9.4. RAW SCORE MEANS, STANDARD DEVIATIONS, AND TRANSFORMED SCORE ITEM-DOMAIN CORRELATIONS OF THE VARIABLES USED IN EACH DOMAIN OF QUALITY OF LIFE—ESTEEM, INDEPENDENCE, AND SECURITY AND SUPPORT

TABLE 9.4. RAW SCORE MEANS, STANDARD DEVIATIONS, AND TRANSFORMED SCORE ITEM-DOMAIN CORRELATIONS OF THE VARIABLES USED IN EACH DOMAIN OF QUALITY OF LIFE—ESTEEM, INDEPENDENCE, AND SECURITY AND SUPPORT

a Item required reverse scoring to make low values reflect low quality of life and high values reflect high quality of life.
b r is the correlation of the item with the remaining items from its dimension. Scores for this analysis were the cumulative percentages associated with each of the ordered categories for each variable.
*p <.001.

2.6 RESULTS

19The results are summarized in Tables 9.4 through 9.6. Table 9.4 shows the raw score means and standard deviations of the component variables of the three quality-of-life domains as well as the correlations of each cumulative percentage score with its domains and the intercorrelations among the three. Intercorrelations among the three domains indicate that esteem and independence are positively correlated, and that both are negatively correlated with support. Cronbach’s alphas were 0.494, 0.796, and 0.277, for esteem, independence, and security and support respectively.

20Table 9.5 shows the means and standard deviations of the predictor variables as well as their correlations with each of the three quality-of-life domains. Each quality-of-life domain is characterized by a different set of predictor correlates. Esteem is greater for those with mild disabilities (learning disabilities, speech disorders, and emotional disturbance) who have had minimal special education. Support is greater for those minimal special education. Independence is greater for with more severe disabilities who have had substantial younger students with mild disabilities who have had special education.

TABLE 9.5. MEANS AND STANDARD DEVIATIONS OF POTENTIAL PREDICTORS OF QUALITY OF LIFE AND THEIR CORRELATIONS WITH EACH OF THE THREE QUALITY-OF-LIFE DOMAIN VARIABLES

TABLE 9.5. MEANS AND STANDARD DEVIATIONS OF POTENTIAL PREDICTORS OF QUALITY OF LIFE AND THEIR CORRELATIONS WITH EACH OF THE THREE QUALITY-OF-LIFE DOMAIN VARIABLES

a p <. 001 = ±0.13.
*|
r | ≥ 0.30.

FIGURE 9.1. PROPORTION OF VARIANCE (R2 CHANGE) ATTRIBUTABLE TO SUCCESSIVE BLOCKS OF PREDICTOR VARIABLES FOR EACH OF THE THREE QUALITY-OF-LIFE DOMAINS

FIGURE 9.1. PROPORTION OF VARIANCE (R2 CHANGE) ATTRIBUTABLE TO SUCCESSIVE BLOCKS OF PREDICTOR VARIABLES FOR EACH OF THE THREE QUALITY-OF-LIFE DOMAINS

Predictor block labels are abbreviations of the eight successive numbered sections from Table 9.3

21A separate multiple regression analysis was conducted for each of the three quality-of-life domains as its dependent variable and the 28 variables from Table 9.3 as the predictor variables entered in the eight ordered blocks shown in Table 9.3: county economic characteristics, family characteristics, noncognitive personal characteristics, cognitive competence, maladaptive behaviors, primary disability, school characteristics, and school programs. Figure 9.1 presents the cumulative proportion of variance (R2) for each domain of quality of life for each successive block of variables. The final multiple R2 was 0.286 for esteem, 0.451 for independence, and 0.368 for support. The sizes of these multiple R2s were ordered (limited) as one would expect from their alpha reliabilities. The location and family characteristics chosen for this analysis contributed about 0.05 to the total R2; individual characteristics—demographic, IQ, behavior problems, and especially disability category—accounted for an additional 20% in esteem, 40% in independence, and 24% in support.

22School characteristics and individual students’ school programs accounted for a final 5% of the variance in each of the three quality-of-life domains. This final increment added to the proportion of variance was 0.046 for independence, F(8,682) = 10.31, p < 0.001; 0.066 for support, F(8,682) = 6.23, p < 0.001; and 0.045 for esteem, F(8,682) = 4.54, p< 0.001.

23The canonical loadings (the correlation between each variable in the predictor or criterion set and its respective canonical variate), proportion of variance (R2) in each set of variables (criterion or predictor set) that is explained by the respective canonical variates, and the canonical correlations (correlations between the criterion and predictor canonical variates) and redundancy coefficients (proportion of variance in a predictor set multiplied by the squared canonical correlation) are presented in Table 9.6. Three canonical variates were identified—the most possible because the smaller set of variables (i.e., the criterion set) had only three domains—each one defined by positive correlations with two of the QOL domains and a negative or negligible correlation with the third. The first might be labeled “competence” because of its association with youth, high IQ, bad behavior record, mild disability classification, and minimal exposure to special education services. The second might be labeled “sensory disability” because of its association with sensory disabilities, IQ, and academic and/or vocational (as opposed to special) school placements. The third might be called “valued support” because of its positive association with the esteem and support quality-of-life domains, with having other disabled siblings, and with attendance in special employment and vocational school programs.

24The proportion of criterion set variance accounted for by the canonical variates is inflated by the fact that there are as many quality-of-life variables as canonical variates, assuring that the sum of the proportion of variances accounted for by three canonical variates is 1.0.

25The first canonical variate (competence) accounts for 59.2% of the variability in the criterion set. The second (sensory disability) and third (valued support) canonical variates account for 20.2% and 20.6% of the criterion set variability, respectively. The squared canonical correlations indicate that the first canonical variate of the criterion set shares 50.7% of its variability with the first canonical variate of the predictor set. The second and third criterion canonical variates share 24.3% and 9.6% of their variability with the corresponding canonical variates of the predictor set, respectively. These result in the redundancy coefficients of 0.300, 0.049, and 0.020 for the first, second, and third canonical variates of the predictor set. The redundancy coefficient is the variance of one set of variables that can be accounted for by a canonical variate from the other set (Dillon & Goldstein, 1984). In other words, the first canonical variate of the predictor set accounts for 30% of the variability of the criterion set (0.592 x 0.507). The second and third canonical variates of the predictor set account for 4.9% (0.202 x 0.243) and 2.0% (0.206 x 0.096) of the criterion set variability, respectively. These three redundancy coefficients sum to 36.9%, the percentage of the criterion set variability accounted for by the linear combination predictor set of variables.

TABLE 9.6. CANONICAL LOADINGS, ASSOCIATED SQUARED MULTIPLE CORRELATIONS, CANONICAL CORRELA TIONS, AND REDUNDANCY COEFFICIENTS FOR QUALITY OF LIFE AND ALL 28 PREDICTORS Canonical Variates (Dimensions)

TABLE 9.6. CANONICAL LOADINGS, ASSOCIATED SQUARED MULTIPLE CORRELATIONS, CANONICAL CORRELA TIONS, AND REDUNDANCY COEFFICIENTS FOR QUALITY OF LIFE AND ALL 28 PREDICTORS Canonical Variates (Dimensions)

*Indicates canonical loading >0.3.
*Indicates canonical loading > 0.5

2.7 DISCUSSION

26In summary, the primary dimension along which these former special education students varied was competence (a dimension whose positive pole was characterized by high independence), high esteem (few stigmata and high employment), and minimal dependency on family or government. Correlates of this factor from the predictor set were mild disability, more integrated and less special school programs, and young age. Presumably this last correlate reflected the common finding that more competent individuals disappear from disability service networks soon after high school graduation, leaving only younger, more competent graduates in a follow-up cohort.

27The second dimension might be called “sensory disability” because of its high loadings from hearing and visual disabilities. It had moderate positive correlations with independence (adaptive skills and estimated future independence), family and government support, and a moderate negative correlation with esteemed status, including employment. The negative pole of this dimension was also characterized by mild disability and mental retardation. The second canonical variate of the predictor set accounted for only 4.9% (redundancy coefficient = 0.049) of the variability in the criterion set. Finally, the third dimension, which had an almost negligible redundancy coefficient of 0.020, was labeled “valued support” because of its high positive correlation with the esteem criterion domain and moderate positive correlation with the support criterion domain.

28The identification and labeling of factors are always subject to interpretations. This problemis compounded in retrospective studies, where the selection of variables is limited to those available. Furthermore, once QOL variables have been identified, their assignment to QOL subscales is subject to the judgment of the investigator. For example, the dimension of esteem reflected evidence of achievement or absence of stigmata, whereas security and support reflected evidence of family or government support. One could challenge the assignment of specialized transportation to government support instead of stigmata. Independence is probably less controversial, being comprised of three adaptive skill variables and two parents’ attribution-of-skill variables. Given their composition, it is not surprising that these subscales were factorally multidimensional when they were entered into a canonical analysis.

29The most salient implication of these results is that quality of life for young adults with disabilities may be defined primarily by a single dimension, competence, which is positively correlated with esteem and independence and negatively correlated with family and governmental support. Furthermore, to the degree that multidimensionality is indicated in these results, it appears related primarily to type of disability—severity of disability, especially mental disability, defining the first factor, and sensory disability defining the second.

3 DISENTANGLING COMPETENCE AND SATISFACTION

30Notwithstanding the seductive parsimony of such a straightforward interpretation, closer scrutiny is probably warranted. While competence and mastery are undoubtedly very satisfying and may be a central developmental drive (e.g., Harter, 1977, 1978), quality of life is likely to be manifested on several planes, not all of which are variations of competence. Like all analyses, the one just reported is a slave to its data set. The National Longitudinal Transition Study (NLTS) focused on disability and competence, not quality of life, limiting the range of variables that addressed QOL nuances. Table 9.1 shows the large number of QOL topics for which there were insufficient or incomplete NLTS items. Excluded were material comforts, social relationships with family and friends, the social responsibilities that accompany these relationships, spiritual fulfillment, creative expression and active recreation, passive recreation, and Normalization and Social Role Valorization. Other multivariate analyses (Borthwick-Duffy, 1986; Halpern, 1993; Harner & Heal, 1993; Heal & Chadsey-Rusch, 1985; Schalock, Keith, Hoffman, & Karan, 1989) have revealed convincing multidimensional structures for these constructs. Furthermore, recent thinking has accorded personal choice the highest value in calculating quality of life for people with disabilities (Goode, 1994; Heal, Borthwick-Duffy, & Saunders, 1996; Saunders & Spradlin, 1991), a dimension that was precluded by the nature of the NLTS data set. Exercising ones own choices is empirically, logically, and legally correlated with competence, but within each competence level these authors propose or imply that quality of life be indexed by the degree of control that each person with a disability can exercise over his or her environment.

31Finally, the NLTS data set has no variables that assess, directly or indirectly, the former students’ own claims of lifestyle satisfaction or subjective well-being. It is illogical and arrogant to select and assess quality-of-life components without asking the subject of study whether he or she values them. In terms of the classification of quality of life shown in Table 9.1 (Hughes et al., 1995), this analysis requires a separation of the well-being domain into that inferred from psychological indicators and informants’ judgments on the one hand and that claimed by the respondent with disabilities herself on the other. To this end, Harner (1991) gathered parallel data by two methods, a client interview and an informant interview. The second investigation compared clients’ and informants’ perceptions of clients’ well-being, using what appeared to be comparable components of these two subscales.

TABLE 9.7. DESCRIPTIVE STATISTICS FOR THE LIFESTYLE SATISFACTION SCALE (LSS) AND CONSTRUCTED QUALITY-OF-LIFE QUESTIONNAIRE (QOLQ) SUBSCALES (N = 46)

TABLE 9.7. DESCRIPTIVE STATISTICS FOR THE LIFESTYLE SATISFACTION SCALE (LSS) AND CONSTRUCTED QUALITY-OF-LIFE QUESTIONNAIRE (QOLQ) SUBSCALES (N = 46)

a The QOLQ item numbers are taken from Schalock, Keith, & Hoffman, 1990; the LSS item numbers are taken from the 76-item trial LSS (Harner, 1991). Both lists are available from the author. The value of each QOLQ item could range from 1 (low) to 3 (high); the value of each LSS item could range from-2 (very dissatisfied) to +2 (very satisfied). b Cronbach’s alpha coefficient of internal consistency. LSS coefficients are from a crossvalidation sample. c Abbreviations: QOL = quality of life assessed by the newly constructed QOLQ subscales; SAT = satisfaction assessed by the LSS subscales; JOB = job; COM = community, home, and neighborhood; PAL = friends and most social activities; REC = recreation and most leisure activities; CTL = control of one’s choices, self-determination

3.1 METHOD

32Forty-six clients from developmental disabilities service agencies were recruited for Multifaceted Lifestyle Satisfaction (LSS; Harner, 1991) interviews while their direct care supervisors were recruited for 1990 Quality of Life Questionnaire (QOLQ; Schalock Keith, & Hoffman, 1990) interviews. Scores from presumably parallel items of the two instruments were correlated to estimate the agreement of caretakers with their clients regarding the quality of their clients’ lives.

3.2 SUBJECTS

33Subjects were 46 adults (21 males and 25 females) with mental retardation who were drawn from five service provider agencies in west central Indiana and east central Illinois. These adults were drawn unsystematically for assessment with both the LSS and QOLQ questionnaires from a larger sample of 149 adults who were assessed with only the LSS. The mean IQ from agency records was 60.8 (SD = 13.7), 4 being classified as severely retarded, 7 as moderately retarded, and 35 as mildly retarded or borderline. Ages ranged from 22 to 65 with a mean of 35.7. Individuals in the sample lived in one of five out-of-home community residential placements: large (16 or more beds) intermediate-care facilities (n = 9), small (15 or fewer beds) intermediate-care facilities (n = 7), sheltered group homes (eight or fewer beds) (n = 11), apartment training programs (three or fewer beds) (n = 9), or semi-independent and independent programs (n = 10).

3.3 ASSESSMENT INSTRUMENTS—LSS AND QOLQ

34The 1990 Quality of Life Questionnaire (Schalock et al., 1990) contains four subscales, each with 10 items: Satisfaction; Empowerment and Independence; Competence and Productivity; and Social Belonging and Community Integration. Each item requires the informant to decide on behalf of each client among three levels of quality of life.

35For instance, Item 1 (Satisfaction subscale) was

Overall, would you say that life:
(3) Brings out the best in you
(2) Treats you like everybody else
(1) Doesn’t give you a chance.

36Item 26 (Empowerment subscale) was

Do you have a key to your home?
(3) Yes, I have a key and use it as I wish
(2) Yes, I have a key but it only unlocks certain areas
(1) No.

37The 1991 LSS is comprised of 53 items in six subscales, five of which were used in the present investigation. Each question requires a response of yes or no. For instance, question 1 is “How do you like living here?”; question 10 is “Do you have enough things to do in your free time?”; and question 23 is “Do you like your job?” (Harner, 1991).

38Although both instruments stress quality of life from the client’s perspective, the QOLQ prompts informants to evaluate clients’ activities and control over personal decisions, whereas the LSS uniformly elicits the clients’ self-report of their subjective satisfaction with their conditions of life.

3.4 ITEM SELECTION

39For the present analysis, only those QOLQ items that appeared conceptually to measure the same constructs as the LSS subscales were selected, so that a QOLQ subscale was constructed to parallel each of the LSS subscales. The items selected and descriptive statistics appear in Table 9.7. Final selection of items was based on an internal consistency reliability analysis. For each subscale the Cronbach’s alpha was calculated with each item included or excluded. Items were retained if they increased the alpha and deleted if they decreased the alpha. The alpha coefficients for the items in the final subscales are presented in Table 9.7. The reliability of selecting the items for the newly constructed QOLQ was also evaluated. The author and two colleagues independently selected items for each of the five subscales to establish agreement with the items that had been nominated for selection by the three as a team several months earlier. The 15 agreement statistics (three colleagues by five subscales) ranged from 72.5% to 97.5% (kappa = 0.290 to 0.925) with a median of 85% (median kappa = 0.490). Agreement with the items selected after item analysis ranged from 77.5% to 97.5% (median = 92.5%); kappa ranged from 0.05 to 0.925 (median = 0.545). For interpretation of the comparison of percentage agreement and kappa, see Johnson and Heal (1987).

40The LSS had 1.8% missing scores, 1.7% of which was due to clients having no jobs, and the QOLQ had 4.9% missing scores. In order to avoid losing data from cases with some missing scores, the mean of an item was substituted into the place of the missing data of any subject.

3.5 INTERVIEW PROCEDURES

41Each LSS interview took about 20 minutes. A question was repeated and/or rephrased until the interviewer judged that the client understood it. In accordance with the instructions in its manual, the QOLQ was administered to two informants who knew the client well, and their scores were averaged. One informant was always a direct service person (group home manager, behavioral technician, or staff person), and the other was a qualified mental retardation professional who had a direct service relationship with the client.

3.6 RESULTS

42The correlations between the LSS subscales and the newly constructed QOLQ subscales are presented in Table 9.8. It is disappointing that the only statistically significant diagonal coefficient, that for RECSAT, was negative in valence, indicating that informants’ judgments of recreational satisfaction were at odds with clients’ judgments. Three off-diagonal correlations—COMSAT with CTLQOL, CTLSAT with JOBQOL, and PALSAT with RECQOL—were statistically significant; two of the three were negative.

FIGURE 9.2. CORRELATIONS OF THE MLSS AND THE MQOLQ SUBSCALES WITH THE QOL DIMENSIONS OF SATISFACTION (I) AND COMPETENCE (II)

FIGURE 9.2. CORRELATIONS OF THE MLSS AND THE MQOLQ SUBSCALES WITH THE QOL DIMENSIONS OF SATISFACTION (I) AND COMPETENCE (II)

43A canonical correlation analysis was performed to evaluate the predictive accuracy of the LSS subscales from the QOLQ subscales. The results are presented in Table 9.9 and Figure 9.2, which show the canonical loadings for the first two canonical variates.

44The first canonical variate was associated with PALSAT and RECSAT from the LSS subscales as indicated by their high canonical loadings. On the other hand, RECQOL and JOBQOL from the QOLQ subscales were associated with the first canonical variate. The negative algebraic signs of these canonical loadings imply that PALSAT and RECSAT from the LSS set of subscales are associated negatively with RECQOL and JOBQOL from the QOLQ set of subscales. The second canonical variate was more interpretable than the first. All subscales of both sets of scales were positively correlated with this canonical variate, making it a good candidate for a general competence dimension. The redundancy coefficient of 0.192 in Table 9.9 indicates that only 19.2% of LSS set variability is accounted for by the first canonical variate from the QOLQ when LSS and QOLQ are treated as criterion and predictor sets respectively. This result follows because (a) the first canonical variate of the LSS set of subscales contains only 34.6% of the LSS variability and (b) this canonical variate shares only 55.7% (squared canonical correlation) of its variance with the first canonical variate from QOLQ set. If both the canonical variates are considered, a total of 27.5% of the LSS variability is accounted for by the QOLQ subscales. When the criterion and predictor sets are reversed, only 20.2% of variability of the QOLQ set is predicted by the two LSS canonical variates.

3.7 DISCUSSION

45The results of this second investigation are very puzzling. Why would two reasonably reliable scales that were designed to measure the same constructs have such subscale by subscale incongruence? Only 4 of the 25 correlations between the 5 LSS and 5 QOLQ subscales were statistically reliable, 3 in counterintuitive directions.

46We suggest that this pattern of results reflects (a) the real differences in the perceptions of clients and managers, (b) the differences in response biases in the two groups of respondents, and (c) imperfect substantive comparability of the two instruments. Looking first to response biases, we would expect greater acquiescence bias (saying “yes” regardless of the question asked) in the clients than in the informants, as it is correlated with intelligence (Heal & Sigelman, 1990, 1995). In this regard, RECSAT consisted entirely of reverse-scored items (e.g., “Would you like to do x more?”), and so an acquiescence bias would lower RECSAT scores and tend to produce a negative correlation with a more objective measure of recreation satisfaction, which RECQOL may have been.

47The negative mean score for RECSAT in Table 9.1 implies that many more respondents said “yes” to these reverse-scored “more” items (e.g., “Would you like to play more board games?” “Yes” = dissatisfaction) than said “no.” On the other hand, PALSAT had a balance of reverse-scored and normally scored items, which logically removes acquiescence bias. Because RECSAT and PALSAT from the LSS were positively correlated, it is likely that RECSAT scores were driven by true satisfaction as well as acquiescence bias. Nevertheless, the possibility exists that “yes” to these items really means “yes,” that they were wrongly reverse-scored, and that satisfaction with friends is, in fact, negatively correlated with the recreation activities because they tend to preclude each other. Turning to respondents’ choices, the informants were likely to let their understanding of current best practices influence their responses. For example, the moderate negative loading of CTLSAT and the moderate positive loading of CTLQOL on the first canonical variate could reflect the clients’ satisfaction with decision-making support (i.e., relief from the stress of making one’s own decisions), whereas the informants could feel obligated to embrace the principle of maximizing clients’ freedom of choice and accordingly to attribute higher quality of life to those clients who have the least assistance in making choices.

TABLE 9.8. CORRELATIONS BETWEEN LSS AND CONSTRUCTED QOLQ SUBSCALES (N = 46)

TABLE 9.8. CORRELATIONS BETWEEN LSS AND CONSTRUCTED QOLQ SUBSCALES (N = 46)

* p <.01.
a See Table 9.7 for a legend of abbreviations

TABLE 9.9. CANONICAL LOADINGS CORRESPONDING TO THE FIRST TWO CANONICAL VARIATES, CANONICAL CORRELATIONS, AND REDUNDANCY COEFFICIENTS BETWEEN THE LSS AND QOLQ SUBSCALES

TABLE 9.9. CANONICAL LOADINGS CORRESPONDING TO THE FIRST TWO CANONICAL VARIATES, CANONICAL CORRELATIONS, AND REDUNDANCY COEFFICIENTS BETWEEN THE LSS AND QOLQ SUBSCALES

a The product of the squared canonical correlation multiplied by the proportion of the total variance of the set of subscales explained by the canonical variate

48Other evidence indicated that the discrepancies between clients and informants on these two instruments should not be attributed to the inability of these informants to predict clients’ responses. As one of her validity evaluations, Harner (1991) selected two items from each LSS subscale and asked 75 informants to predict their clients’ responses. Informants predicted their clients’ LSS responses quite well. All 5 of the subscale correlations were positive, 4 significantly so, p <.01. Similarly Schalock et al. (1990) reported that the correlations between the 374 clients who provided their own responses on the QOLQ and the responses of their caretakers were high and statistically significant: Correlations were 0.66, 0.73, 0.81, 0.46, and 0.73, respectively, for their four subscales of Satisfaction, Social Belonging and Community Integration, Empowerment and Independence, and Competence and Productivity, and the total QOLQ score. Thus, the ability of caretakers to predict their clients’ responses on the LSS and the QOLQ contrasts with the mismatch between their quality-of-life judgments and corresponding clients’ LSS claims of lifestyle satisfaction. Given all these considerations, the first canonical variate appears to reflect different values of the clients and their caretakers that are exposed when caretakers are not explicitly instructed to guess their clients’ responses, and to greater acquiescence by clients than by informants.

49The second canonical variate appears to reflect a general competence factor, with informants’ attributions of self-determination and quality of employment dominating the variate and clients’ satisfaction with their community and self-determination contributing to their linear combination. The agreement on the competence dimension is reminiscent of Halpern, Irvin, and Landman (1979), who found extraordinarily good agreement between scores on an informant-completed employability scale and self-assessment by employment candidates with mental retardation.

4 GENERAL DISCUSSION

50Returning to our theme “Are Normalization and Social Role Valorization limited by competence?”, we must accept very tentative conclusions. Normalization and Social Role Valorization are presumably only two of the many constructs that comprise quality of life. This diminution is supported by the Hughes et al. (1995) systematic review of the quality-of-life literature, which was summarized in Table 9.1.

51If quality of life looms so large conceptually, can it be operationalized as well as Normalization and Social Role Valorization have been through PASS (Wolfensberger & Glenn, 1975) and PASSING (Wolfensberger & Thomas, 1983)? The present analyses have exposed two problems in this operationalization, both of which may yield to subsequent research. First, it appears that the many conceptually discrete features of quality of life will be difficult to disentangle from competence. This was seen in the first investigation from the high correlation of both the esteem and independence quality-of-life subscales with several competency predictors, accounting for 30% from the 37% total in the redundancy coefficients. Independence, of course, weighs heavily as a Normalization and SRV value as well as being a premier proxy for competence and self-determination in quality-of-life conceptualizations. The rather distasteful implication is that the most highly valued outcome for human service practice, independence, is best predicted by the entry skill of the client and almost trivially by human service practice. “Blaming the victim” for his or her constraints seems justified by this result. Elsewhere in this book, Flynn (chapter 14) interestingly suggests that PASS and PASSING results are correlated with attributes of individuals such as their level of adaptive behavior. The second canonical analysis supported the first in its detection of a competence dimension, but in this analysis it was swamped by another dimension that, while somewhat enigmatic, appeared to reflect in part areas in which the client and caretaker disagreed on what constituted the good life for the client.

52Is there Normalization, Social Role Valorization, or quality of life beyond competence? Scholars and practitioners in human services would like to think so. Only more refined assessment of these constructs and more concerted efforts to find valued and satisfying roles that are independent of competence will help us understand and separate one’s quality of life from one’s ability.

Bibliographie

REFERENCES

AAMR (American Association on Mental Retardation) (1992). Mental retardation: Definition, classification, and system of supports. Washington DC: Author.

Borthwick-Duffy, S. A. (1986). Quality of life of mentally retarded people: Development of a model. Unpublished doctoral dissertation, University of California, School of Education, Riverside, CA.

Dillon, W. R., & Goldstein, M. (1984). Multivariate analysis: Methods and applications. New York: Wiley.

Goode, D. (1994). Quality of life for persons with disabilities. Cambridge, MA: Brookline Books.

Halpern, A. S. (1993). Quality of life as a conceptual framework for evaluating transition outcomes. Exceptional Children, 59, 486-498.

Halpern, A. S., Irvin, L. K., & Landman, J. T. (1979). Alternative approaches to the measurement of adaptive behavior. American Journal of Mental Deficiency, 84, 304-310.

Harner, C. (1991). Assessing the satisfaction of adults with mental retardation living in the community. Unpublished doctoral dissertation, University of Illinois, Urbana, IL.

Harner, C. J., & Heal, L. W. (1993). The multifaceted lifestyle satisfaction scale (MISS). Psychometric properties of an interview schedule for assessing personal satisfaction of adults with limited intelligence. Research in Developmental Disabilities, 14, 221-236.

Harter, S. (1977). The effects of social reinforcement and task difficulty on the pleasure derived by normal and retarded children from cognitive challenge and mastery. Journal of Experimental Child Psychology, 24, 476-494.

Harter, S. (1978). Effectance motivation reconsidered. Human Development, 21, 34-64.

Heal, L. W., Borthwick-Duffy, S. A., & Saunders, R. R. (1996). Assessment of quality of life. In J. W. Jacobson & J. A. Mulick (Eds.), Manual of Diagnosis and Practice in Mental Retardation. Washington, DC: American Psychological Association.

Heal, L. W., & Chadsey-Rusch, J. (1985). The lifestyle satisfaction scale (LSS): Assessing individuals’ satisfaction with residence, community setting, and associated services. Applied Research in Mental Retardation, 6, 475-490.

Heal, L. W., & Rusch, F. R. (1994). Prediction of residential independence of special education high school students. Research in Developmental Disabilities, 15(3), 223-243.

Heal, L. W., & Rusch, F. R. (1995). Predicting employment for students who leave special education high school programs. Exceptional Children, 61, 472-487.

Heal, L. W., & Sigelman, C. K. (1990). Methodological issues in measuring the quality of life of individuals with mental retardation. In R. L. Schalock (Ed.), Quality of life: Perspectives and issues (pp. 161-176). Washington, DC: American Association on Mental Retardation.

Heal, L. W., & Sigelman, C. K. (1995). Response biases in interviews of individuals with limited mental ability. Journal of Intellectual Disability Research.

Hughes, C.; Huang, B., Kim, J., Eisenman, L. T., & Killian, D. J. (1995). Quality of life in applied research: Conceptual model and analysis of measures. American Journal of Mental Retardation.

Jamieson, J. (1993). Adults with mental handicap: Their quality of life. Vancouver, BC: British Columbia Ministry of Social Services.

Javitz, H. S., & Wagner, M. (1987). The national longitudinal transition study of special education students: Report on sample design and limitations, Wave 1. Menlo Park, CA: SRI International.

Johnson, L. J., & Heal, L. W. (1987). Inter-observer agreement: How large should kappa be? Capstone Journal of Education, 7, 51-73.

Nirje, B. (1969). The Normalization principle and its human management implications. In R. B. Kugel And W. Wolfensberger (Eds.), Changing patterns in residential services for the mentally retarded. Washington, DC: President’s Committee on Mental Retardation.

Nirje, B. (1992). The normalization principle—25 years later. In U. Lehtinen & R. Pirttimaa (Eds.), Arjessa Tapatuu: Comments on mental retardation and adult education (pp. 1-25). Jyväskylä, Finland: University of Jyväskylä, Institute for Educational Research.

Parent, W. (1993). Quality of life and consumer choice. In P. Wehman (Ed.), The ADA mandate for social change. Baltimore: Paul H. Brookes Publishers.

Parmenter, R.R. (1992). Quality of life for people with developmental disabilities. In N. W. Bray (Ed.), International review of research in mental retardation, 18, 247-287. New York: Academic Press.

Saunders, R. R., & Spradlin, J. E. (1991). A supported routines approach to active treatment for enhancing independence, competence, and self-worth. Behavioral Residential Treatment, 6, 11-37.

Schalock, R. L. (1990). Quality of life: Perspectives and issues. Washington, DC: American Association on Mental Retardation.

Schalock, R. L. (1996). Quality of life: Perspectives and issues, Vol. 1, Conceptualization and measurement (2nd ed.). Washington, DC: American Association on Mental Retardation.

Schalock, R. L., & Jensen, C. M. (1986). Assessing the goodness-of-fit between persons and their environments. Journal of the Association for Persons with Severe Handicaps, 11(2), 103-109.

Schalock, R. L., Keith, K. D., & Hoffman, K. (1990). 1990 Quality of Life Questionnaire Standardization Manual. Hastings, NE: Mid-Nebraska Mental Retardation Services.

Schalock, R. L.; Keith, K. D.; Hoffman, K.;& Karan, O. C. (1989). Quality of life: Its measurement and use in human services programs. Mental Retardation, 27(1), 25-31.

Valdes, K. A. (1989). The national longitudinal transition study of special education students. Menlo Park, CA: SRI International.

Wagner, M. (1989). Youth with disabilities during transition: An overview of descriptive findings from the national longitudinal transition study. Menlo Park, CA: SRI International.

Wagner, M., Newman, L., D’Amico, R., Jay, E. D., Butler-Walin, P., Marder, C., & Cox, R. (1991). Youth with disabilities: How are they doing? The first comprehensive report from the national longitudinal transition study of special education students. Menlo Park, CA: SRI International.

Wolfensberger, W. (1980). “The definition of normalization: Update, problems, disagreements and misunderstandings.” In R. J. Flynn & K. E. Nitsch (Eds.) Normalization, social integration, and community services. Baltimore: Pro-ED.

Wolfensberger, W. (1992). A brief introduction to Social Role Valorization as a high-order concept for structuring human services. (2nd ed., rev.). Syracuse, NY: Syracuse University, Training Institute for Human Service Planning, Leadership and Change Agentry.

Wolfensberger, W. (1994). “Let’s hang up’ Quality of Life’ as a hopeless term.” In D. Goode (Ed.), Quality of life for persons with disabilities. Cambridge, MA: Brookline Books.

Wolfensberger, W., & Glenn, L. (1975). Pass 3. Program analysis of service systems handbook. Toronto, ON: National Institute on Mental Retardation.

Wolfensberger, W., & Thomas, S. (1983). Passing (Program Analysis of Service Systems’ Implementation of Normalization Goals. Toronto, ON: National Institute on Mental Retardation.

Notes

1 The present report was supported in part by the Office of Special Education and Rehabilitative Services, U.S. Department of Education under a contract (300-85-0160) to the Transition Research Institute at the University of Illinois. I am grateful to Dr. Madhab Raj Khoju for his statistical consultation on the analyses reported herein, to Cathy J. Harner for the use of the data from her dissertation, and to Mary Wagner, Ron D’Amico, and Kathryn Valdes for their assistance in interpreting the data collected in the National Longitudinal Transition Study.

Table des illustrations

Titre TABLE 9.1. CLASSIFICATION OF QUALITY OF LIFEa
Légende a Adapted from Hughes et al. (1995). b The number of references in which Hughes et al. (1995) reported an application
URL http://books.openedition.org/uop/docannexe/image/2488/img-1.jpg
Fichier image/, 132k
Titre TABLE 9.2. COMPOSITION OF EACH DOMAIN OF THE QUALITY-OF-LIFE SCALE: SELF-ESTEEM, INDEPENDENCE, AND SECURITY AND SUPPORT
Légende a Item required reverse scoring to make low values reflect low quality of life and high values reflect high quality of life
URL http://books.openedition.org/uop/docannexe/image/2488/img-2.jpg
Fichier image/, 228k
Titre TABLE 9.3. PREDICTORS OF QUALITY OF LIFE GROUPED BY CHARACTERISTICS
URL http://books.openedition.org/uop/docannexe/image/2488/img-3.jpg
Fichier image/, 264k
Titre TABLE 9.4. RAW SCORE MEANS, STANDARD DEVIATIONS, AND TRANSFORMED SCORE ITEM-DOMAIN CORRELATIONS OF THE VARIABLES USED IN EACH DOMAIN OF QUALITY OF LIFE—ESTEEM, INDEPENDENCE, AND SECURITY AND SUPPORT
Légende a Item required reverse scoring to make low values reflect low quality of life and high values reflect high quality of life.b r is the correlation of the item with the remaining items from its dimension. Scores for this analysis were the cumulative percentages associated with each of the ordered categories for each variable.*p <.001.
URL http://books.openedition.org/uop/docannexe/image/2488/img-4.jpg
Fichier image/, 160k
Titre TABLE 9.5. MEANS AND STANDARD DEVIATIONS OF POTENTIAL PREDICTORS OF QUALITY OF LIFE AND THEIR CORRELATIONS WITH EACH OF THE THREE QUALITY-OF-LIFE DOMAIN VARIABLES
Légende a p <. 001 = ±0.13.*| r | ≥ 0.30.
URL http://books.openedition.org/uop/docannexe/image/2488/img-5.jpg
Fichier image/, 300k
Titre FIGURE 9.1. PROPORTION OF VARIANCE (R2 CHANGE) ATTRIBUTABLE TO SUCCESSIVE BLOCKS OF PREDICTOR VARIABLES FOR EACH OF THE THREE QUALITY-OF-LIFE DOMAINS
Légende Predictor block labels are abbreviations of the eight successive numbered sections from Table 9.3
URL http://books.openedition.org/uop/docannexe/image/2488/img-6.jpg
Fichier image/, 56k
Titre TABLE 9.6. CANONICAL LOADINGS, ASSOCIATED SQUARED MULTIPLE CORRELATIONS, CANONICAL CORRELA TIONS, AND REDUNDANCY COEFFICIENTS FOR QUALITY OF LIFE AND ALL 28 PREDICTORS Canonical Variates (Dimensions)
Légende *Indicates canonical loading >0.3.*Indicates canonical loading > 0.5
URL http://books.openedition.org/uop/docannexe/image/2488/img-7.jpg
Fichier image/, 452k
Titre TABLE 9.7. DESCRIPTIVE STATISTICS FOR THE LIFESTYLE SATISFACTION SCALE (LSS) AND CONSTRUCTED QUALITY-OF-LIFE QUESTIONNAIRE (QOLQ) SUBSCALES (N = 46)
Légende a The QOLQ item numbers are taken from Schalock, Keith, & Hoffman, 1990; the LSS item numbers are taken from the 76-item trial LSS (Harner, 1991). Both lists are available from the author. The value of each QOLQ item could range from 1 (low) to 3 (high); the value of each LSS item could range from-2 (very dissatisfied) to +2 (very satisfied). b Cronbach’s alpha coefficient of internal consistency. LSS coefficients are from a crossvalidation sample. c Abbreviations: QOL = quality of life assessed by the newly constructed QOLQ subscales; SAT = satisfaction assessed by the LSS subscales; JOB = job; COM = community, home, and neighborhood; PAL = friends and most social activities; REC = recreation and most leisure activities; CTL = control of one’s choices, self-determination
URL http://books.openedition.org/uop/docannexe/image/2488/img-8.jpg
Fichier image/, 144k
Titre FIGURE 9.2. CORRELATIONS OF THE MLSS AND THE MQOLQ SUBSCALES WITH THE QOL DIMENSIONS OF SATISFACTION (I) AND COMPETENCE (II)
URL http://books.openedition.org/uop/docannexe/image/2488/img-9.jpg
Fichier image/, 68k
Titre TABLE 9.8. CORRELATIONS BETWEEN LSS AND CONSTRUCTED QOLQ SUBSCALES (N = 46)
Légende * p <.01.a See Table 9.7 for a legend of abbreviations
URL http://books.openedition.org/uop/docannexe/image/2488/img-10.jpg
Fichier image/, 72k
Titre TABLE 9.9. CANONICAL LOADINGS CORRESPONDING TO THE FIRST TWO CANONICAL VARIATES, CANONICAL CORRELATIONS, AND REDUNDANCY COEFFICIENTS BETWEEN THE LSS AND QOLQ SUBSCALES
Légende a The product of the squared canonical correlation multiplied by the proportion of the total variance of the set of subscales explained by the canonical variate
URL http://books.openedition.org/uop/docannexe/image/2488/img-11.jpg
Fichier image/, 159k

Auteur

Ph.D., Professor Emeritus of Special Education, Psychology, and Social Work, University of Illinois, Urbana-Champaign, IL, USA

© Les Presses de l’Université d’Ottawa | University of Ottawa Press, 1999

Conditions d’utilisation : http://www.openedition.org/6540