|
|
Chapter 1: IntroductionThe concept of independent living has slowly gained strength since the late 1960s. The most important factor that differentiates the independent living model from previous approaches is the rejection of the rehabilitation model, which was common at least since the end of World War II. The rehabilitation model conceives of people with disabilities as patients with some sort of medical condition or impairment, and the goal is to perform treatments or interventions with the limited vision of creating maximum self-care or gainful employment. The independent living model does not eschew the provision of adequate professional and medical services to meet the needs of people with disabilities in every community. It does emphasize the empowerment of people with disabilities to take control over their own lives to the maximum extent possible. Instead of conceiving of people with disabilities as passive patients who need treatments or help, the independent living model views them as active consumers who control the services they use. Instead of defining the problem as a medical condition or an impairment, the problem is defined as the dependence on professional and medical services and personnel that stems from the condition itself. The problem is not in the individual, it is in the environmental barriers to independent living, and the solution is to remove barriers through system change, advocacy, and self-help in order to create independence through control over available options for everyday life. The independent living movement in America stresses the empowerment of people with disabilities. The mission of the Idaho State Independent Living Council is to "promote a philosophy of independent living, including a philosophy of consumer control, peer support, self-help, self determination, equal access, and individual and system advocacy, in order to maximize opportunities for individuals with disabilities, and the integration and full inclusion of individuals with disabilities into the mainstream of society" (emphasis added). The goal is to maximize accessible opportunities and to incorporate people with disabilities into all walks of life. The means to achieve this goal concentrate on empowering people with disabilities. The road to empowerment includes changes in the individual and changes in the society. Changes in the individual include recognizing and using the power of consumerism as well as peer support and self-help groups to raise awareness that self-determination is within reach. Changes in society also are required, including further demedicalization, deinstitutionalization, strengthened public policies for equal access, and further use of individual and system advocacy. In order to realize its goals, the Idaho State Independent Living Council must obtain detailed information on the prevalence, needs, and concerns of Idahoans with disabilities. This information is crucial for the Council's efforts to develop an effective three-year strategic plan. The key question confronting the Council is: What are the most important things that need to be done to "promote a philosophy of independent living, including a philosophy of consumer control, peer support, self-help, self- determination, equal access, and individual and system advocacy, in order to maximize opportunities for individuals with disabilities, and the integration and full inclusion of individuals with disabilities into the mainstream of society?" At the beginning of the project, the Council was already aware of many of the broad unmet needs throughout the state, such as the fact that many people with disabilities do not have adequate access to adequate transportation services. Through the survey, the Council sought to determine at a detailed level what other kinds of unmet needs exist and for whom are they the most important. On September 28, 1994, the Idaho State Independent Living Council awarded a contract, funded jointly by the federal government (90 percent) and the state government (10 percent), to the Survey Research Center at Boise State University to conduct a statewide needs assessment survey of Idahoans with disabilities. This monograph presents the results of the Center's work. OBJECTIVES The Idaho State Independent Living Council sought to conduct a statewide needs assessment survey of Idahoans with disabilities in order to identify the greatest unmet needs for three-year strategic planning. In order to accomplish this goal the Council developed several priorities for the survey. Most importantly, the Council desired a survey based on probability sampling that would be statistically valid. The Council also wanted to reach those Idahoans with disabilities that were not currently receiving services in order to assess unmet need. In addition, the Council strongly desired that the survey provide new information that was not already known. Further, the Council sought to ensure that the needs and concerns of people with communication-related disabilities were well represented among the respondents. Finally, the Council sought to ensure that minority populations were represented among the respondents. The Council was interested primarily in the detailed, unknown needs of Idahoans with disabilities. However, the Council also recognized the need for accurate information about the magnitude of problems that are already well known at a general level, such as the lack of adequate transportation. These considerations led to the following specific survey objectives:
The methodology developed by the Survey Research Center at Boise State University to meet these objectives is presented in the following sections, including sampling plan, sample size considerations, questionnaire development, and data collection and processing. METHODOLOGY Sampling Plan The highest priority objective for the Council was to base the survey on a statewide, representative sample of Idahoans with disabilities. Several options for developing a probability sample of Idahoans with disabilities were considered. The options included (1) drawing a sample from lists of persons maintained by service providers or Centers for Independent Living, (2) drawing a sample from lists of persons who indicated on previous surveys that they or someone in their household had a "physical disability," (3) drawing a probability sample from official listings of people who report on their income tax returns being disabled or having a persons with a disability in the household, (4) drawing a probability sample from state listings of people receiving social security disability income, (5) the use of random digit dialing telephone survey methods. This section presents the advantages and disadvantages of each of these options and the reasons for finally selecting the random digit dialing survey method. It is important to remember that there is no single, perfect option. Each option has some strengths and some weaknesses, including time and cost. One possibility is always tempting: to add the options together so as to take advantage of all their strengths. However, it is almost never the case that a survey sponsor has sufficient time or funds for this. The objective is to maximize the information that can be obtained with the available resources. Client Lists. The first and most obvious method of identifying people with disabilities is to use client lists maintained by service providers or Centers for Independent Living. However, these lists include only those people who receive or who have received services. The Council fully realized that, in a needs assessment survey, it is essential to identify and estimate the prevalence of those people who have not received services in order to be able to develop strategic plans based on the types and magnitude of unmet needs. To the extent that minorities are less likely to receive services, this method would result in under coverage of minorities in the sampling frame. The use of client lists as a sampling frame was not considered seriously as a primary sampling frame, although the use of service provider lists was considered as a supplement to other methods. However, the strong advantage of using client lists is that they include only people with disabilities. All other methods considered make it increasingly difficult, time-consuming, and costly to identify people with disabilities. Because it is the easiest and the least costly option, it serves as a backdrop against which to compare the other alternatives. That is, any other more costly option that does not significantly improve on this option would not be considered. Previous Respondent Lists. A second option involved developing or purchasing a list of people with disabilities. This technique involves selecting households from past surveys in which respondents indicated that they or someone in their household had a "physical disability." The Survey Research Center has conducted previous surveys in which respondents were asked whether they or anyone in their household had a "physical disability." A national firm called Survey Sampling Inc. has also asked this question. This method would provide a way of identifying those who do not currently receive services. It is also relatively inexpensive and would not require an extensive amount of time to prepare. Previous respondent lists, however, have two serious shortcomings. The first is the potential of the method to violate the strong pledge of confidentiality that is incumbent upon all public opinion researchers. Using respondents from previous surveys that possess certain attributes in new surveys is considered suspect by many public opinion researchers, even when the respondents have given permission to be recontacted. The reasons are: (1) respondents cannot know the reasons for any recontact when they ostensibly give their permission, and (2) when purchasing lists from other organizations, researchers cannot be certain that the original respondents actually ever gave their permission. At the Survey Research Center, once a survey is completed the telephone numbers and any other identifier of survey respondents that would make recontact possible are permanently removed from the data to ensure respondent confidentiality. However, such a list can be purchased from a company that mailed questionnaires to 10 million U.S. households in 1993, all containing a question that asked whether any person with a "physical disability" resided in the household. The list contains 110,000 households (1.1 percent of the sample) that reportedly contained a resident with a physical disability. This percentage is consistent with the Center's finding that between one and two percent of survey respondents in the Boise area report that a person with a physical disability resides in the household. The problem is that the survey question, by using the word "disability," inadvertently invokes a social stigma such that the vast majority of people with significant health related conditions that would, for example, qualify them as having a disability under the Americans with Disabilities Act or other relevant legislation do not answer the question affirmatively. Thus the second problem with previous respondent lists is that they have a very large and unknown amount of under coverage. Minorities would tend to be under-represented at least to the same extent as others with disabilities. The number of Idaho residents among the 110,000 who self-identified as having a "physical disability" is likely to be quite small, and the number of Idaho minorities among the respondents would be even smaller. Official Records. Two methods were considered that rely on the possibility of drawing a probability sample from lists of persons with disabilities taken from official government records. The first was to draw a probability sample from official listings of those who report on their income tax returns being disabled or having a person with a disability in the household. The second was to draw probability samples from lists of persons receiving social security disability income. These lists would provide a more complete listing of people with disabilities who have not received Center for Independent Living services in the past than the previous options. At the outset, these two methods appeared to hold the most promise. However, the methods do have significant limitations. Although the Center maintains strong relations with state government agencies that have fostered successful cooperation, such a sampling procedure would require a memorandum of understanding concerning the safeguarding of confidentiality between the Council and the agency responsible for maintaining the records. In a recent needs assessment study of the over-representation of minority youth in the criminal justice system, the Center worked extensively with the courts and other state entities to develop such a memorandum. The trouble was that it took a full three months to secure the necessary assurances of confidentiality, which raised the cost of the project substantially. Over the course of studying the available alternatives for this research, it became apparent that many people who qualify as having a disability under the Americans with Disabilities Act or other relevant legislation may not qualify for Social Security Disability Income and may not claim themselves or members of their family as having a disability on their income tax returns. Although less of a problem here than in the previously discussed methods, under coverage is still a potentially serious threat. In addition, such listings, especially tax returns, would in all likelihood under-represent minorities more than others. Moreover, such listings are likely not to include home telephone numbers. This could limit the approach to a mail survey, which would greatly increase the time required to complete the survey. Finally, obtaining lists from official records frequently requires a great deal of effort to transform the computer files from what the Center receives into something that is useful as a survey sampling frame, which could greatly increase the time and expense of the survey. Moreover, technical programming work is not something that the Council would derive any benefit from since it is related to the mechanics of making the survey work and not to the results. Random Digit Dialing. Random digit dialing telephone survey methods did not at first appear to be a candidate for this project. Despite the well-known advantages of this methodology, it is not widely applicable when the population of interest is a small fraction of the general population. Among the advantages of the random digit dialing telephone survey methodology are that is quite expeditious, it is relatively affordable under optimum conditions, it includes the total population with the exception of the increasingly small proportion of households without telephones, and it is statistically precise. Although minorities are less likely to have a telephone than others, the degree of under coverage of minorities is less here than in any other method considered practical for a statewide survey with high statistical validity. The reason that random digit dialing methods are not generally applicable when the population of interest is small is time and cost. In telephone surveys conducted by the Survey Research Center at Boise State University (an institution with high credibility and thus a relatively low refusal rate) in which the adult in the household with the most recent birthday (the Center's usual selection criteria) qualifies as a respondent, approximately 10 telephone calls must be made in order to complete one interview. If the population to be surveyed represented only one percent of the total population, the number of calls required to complete one interview would increase to the point where the cost, if not the time, would become prohibitive for most survey sponsors. As mentioned earlier, surveys have routinely found that approximately one percent of respondents indicate that someone in the household has a "physical disability". In the course of the conversations between the Center and the Council, however, it became clear that a much larger percentage of the total population qualified as having a disability under the Americans with Disabilities Act and other relevant legislation. It became apparent that survey respondents tend not to self-identify as having a disability due to the social stigma attached to the word "disability." In the course of the background research on this project, Center staff catalogued numerous examples in which people said, for example, my father only has one leg, but he does not have a "disability". With this discovery, the use of random digit dialing telephone survey methods became a real possibility. Two additional problems needed to be solved. The first was to determine just what is the proportion of the general population that qualifies as having a disability under the Americans with Disabilities Act and other relevant legislation. The second was what could be done to screen respondents to determine whether they qualified as having a disability under relevant legislation without using the term disability or otherwise stigmatizing respondents in a way that would make it hard for those with qualifying conditions to answer in the affirmative. Center staff accumulated estimates of the proportion of the general population that might qualify as having a disability under relevant legislation from several sources, including the 1990 U.S. Census. These estimates ranged from a low of 12 percent to a high of nearly one-third. The Center deemed this range high enough to consider seriously the use of random digit dialing methods telephone survey methods. The next question was how to develop screening questions to identify people who might qualify as having a disability under relevant legislation. At its December meeting, the Council provided the Center with a set of federal criteria that identify health-related conditions that are necessary (if not sufficient) conditions for qualifying people as having a disability under relevant legislation. Center staff were able to transform the 17 technical criteria provided by the Council into language suitable for telephone interviews and to make them short enough to minimize the number of potential respondents who would lose interest in the survey and refuse to complete them. It is doubtful that these questions identify only people who currently qualify for public services under existing law, such as the Social Security Act, but they most certainly identify people who report having health-related conditions covered by the Americans with Disabilities Act. The screening questions (along with the entire survey) were pretested in February 1995. The results showed that the method was successful and the Center proceeded to complete the survey using this method. Sample Size Considerations Initially, the Council desired to complete 700 interviews with Idahoans with disabilities, approximately 100 in each of the seven Idaho regions defined by the Idaho Division of Vocational Rehabilitation. However, 700 is more than are needed for a statewide survey of this type, and 100 is too few to provide reasonably accurate estimates for each region. With five point rating scale measures such as were used predominantly in this survey, 600 completed interviews are needed to produce statewide confidence intervals of approximately plus or minus three to four percent for the mean, with the most extreme being plus or minus four to five percent. Approximately 500 interviews would be necessary to produce statewide confidence intervals of approximately plus or minus four to five percent, with the most extreme being approximately plus or minus five to six percent. Of course, the confidence intervals will be larger for subsets of the state population, such as regions or specific types of people, and for estimates of the effects of demographic factors on the reported level of need. For example, in order to achieve confidence intervals of plus or minus four to five percent for each region, 500 completed interviews would be required for each region for a total of 3,500 completed interviews. In order to achieve the same statewide confidence interval of plus or minus four to five percent for a single minority group, nearly another 500 interviews would have to be completed. In view of the needs and resources of the Council and the resources already devoted to developing the screening methodology and the questionnaire, the Council agreed that 500 completed interviews statewide would be sufficient. In all instances, the results are statistically valid: The only limiting factor is that the wider the confidence intervals the more difficult it is to say with confidence what the true result for a given subset of the population is likely to be. The possibility of oversampling minorities in order to have a sufficiently large sample to make precise statements about specific minorities was considered. However, since the proportions of minorities are so small in Idaho and since the proportion of the general public with qualifying conditions was already small, this option was rejected as too costly. Nevertheless, steps were taken to ensure that those minorities reached by telephone could be interviewed. These are discussed later in the section on data collection and processing. The Center purchases random digit dialing samples from the firm Survey Sampling Inc. (SSI) which are well-documented, credible, and have successfully withstood many court challenges. The probability samples are stratified proportional to the number of active blocks (multiplied by 100) in each telephone exchange in each county of the test area. A block is a group of 100 telephone numbers in a single exchange. For example 343-1100 to 343-1199 is one block. There are 100 blocks or 10,000 numbers per exchange. In order for a block to be active, at least one number in the block must be listed in the telephone directory. (The assessment of active blocks is updated every six months by SSI.) For example, in the Boise exchange "343" all 100 blocks are active and the sampling frame includes 10,000 numbers for that exchange. In the town of Deary, Idaho, the exchange "877" has only three active blocks, giving a total of 300 numbers included in the sampling frame for that exchange. In order to ensure that every Idahoan had a chance to be included in the survey, the Center included telephone exchanges that border on Idaho and have some Idahoans even though the majority of the numbers are in adjacent states. These included exchanges in Whitman and Spokane Counties in Washington. Accordingly, there were a total of 782,700 telephone numbers from which the sample was drawn. Once the sample is drawn, SSI removes numbers that they have identified through their screening procedures as business or nonworking numbers. However, not all of these numbers can be identified and removed: Some must be identified by the Center over the course of the survey. In addition, the Center cannot contact all of the valid residential numbers in the sample during the course of the survey. Some will remain unanswered or busy at the end of the survey. Moreover, between seven and nine out of every ten valid numbers contacted would not report a person who qualified for the survey, and an additional one out of ten valid numbers reached would refuse to conduct the interview. The number of calls that result in no answers, busy signals, businesses, refusals, unqualified households, and so forth before one interview can be completed is termed the "hit rate. " For this survey, SSI determined that the hit rate could be as high as twenty-six to one. Accordingly, the Center purchased a sample of 13,000 telephone numbers. As just described, these 13,000 numbers were drawn from each county based on the proportion of active telephone blocks (multiplied by 100) in each telephone exchange. For example, Ada County contains over 20 percent of the total active blocks (multiplied by 100), whereas Camas County contains 0.16 percent of the active blocks (multiplied by 100). Accordingly, 20 percent of the 13,000 numbers or 2,658 numbers were drawn from active blocks in Ada County exchanges and 0.16 percent of the 13,000 or 21 numbers were drawn from the active blocks in Camas County exchanges. The numbers are drawn independently from each exchange in each county using systematic sampling procedures. For example, in Camas County there were approximately 13 active blocks for a total of 1,300 numbers in the sampling frame. The 1,300 numbers were divided by the 21 necessary for the sample to determine a sampling interval of approximately 61. After a random start every sixty-first number was selected from the 1,300 numerically order telephone numbers. The 13,000 numbers were further divided into 130 replicates of 100 numbers each. Each replicate is an independent sample with the same specifications as just discussed. This procedure is adopted so that the Center can better control the response rate by adding new replicates as the current replicates are exhausted by the interviewers. After SSI screened the sample for business and non-working numbers, the Center actually received a total of 11,015 numbers, divided into 130 replicates ranging from 76 to 92 numbers in each replicate. The sampled telephone numbers were spread across the 44 Idaho counties in rough proportion to the number of households in each county. This procedure allows the Center to aggregate the results into regions of the state along county lines and ensure that statistics for each region can be calculated. In the case of the Idaho Division of Vocational Rehabilitation, however, two of the seven regions do not follow county lines. Region Seven includes a portion of Owyhee County and a portion of Adams County (with the remainder of those counties included in Region Three) as well as Washington, Payette, Gem, and Canyon Counties. Bingham County is split between Regions Five and Six. The Center included all of Region Seven in Region Three, which also includes Ada, Boise, and Elmore Counties. Bingham County was included in Region Five. This approach represents one of the standard ways of representing regions in Idaho that conform to county boundaries. A final issue related to sample size involved the ability of a general population survey of this size to reach minorities and people with relatively rare types of disabilities. In a general population survey such as this, respondents with the most prevalent characteristics will be most frequently represented. Although this is appropriate for a statewide needs assessment survey, it does mean that very few of some types of Idahoans will be reached. In order to collect additional information on relatively rare respondents, the four Centers for Independent Living together completed approximately 100 additional interviews. These interviews do not represent a statistically valid sample. Their purpose was to provide additional qualitative information about Idahoans with relatively rare characteristics. Questionnaire Development The Center supplemented its own detailed knowledge of the methods of needs assessment by working closely with the Council to determine what the members wanted to find out about Idahoans with disabilities. Council members developed the questions in order to ensure that the items addressed the detailed information about issues that were not well known as well as those that were well known at the general level. Center staff transformed the draft questions provided by the Council into professional telephone interview questions. Identifying Idahoans with Disabilities. Center staff were most concerned with the screening aspect of the survey--the ability of the survey to identify Idahoans who have characteristics relevant for qualifying people as having a disability under existing laws. One thing that, to the Center's knowledge, has not been done before is to develop a practical set of screening questions designed to produce an overall count of the number of people that meet one or more of the relevant criteria. Since the proportion of the population that has a "physical disability" reportedly has varied between one and 33 percent of the total general population, the Center was concerned to help explain why the estimates vary so greatly. The Center did not use the standard survey question: "Does anyone in your household have a physical disability" because the biased results from this type of question are known already. The Center used the criteria provided by the Council to develop a detailed and neutral set of 17 screening questions. In addition, the Center used the questions from the long form of the U.S. Census for determining the number of people with disabilities. The Census questions ask whether the respondent has experienced one of four possible consequences of a health-related condition for six months or more: limiting the ability to work, preventing holding a job, interfering with leaving home, and interfering with taking care of personal needs. The Center asked whether respondents possessed each of these conditions at all, whether it had lasted six months or more, and whether it had lasted 12 months or more. Thus the center included 12 Census-related questions. Open-Ended Questions. One question that has worked well for the Center on several different types of surveys is to ask the respondent an open-ended question about the most important problem they face. In this project, the Center asked about the most important problem facing Idahoans with disabilities and about the most important problem facing the respondent as a person with a disability. Independent Living and Self-Efficacy. One of the Council's objectives was to find a way to measure the extent to which Idahoans with disabilities live independently, the extent to which they feel independent, or the extent to which they share the independent living philosophy. This presented a unique measurement challenge because the vast majority of the potential respondents probably would not be aware of the terms independent living, independent living movement, or independent living philosophy, yet many could be expected to live independently nonetheless. In the sociological literature a set of questions has been designed to assess the extent to which people believe themselves to be self-efficacious at handling the numerous problems that confront people in life. Closely related to self-concept, the idea of self-efficacy includes the extent to which people take control over the environmental constraints or barriers they confront. The concept has been used, for example, to try to determine the kinds of people who become political activists. According to the theory, people with high self-efficacy are more likely to take risks and to challenge the status quo; accordingly they should be more likely to become political activists. In this sense, the idea of self-efficacy is closely related to the extent to which people feel independent or the extent to which they share the independent living philosophy, even though they may not be familiar with the terminology of the independent living movement. The Center included Gurin's "personal control" subscale of the full self-efficacy scale in order to help supplement other measures of independence or independent living. In addition, the Center and the Council developed a question that presented a definition of independent living to the respondents and then asked the extent to which the definition characterized their life. Data Collection and Processing Once the final, pretested draft of the questionnaire was approved by the Council at its meeting in February 1995, the Center purchased the full sample from Survey Sampling Inc. and began the survey. Computer-Assisted Telephone Interviewing System. The questionnaire items were entered into the computer assisted, telephone interviewing software, which transforms the hard-copy questionnaire into data entry screens on a computer. The sample telephone numbers are fed through the computer network to the interviewer work stations. The disposition of every telephone call made as well as the data on all completed interviews are returned through the network to a central computer. Techniques for Interviewing Respondents with Communication-Related Disabilities. In the last several years, the Center has developed substantial expertise in conducting interviews with people who have communication-related disabilities. For two years the Center implemented demonstration programs that provided innovative services for people involved in custody or visitation disputes, some of which involved parents or children with disabilities. The Center is now in the third year of a project involving juveniles and families at risk, in a number of which the parents, children, or both had disabilities. In the summer of 1994, the Center completed a survey for the Ada PIanning Association as part of a needs assessment for the current riders of the Boise Urban Stages bus system, many of whom are people with disabilities. During the course of this survey, which included both a mail and a telephone component, Center staff and interviewers gained extensive experience interviewing people with cognitive and communication-related disabilities. In most instances, barriers involving communication-related disabilities were overcome by having another family member present to relay the survey questions to the respondent and then to relay the answers to our trained interviewers. Some people with disabilities who received the questionnaire in the mail preferred to call our offices and complete the interview by telephone. These experiences made the Survey Research Center at Boise State University eager to undertake a project in which Idahoans with disabilities were the primary population of interest. As just discussed, the Center's research experience shows that the most effective method for reducing environmental barriers involving communication-related disabilities is to have another family member act as a translator. One thing the Center has discovered is that in many instances where the potential respondent's disability makes it difficult to communicate over the phone, the surveyor must gain trust and rapport with someone else in the household who, in effect, acts as a gatekeeper. When the actual respondent is unable to communicate over the phone, another household member also can act as a proxy respondent who provides their own answers to the question on behalf of the actual respondent. The Council was divided on the extent to which other family members would be able to act effectively either as translators or as proxy respondents to provide information that would accurately represent the needs and concerns of the actual respondent. In some instances, caretakers may be an obstruction to identifying needs and concerns, but in others they are crucial advocates. However, the Council was not divided on the question of whether to include the responses of other household members versus getting no information at all about the actual respondent. When the actual respondent was unable to speak on the phone, the Center included the responses of other household members who acted as translators or proxy respondents. The Center recorded the relationship of the other household member to the actual respondent, as well as whether the other household member was the legal guardian of the actual respondent. In order to ensure that all respondents could use the most convenient method of response for their situation, the respondents were invited to request a mail version of the questionnaire or to call the Center directly at a time when another household member was available to act as a translator. Minorities. To ensure that potential respondents who speak only Spanish were included, the Center recorded instances where the person answering the phone spoke only Spanish. The Center's Spanish speaking interviewers recontacted these households to determine if any of these potential respondents qualified under one or more of the screening questions and to conduct the interview with those that did.
|
|
|
