THE DEVELOPMENT AND VALIDATION OF THE FAMILY EXPERIENCES WITH AUTISM SPECTRUM DISORDERS (FEASD) SCALE

Autism Spectrum Disorders (ASD) impacts 1 in 50 children in the United States (Centers for Disease Control and Prevention [CDC], 2013). This striking increase in the number of children with ASD affects families in a variety of ways. The purpose of this study was to develop and validate a scale that measured the experiences of families of children with ASD in schools, health care settings, and in their families/communities. Family-centered care (American Academy of Pediatrics, 2003) principles were used to create the items in the scale. In addition to determining the psychometric properties of the scale, the second purpose of this study was to assess families’ experiences with professionals in the health care, education, and community/familial settings with respect to the child’s race/ethnicity, family household income, level of educational attainment of caregiver, race/ethnicity of doctor who diagnosed the child with ASD, and kind of doctor that diagnosed the child with ASD. Four hundred sixty-six respondents completed the online scale and personal background questions over a period of four months. Principal components analysis was conducted on the “Family Experiences with Autism Spectrum Disorders” (FEASD) Scale, which indicated 3 factors were present. The three factors, “Family Support,” “School Quality,” and “Health Care Quality,” account for 48.58% of the variance and had an overall coefficient alpha level of .92. Each factor was found to have coefficient alpha levels of .96, .89 and .70, all acceptable internal consistency values for a new scale (Stevens, 2002). The multivariate analyses indicated two significant differences. First, families of children with ASD who reported a household income of $100,000 $124,999 had more positive experiences on the “Family Support” Scale, F(9, 456) = 2.97, p = .002. Second, families who reported that a pediatrician diagnosed their child with ASD had more positive experiences on the “Total FEASD” Scale F(3, 410) = 4.36, p = .005, compared to those families who had a psychologist make the diagnosis. Limitations of the present study and future directions of research are included.

. This is an increase in over 78% of ASD cases reported compared to ten years ago (CDC, 2013).
The Individuals with Disabilities Education Improvement Act (IDEIA) (2004) defines Autism as "a developmental disability significantly affecting verbal and nonverbal communication and social interaction, usually evident before age 3 that adversely affects a child's educational performance. Other characteristics often associated with ASD are engagement in repetitive activities and stereotyped movements, resistance to environmental change or change in daily routines, and unusual responses to sensory experiences. The term does not apply if a child's educational performance is adversely affected because the child has an emotional disturbance" [34 C.F.R. 300.8(c) (1)].
The Autism can impact a person in a myriad of ways (CDC, 2013). Some people with ASD have intellectual disabilities, while others have superior intelligence quotients. In addition, there may be a person with ASD who is unable to communicate verbally, often referred to as being "non-verbal," while another is a strong verbal communicator, who has no trouble speaking. While people with ASD have common challenges such as difficulty with social interaction, understanding non-verbal cues, and comprehending abstract language, there are differences. The severity of the symptoms, how they start, and the particular nature of the symptoms vary greatly.
Each individual with ASD has strengths, challenges, hopes, and dreams. The physicians, educators, and other professionals who support people with ASD must treat each as an individual (CDC, 2013).
Due to the varied impact that ASD can have on the neurological development and functioning of a person, researchers are urged to examine the various settings that impact the success of a person diagnosed with ASD (Bellin, Osteen, Heffernan, Levey, & Snyder-Vogel, 2011). For persons with significant disabilities, especially children, family members are often relied upon for sharing their perceptions of the quality of interaction with the child's doctors, teachers, extended family members, and people in the community (Denboba, McPherson, Kenney, Strickland, & Newacheck, 2006;Strickland, McPherson, Weissman, van Dyck, Huang, & Newacheck, 2004).

Selection of the Problem
The literature presented above supports the idea that research be conducted to determine the experiences of families who have children diagnosed with ASD to better understand any perceptions of differences that may exist in their quality of support in various settings. It is well documented that racial/ethnic minorities are underrepresented in receiving a diagnosis of ASD (Lord & Bishop, 2010;Mandell, Ittenbach, Levy, & Pinto-Martin, 2007;Mandell, Listerud, Levy, & Pinto-Martin, 2002;Morrier, Hess, & Heflin, 2008) which impacts the services that are provided to these children and their overall ability in making satisfactory social, emotional, and academic progress (Birkin, Anderson, Seymour, & Moore, 2008;Mandell, Morales, Xie, Polsky, Stahmer, & Marcus, 2010;National Research Council, 2001;Zoints, Zoints, Harrison, & Bellinger, 2003). The studies examining the satisfaction of caregivers of children with ASD in regard to the education, health care, and community/familial support provided are limited (Mandell et al., 2007;Mandell et al., 2002;Morrier, et al., 2008). Addressing the challenges that families face in each of the three areas (education, health care, and community/familial support) is key to understanding the steps necessary to improve access and quality of interventions provided to all families.

Statement of Purpose
The first purpose of this study was to develop a scale that measured the experiences of families with children with ASD. Family-centered care (American Academy of Pediatrics, 2003) was particularly important in the current study, as it provided a theoretical framework for the creation of the items in the scale used to measure the experiences of families of children with ASD.
Once the scale was developed with input from content experts and family members, the researcher determined the factor structure of the scale and the instrument's internal consistency. In addition to determining the psychometric properties of the scale, the second purpose of this study was to use the validated scale to assess families' experiences with professionals in the health care, education, and community/familial settings with respect to the child's race/ethnicity, family household income, level of educational attainment of caregiver, race or ethnicity of doctor who diagnosed the child with ASD, and kind of doctor that diagnosed the child with ASD.

Disparities in ASD
Like many families of children with disabilities, families of children with ASD navigate complex education and health care systems. While these systems pose difficulties for all families, caregivers of children with ASD have a particularly difficult journey, starting with receiving a diagnosis of ASD.
With the increase in children being diagnosed with ASD, research has begun to look at the epidemiological, social, racial, and environmental factors associated with the diagnosis. Fombonne (2003) and Yeargin-Allsopp, Rice, Karapurkar, Doernberg, Boyle, and Murphy (2003) have established that there is no known ethnic or racial difference in the epidemiology of autism. While biologically no difference exists, national data trends in ASD research suggest an under representation of racial/ethnic minorities diagnosed with ASD (Mandell et al., 2002). What factors account for these disparities? Mandell et al. (2002) found that African American children were less likely to be given a diagnosis of ASD on the first visit to their health care provider. The researchers revealed that African American children were referred to specialty care later and required more specialty care visits to receive an ASD diagnosis than White children. In another study, Mandell et al. (2007) found that African American children were 2.6 times less likely to be diagnosed with ASD on their first visit to a physician compared to White children. Palmer, Walker, Mandell, Bayles, and Miller (2010) studied 1184 Texas school districts to identify the incidence of ASD among Hispanic students. They found that Whites were two to three times more likely to have an ASD diagnosis compared to Hispanics. They found that socio-economic factors could not explain the under representation of Hispanic children diagnosed with ASD. Similarly, Begeer, El Bouk, Boussaid, Terwogt, and Koot (2008) found that pediatricians more often diagnosed Whites with ASD compared to their racial/ethnic minority counterparts.
Research has consistently shown that most states have some disproportional representation of children of color in the ASD category (Lord & Bishop, 2010). The findings of Morrier et al. (2008) are consistent with other studies of the disproportionality of racial/ethnic minorities with ASD: They suggest that socioeconomic factors combined with race/ethnicity be studied to better determine the causes of the under representation. Furthermore, families of color who have been provided a diagnosis for their child must be included in research to better understand their experiences once diagnosed (Morrier et al., 2008).

Disparities Post Diagnosis
Once a child receives a diagnosis of ASD, families continue to collaborate with health care and educational professionals. Carbone, Behl, Azor, and Murphy (2010) conducted a qualitative study to examine the differences in perspectives of pediatricians and families of children with ASD. The researchers interviewed five parents and nine pediatricians in separate focus groups. The pediatricians in the study cited lack of time and lack of care coordination as the major barriers to providing quality care to families. The parents shared that it was difficult to find a physician that used family-centered practices. The parents indicated that they sought support from their provider in referring them to the services available for their child and felt frustrated and angry when physicians disregarded their concerns about their child's development and behavior. The five families also shared that they were often "isolated, angry, frustrated, and fatigued" in identifying services on their own (p. 320).
Studies have also examined the specific frustrations that families of color have with the educational system. Zoints et al. (2003) examined 24 African American families' experiences within the special education system. Specifically, families were interviewed about their perceptions of cultural sensitivity by teachers and other school professionals. Forty-one percent of the parents interviewed in the study were unaware of trainings to improve cultural sensitivity and understanding of teachers. Of those parents who were aware of trainings, 57% reported not seeing outward evidence of cross-cultural sensitivity from their child's teachers. Additionally, one of the six themes from the study was "issue of quality training among teachers and other school personnel" in regard to developing cultural sensitivity. This study indicates a further need to ensure culturally responsive educators teach all students.
In addition to differences in education and health care quality, the availability of services for families of children with ASD varies. Mandell et al. (2010) revealed that 2004 Medicaid claims for children with ASD were from predominantly White communities, with higher number of specialty pediatricians in the area, and with a greater number of students in special education based on the ASD diagnosis. This study suggests a great need to provide racial/ethnic minority groups with targeted support to access the services provided through Medicaid at a rate similar to their White counterparts.

Family Support
In addition to the impact of the education and health care systems, family member support is also an important aspect of raising a child with ASD. Bayat (2007) examined the experiences of 175 guardians with children diagnosed with autism between the ages 2 and 18. In their responses to three open-ended questions about raising a child with ASD, the researcher found subcategories of family resilience themes: (1) pulling resources together; (2) being connected; (3) making meaning out of adversity; (4) affirmation of strength and being more compassionate; and (5) spiritual experience and belief system. Sixty-two percent of families identified being closer as a family because of the diagnosis and 63% percent were able to make meaning out of the diagnosis. This research suggests that families need services that support family strengths and characteristics.
Families of children with autism have also indicated that religious involvement has been a positive support in coping with the challenges of having a child with special needs. Ekas, Whitman, and Shivers (2009)
Family-centered care principles emphasize the role, experiences, and needs of the family in caring for children with disabilities. "The family, (not the professional) is the constant in the child's life; the family is the ultimate expert on the needs and wellbeing of the child; one cannot help a child without simultaneously helping a family (and often involve the community within which the family is nested); and whenever possible parents should be senior partners with professionals in the creation of service plans for their child" (Trute, 2007, p. 284).
Families of children with disabilities report that professionals need to listen and learn from parents, be culturally responsive to families, work as partners, and individualize how to support a family's unique needs (Goldfarb, Devine, Yingling, Hill, Moss, Ogburn, Roberts, Smith, & Pariseau, 2010).
Health care professionals that use family-centered practices recognize the cultural, linguistic, socioeconomic, ethnic and racial perspectives of families. This approach to medical care emphasizes that professionals highlight strengths of the family, and provide useful non-biased information and supports to families. When health care professionals utilize family-centered care, better outcomes in overall health care are reported (American Academy of Pediatrics, 2003). In addition, families report higher levels of satisfaction with their child's overall health care experience when family-centered care is reported (King et al., 2004). Within educational settings, families report better educational outcomes for their children when family-centered practices are used (Davies, 1995;Dunst & Trivette, 1996). However, family-centered care must be further investigated in community settings with diverse populations (Bellin et al., 2011).
Measuring the impact of family-centered care among various groups is important in understanding how to better care for families of children with ASD. (2011)
The racial/ethnic background has been used to determine differences in care in both health care and educational settings (Knapp, Madden, & Marcu, 2010;Montes & Halterman, 2011;Morrier, Hess, & Heflin, 2008;Ngui & Flores, 2006). Ngui and Flores (2006) found that African American and Hispanic families with children with disabilities had more dissatisfaction in their health care than their White counterparts.
Thirteen percent of African Americans and 16% of Hispanics were dissatisfied with the level of family-centered medical care, compared to 7% of White parents. Also, Montes and Halterman (2011) found that African American families with children with ASD were less likely to receive family-centered care from physicians than White families. Similarly, families of color were less likely to report feeling like a partner in their child's health care planning with their child's physician (Knapp, Madden, & Marcu, 2010).
In addition to the differences within the health care system, families of color, in particular families of color raising a child with a disability, have reported more negative experiences in regard to educational quality than White families (De Valenzuela et al., 2006;Fierros & Conroy, 2002). Fierros and Conroy (2002) suggest that African American and Hispanic students with disabilities are more likely to be placed in restrictive settings, limiting their access to the general education curriculum. Additionally, the quality of special education services for students of color has been of concern. Approximately 75% of African American students with disabilities do not have employment two years after high school graduation compared to 47% of White students with disabilities. Five years after high school graduation, 50% of African Americans students with disabilities are not employed compared to 39% of White students with disabilities.
De Valenzuela et al. (2006) examined students with disabilities in a large urban school district in the southwestern United States. The researchers found the overall educational quality for students with disabilities to be significantly worse for students of color. African American, Hispanic and Native American students were more likely to be placed in the most restrictive setting or a placement in a separate class 60% or more of the time than White, Asian, and other students. The researcher suggested that students from these ethnic/racial minority groups had a lesser chance of access to general education.
The income level of a family also has been used to determine differences in care in both health care and educational settings (Knapp et al., 2010;Montes & Halterman, 2011;Morrier et al., 2008.) Knapp et al. (2010)  Lastly, the level of education of a parent of a child with ASD also suggests differences in experiences in schools, health care settings, and their communities. Thomas et al. (2007) surveyed 383 caregivers in North Carolina to determine the characteristics associated with use of ASD services. Parents with more education have shown to have a greater chance of receiving access to quality care for their child with ASD. Caregivers with at least a college degree had two to four times odds of using some type of services. The services cited by families included direct therapy, such as occupational or speech therapies, as well as specific interventions such as Picture Exchange Communication Systems (PECS). Furthermore, Thomas et al. (2007) confirmed the findings of Newacheck, Hung, and Wright (2002) that families of color with children with ASD and parents with less education were less likely to receive services, such as occupational and speech therapy.

Health Care Provider Variables
Health care providers have been shown to affect patients' satisfaction of care they have received (Horn, Mitchell, Wang, Joseph, & Wissow, 2012;Levinson et al., 2008;Rutten, Augustson, & Wanke, 2006). Among them, race was a factor related to the quality of care reported by patients (Cooper, Roter, Johnson, Ford, Steinwachs, & Powe, 2003;Cooper-Patrick, Gallo, Gonzales, Vu, Powe, Nelson, & Ford, 1999;Saha, Arbelaez, & Cooper, 2003 interactions more participatory and more positive than patients with a physician of a different race/ethnicity as their own. Overall, the researchers found African Americans were more likely to rate their experiences with physicians of any race as less participatory and less positive than Whites. In addition to the race of the physician, some medical specialties have utilized family-centered practices with positive outcomes (American Academy of Pediatrics, 2003). Pediatricians have led the field in adopting a policy that supports familycentered care. Johnson and Myers (2007) recommend that all pediatricians screen children at 9, 18, 24 or 30 months for developmental delays. This recommendation, which is endorsed by The American Academy of Pediatrics, uses a family-centered care model.

Scale Development
To develop a scale to measure respondents' experiences, it is important to establish validity and reliability of the instrument. In order to construct a scale that is both valid and reliable, it is critical for researchers to begin with a theory behind the construct that is being measured (DeVellis, 2003). Typically a theory is selected from literature on the topic being measured.
In addition to the importance of theory in scale design, it is also key to determine the specificity of the construct being studied. Scales can be used to measure very specific attitudes or broader constructs that intend to capture a global set of behaviors (DeVellis, 2003). In addition, scales can be intended for use in very specific settings. For example, some scales were designed for use in one setting, such as a school.
Several of these scales posed problems for use in the current study, as some were not developed for families of children with ASD, were not administered to parents, or had poor psychometric properties.
The first concern with the scales currently measuring families' experiences toward working with physicians, educators, family members and community agencies is the level of specificity. For each of the scales, the researchers did not target one particular population, but focused on a wide range of disabilities (Bailey et al., 2011;Hoffman et al., 2006;Kontos & Diamond, 2002;Maijala et al., 2009;Seid et al., 2001;Summers et al., 2005). Other scales were focused on one particular type of setting, such as a hospital critical care unit, limiting the use of the scale (Maijala, et.al. 2009).
A second concern was the psychometric properties of the scales. Tinsley and Tinsley (1987) recommend the number of respondents needed to validate a scale should be approximately ten times the number of items on the scales. For example, an item pool of 60 questions would require 600 respondents. Thompson Tinsley and Tinsley (1987).
"The Parent Perceptions of Care (PPC)" developed by Maijala et al. (2009) measured the experiences of families who had children who were hospitalized for acute care. While this scale was created using a clear theoretical framework, the psychometric properties of the instrument were of major concern considering only 91 items. Kontos and Diamond (2002) validated a scale to examine the differences between ratings of parents with children in early intervention programs toward their early intervention providers in Indiana. "The Early Intervention Scale" contained four subscales: home-based therapies/instruction, centre-based therapies/instruction, medical health services, and service coordination. The initial 33-items were administered to 209 families, short of the recommendation by Tinsley and Tinsley (1987) for the number of respondents needed for validating the scale.
A third concern with previous scales is the absence of information about translations of the scales into other languages. Weeks, Swerissen, and Belfrage (2007) suggest researchers take careful consideration when translating instruments into other languages to avoid unintended cross-cultural differences. Back translation can cause errors in grammar, sentence structure, language difficulty level, inaccuracies, and inconsistencies when an instrument is translated from one language to another (Weeks et al., 2007). Some of the authors were unclear about the processes taken to translate their instrument from English to another language (Seid et al., 2001;Summers et al., 2005).
"The Family-Professional Partnership Scale" developed by Summers et al. (2005) looked only at the experiences of families in the school setting, and did not incorporate the experiences of families in the health care setting. Likewise the scale addressed the professional as "[your] child's service providers." This language did not allow one to determine if a family had varied experiences among different service providers. For example, if a family felt collaboration with a speech and language pathologist was positive, but collaboration with a special educator was negative, the scale did not make this distinction. The scale was not specific to children with ASD.
Additionally the researchers translated the scale into Spanish for families, but did not separate this information out to validate the scale in a second language (Weeks et al., 2007).
An additional scale for measuring health care quality developed by Seid et al.

Research Summary
Literature currently shows that children of color, specifically African

Americans and Hispanics are under-represented in being diagnosed with Autism
Spectrum Disorder nationally (Mandell et al., 2002;Palmer et al., 2010). This limits children from being afforded services that are often found to be most effective at young ages (Johnson & Myers, 2007). Studies show that the earlier a child is diagnosed with ASD, the better the prognosis (National Research Council, 2001).
Given these data, the aim of the current scale is to determine the family-centered experiences that families of children with ASD are having in three settings that impact their care: schools, health care settings, and the community. After a review of scales that measure a family's experiences in these settings, none adequately reflected the specific goals of the current research project.

Research Questions
The first portion of the study was to develop and validate the scale for use in English. Two questions were addressed in this part of the study.  Zoints et al., 2003). Therefore, in this study it was expected that families of color would have more negative experiences in schools, health care settings, and within their communities. Parents with more education have shown to have a greater chance of receiving access to quality care for their child with ASD (Thomas et al., 2007). In this study, 91.9% of the respondents were mothers (n = 434), 4.2% were fathers (n = 20), with grandparents, siblings, and other family members comprising 2.5% (n = 12). There were 1.3% (n = 6) respondents who declined to answer this question.
Respondents from 45 states were represented in the data. Respondents were not represented from Alaska, Delaware, Nevada, North Dakota, or South Dakota. In this study 14.4% (n = 68) of the families were from Massachusetts; 9.5% (n = 45) of the families were from Michigan. These states represented the largest number of respondents in the sample. The remaining 43 states each represented between 7.4% (n = 35 and .2% (n = 1) of the data collected. Table 2 provides a breakdown of the number of respondents represented by each state. Families reported their child's race/ethnicity. In this study, 79.9% (n = 376) of the children were White, 9.7% (n = 46) were Hispanic, 4.9% (n = 23) were Black, 2.3% (n = 11) reported as other racial/ethnic minority, 1.5% (n = 7) were Asian, .4% (n = 2) were American Indian, and 1.5% (n = 7) respondents declined to answer this question. For the purpose of the present study, the researcher collapsed those families who indicated they were racial/ethnic minorities into one group in order to have an adequate sample size to conduct hypothesis testing. To conduct the MANOVA concerning the race/ethnicity of families, those respondents who indicated they were American Indian or Alaska Native, Asian, Black, Hispanic, or Other Minority were collapsed into one group for a total n of 89. See Table 3 for this information.
These data can be compared to data from the U.S. Census Bureau complied in 2012. See Table 4 for this information (United States Census Bureau, 2012b).  Two or more races 2.5 Hispanic or Latino (of any races) 16.9 Families who participated in the survey reported their approximate household income. Families who earned less than $100,000 represented 64.2% (n = 303) of the data. Families who earned more than $100,000 represented 27% (n = 127) of the data.
Families who declined to answer the question about income level represented 7.6% (n = 36) of the data and 1.3% (n = 6) of the data were missing. See Table 5 for the break down of household income reported. (United States Census Bureau, 2012a). Families also reported their highest level of education. Table 6 provides the data of caregiver educational attainment level (United States Census Bureau, 2012a).

Instrumentation Personal Background Information
Respondents were asked to answer thirteen personal background questions in the study. The families reported demographic information about their child with ASD: type of ASD diagnosed, race/ethnicity, year of birth, and gender. In addition, the families reported demographic information about themselves: relationship to the child with ASD, language mostly spoken at home, their race/ethnicity, city or town they resided in, state of residence, approximate household income, and their highest level of education completed. The families also reported the type of professional who diagnosed their child and the race/ethnicity of this professional.

Previously Published Scales
The first part of the study involved completing a review of all of the scales previously developed that measure the experiences of families of children with ASD in schools, health care settings, family environment, and in their communities. Using journal databases at The University of Rhode Island and Brown University, the researcher completed seventeen extensive searches in psychology, medicine, education, sociology, and social work databases for scales that measured the experiences of families of children with ASD. The types of experiences that were searched for included schools, health care settings, their families (extended and immediate), their communities, and their places of worship. In addition, familycentered care scales were also searched for. There was a lack of published research on scales specifically designed for families of children with ASD, thus the researcher broadened the search criteria to include scales created to measure the experiences of families with children with disabilities. This yielded more studies in the search.
As previously mentioned several of these scales posed problems for use in the current study (Bailey et al., 2011;Hoffman et al., 2006;Kontos & Diamond, 2002;Maijala et al., 2009;Seid et al., 2001;Summers et al., 2005;Thompson & Mazer, 2012). This process confirmed for the researcher that understanding the experiences of families of children with ASD were important to add to the literature base.

Item Development
Family-centered care was chosen as the theoretical frame for the items developed for this study. Using the American Academy of Pediatrics (2003)  The Likert-type scaling of the survey included response options "4 = strongly agree," "3 = agree," "2 = disagree," and "1 = strongly disagree." The Likert-type scale contained the option of "Does not Apply" (NA) to ensure that respondents who felt that an item did not apply were able to select this option. For example, a family could have homeschooled their child, thus the items on the "School Support" Scale would not apply.
The item pool was divided into three areas: health care, family/community, and school. These settings were used to get a broad understanding of the types of interactions families of children with ASD experience with individuals in each of these settings. Each item was coded by the researcher to ensure that each of the five components of family-centered care were represented in the three settings being presented in the items. Positively and negatively worded items were constructed as recommended by Fowler (1995). The researcher also applied the guidelines of writing items that were not double-barreled and avoiding unintended question order effects (Dillman, Smyth, & Christian, 2009;Fowler, 1995).
Double-barreled items are problematic in scale design, as they include two concepts that make it difficult or impossible to distinguish what the item is measuring.
An example of a double-barreled item is: "My child's doctor listens to me and is caring." There are two concepts that are being measured: listening and caring.
Instead this item should be separated so that there are two items to assess each concept. In addition to creating items that were not double barreled, the researcher also followed the recommendation of Dillman et al. (2009) andFowler (1995) to avoid unintended question order effects. The researcher grouped the items by topic and asked the demographic questions at the end of the instrument as to not influence the respondents' answers to the items measuring their experiences.
Once the item pool was written, content validity was examined through expert and family feedback. The experts and families were from various locations around the country. The 62 items were sent to three experts based on the following: (1)  The researcher examined each item score for relevancy and clarity for any item that was rated with a 1 in either relevancy or clarity on the Likert-type scales by more than 2 respondents (experts or families). The content validity ratio (CVR) for each item was calculated using the following formula: CVR = n e -N/2) / N/2 where CVR = content validity ratio, n e = number of raters indicating "essential," N = total number of raters (Waltz & Bausell, 1983). The essential values in this study were those items scored by raters with either a 2 (item needs some revision to be relevant) or 3 (item is relevant) for relevancy. The CVR was calculated for each item and any item with a value of .75 or lower was examined more closely. In total, there were 11 items that received a CVR value of .75 or lower and were discarded from the final instrument because of the low interrater agreement (Waltz & Bausell, 1983).
In addition to rating the relevancy of the items, there were twelve items that were commented on by the experts and families in regard to their clarity. This qualitative information was used in discarding an additional 12 items from the instrument. A former educational statistics professor and a current special education professor at The University of Rhode Island reviewed the qualitative feedback provided by the six people and agreed with the researcher to eliminate the 23 items from the scale. The same special education professor was consulted about the final 39 items prior to sending out the final instrument to be piloted.
The next step in the process involved the researcher coding the 39 items.
Using family-centered care literature, the researcher developed five codes describing aspects of family-centered care (American Academy of Pediatrics, 2003). Once the researcher coded the items, three professionals were trained and asked to code the items. Table 7 shows the codes the three professionals used to code the items within the scale. One professional was an ASD researcher and parent of a child with ASD from the Midwest, one was a special education teacher from the Northeast, and the third professional was a director of a youth community center from the Midwest. The researcher trained these professionals to code the items using examples of various sample items that fall into each code. The professionals were able to ask questions about the process and then were asked to independently code the items. None of the professionals had difficulty understanding the directions and were able to code the 39 items successfully. See Appendix C for a copy of the letter sent to professionals who coded the items.  (Stevens, 2002). The interrater agreement for the three professionals was .82, .84, and .87.

Level of Information Sharing &/or Seeking
The extent to which a professional communicates with a family member (through sharing and seeking information)

Level of Respectful & Supportive Interactions
The extent to which a professional values or supports the family member or child

Level of Establishing Collaboration &/or Partnerships
The extent to which a professional offers to collaborate or create a partnership with the family member or another professional

Level of Competency
The extent to which a professional or parent has the level of knowledge, skills, and follow-through in supporting a child with ASD

Level of Access to Services
The extent to which a family is able to involve their child with ASD in supports, interventions, or services Other Please write down the code you feel best represents this item

Web-based Scale
The researcher used a web-based survey program, Survey Monkey, to create the online scale. The online scale was tailored to respondents using visual components such as color, size, and organization logo suggested to improve response rates (Dillman et al., 2009) information, such as the respondent's name or contact information, was collected from the caregiver to ensure subject anonymity. The last page of the online survey allowed a family to be redirected to a separate website to be entered into a drawing. This ensured that no respondent information would be linked to their survey responses. A customized web address was created for the study.
Once the FEASD was placed online, five people piloted the survey to ensure that all links worked correctly. The five people confirmed that the web-based FEASD Scale worked and the researcher began data collection. See Appendix G for a copy of the letter sent to individuals who piloted the survey.

Data Collection using FEASD
Respondents were recruited from organizations that served families of children with ASD. The researcher contacted at least one organization or support group that Organizations were asked if they could provide contact information for other organizations that serve families of children with ASD to the researcher. This convenience sampling methodology is known as snowball sampling (Creswell, 2009;Patton, 2002). Snowball sampling requires the researcher to start with a list of possible participants (in this case ASD organizations) and then each respondent is asked if he/she knows of other parents/guardians who would like to complete the scale (Creswell, 2009;Patton, 2002). This sampling methodology was selected since families of children with ASD are often connected with one another in various organizations and online support groups. Although this methodology was considered a type of convenience sample, the respondents in this study would be best recruited through snowball sampling.
As previously stated, respondents were given the option of clicking on a link to a second survey to provide their email address and their phone number to be entered into a drawing for $25.00 at the end of the survey as an incentive to complete the survey. This procedure was employed so that no information that the respondent entered in the survey was linked to their contact information for the purpose of the drawing. The incentives were thoroughly explained to all respondents who completed the survey. The Organization for Autism Research (OAR) generously provided the researcher with a grant, which funded both the drawing for the respondents and the organizations who participated in the study.
After four months of data collection, the web-based survey was taken down. A total of 472 respondents completed the survey after four months of being available online. The data were downloaded into SPSS, a computer based statistical analysis program. After all data were collected, the incentives were distributed to the families and to an organization. Ten families in total were randomly selected to receive a $25.00 gift card. One organization that posted the link to the online survey was randomly selected to receive a $100.00 donation from the researcher.

Overview of Data Analyses
To answer the first two questions in the study, a data screening process was first employed in the data analysis. First, data were checked for accuracy, outliers, and missing values using SPSS 19. Assumptions for normality, linearity, heterogeneity of variance, and factorability of the correlation matrix were examined to ensure that all assumptions were met to perform the statistical tests. Second, the descriptive statistics of each item were examined using item means, standard deviations, range of scores, skew and kurtosis. Third, the psychometric properties of the scale were examined using principal components analysis. Lastly, the reliability of the scale was examined using coefficient alpha.

Initial Screening of the Data
The first step in the screening process was to examine the missing data. When examining individual responses, there were 503 people who completed the survey.
However, only 472 people completed the items in the survey from the beginning to end. It was discovered that 31 respondents answered only the first demographic question (Q1: What form of Autism Spectrum Disorder (ASD) does your child have?) out of 52 questions. These 31 respondents were eliminated from the study.
Additionally 6 respondents did not answer the remaining demographic/personal questions following the Likert items. These respondents were also eliminated from the study. This results in a final sample of 466 respondents.
Using listwise deletion procedures, there was 18.4% missing data in the overall sample. Upon examining individual items using pairwise deletion, it was found that the most data missing on any one item was 1.5%. Thus, pairwise deletion was selected for all subsequent analyses.
Next, the data set was screened for univariate outliers. Thirteen univariate outliers were found to be greater than 3.29 standardized scores away from the mean.
These items were checked for accuracy as well as patterns of respondent input. It was determined that the data were accurately entered and nothing appeared abnormal in the data set. With additional examination of the Extreme Values tables and charts (histograms and boxplots) for outliers, all scores were in the range of possible scores and no outliers were indicated.
Multivariate outliers were also screened for using Mahalanbois Distance procedure (Stevens, 2002). Mahalanbois distance can determine the multivariate outliers of a data set with fewer than 2,000 participants. Critical values were used for comparing the Mahalanobis distances (Pallant, 2001 Only one respondent had a Mahalanobis value that exceeded the critical value of 26.12 (respondent's value = 28. 24). This respondent's data were reviewed and were checked for accuracy and patterns. It was determined that the data were accurate so this respondent's data were included.

Item Analysis
Prior to conducting the factor analysis, exploratory item analysis of the The data were then examined for normality. Skew and kurtosis were examined for all 39 variables to determine if the data set was a normal distribution. Using the guidelines of two for skew and four for kurtosis, question 39 did not meet these criteria for normal distribution (Stevens, 2002). Examining this item's histogram confirmed this finding. The item was transformed using log10 reflection transformation. The histogram for the transformed item showed that the distribution was more normal than what was presented in the raw data. The transformation data met the assumptions for normality to conduct parametric statistics, thus the transformed data were included in all MANOVA tests. Table 8 shows the item skew and kurtosis in raw data and transformed data.

Questions #1 & 2: What is the factor structure of the scale and what is the internal consistency of the scale?
The researcher used a parallel analysis statistical test to determine the number of components to retain in the principal components analysis (PCA) (Stevens, 2002).
The Monte Carlo simulation for parallel analysis was conducted to determine the number of factors or components to retain in the PCA. Parallel analysis is an alternative method to the scree plot method or the Kaiser rule which suggests retaining eigenvalues greater than 1. The researcher used parallel analysis rather than the previously mentioned methods for determining the number of components to retain because it has been a more robust method for determining the number of principal components to retain (Franklin, Gibson, Robertson, Pohlmann, & Fralish, 1995).
Using the Monte Carlo simulation test in SPSS, the researcher determined that there were 4 components that were statistically significant to include in the PCA.
To determine the type of rotation that would be best for interpretation, the inter-factor correlation matrix was examined to determine if an orthogonal or an oblique rotation would best help interpret the factors. Orthogonal rotations are used when the rotated factors are uncorrelated; whereas oblique rotations are used when the rotated factors are correlated (Stevens, 2002). The FEASD factors were only minimally correlated, thus Varimax orthogonal rotation was selected for the analysis.
The PCA was conducted by forcing four components to be retained. When this analysis was conducted, the fourth component only had two items that loaded.
This component had fewer than three items, the minimum number of items required for a component to be adequate for further analyses (Stevens, 2002). It is recommended that components with fewer than 3 items are unreliable and should be discarded (Stevens, 2002).
The researcher then conducted a PCA by forcing three components to be retained. When this analyses was conducted, all but two items on the scale loaded with values of .4 or above on the three components. To better understand what components did not load on the hypothesized factors, each item was examined to determine on which factor each loaded.
Upon examining the pattern and structure matrix of the Varimax rotation, two items had values less than .4: Item 4 (I pay a lot of money to get ASD services for my child) and Item 7 (There are high quality, free community programs for my child).
Based on this information, these items were excluded from the instrument.
After items 4 and 7 were deleted, the Cronbach's Alpha was calculated. For the first factor ("School Quality") the coefficient alpha was found to be very good at .96. The coefficient alpha for the second factor ("Health Care Quality") was good at .89, and the coefficient alpha for the third factor ("Family Support") was acceptable at .70. The overall coefficient alpha for the entire scale was very good at .92, accounting for 48.58% of the total variance of the scale. Table 9 contains the factor structure of the FEASD.
The researcher named the three factors. Factor 1, "School Quality," contained 18 items that intend to measure a family's experiences working with school officials.
All of these items were hypothesized to load on this factor. Factor 2, "Health Care Quality," contained 11 items that intended to measure a family's experiences working with their child's physician. All of the items were hypothesized to load on this factor, as they were initially developed. Factor 3, "Family Support" contained 8 items that intended to measure a family's rating of how supported they feel from immediate and extended family members in regard to having a child with ASD.  First, the assumptions of multicollinearity and singularity were examined. One way to determine multicollinearity was through analyzing the tolerance and variance inflation factor (VIF) among the variables (Stevens, 2002). The smaller the tolerance value, the more likely the variable is linear. Using the dependent variables of "School Quality," "Health Care Quality," and "Family Support" and the independent variables of "Diagnostician," "Child's Race/Ethnicity," "Physician's Race/Ethnicity," "Parent's Educational Level," and "Family Income" the VIF values were all less than 2 for each combination, suggesting that the dependent variables were only moderately correlated.  (Stevens, 2002). The highest correlation among the dependent variables was .307 (between "School Quality" and "Family Support"), although this value is below the 0.70 value that would be concerning (Stevens, 2002). These correlations and the tolerance and VIF values that were calculated indicate there was no violation of the assumption of multicollinearity for the study.    20.24) in their experiences on the Total FEASD Scale.
These findings indicate that families who had a pediatrician as the person who diagnosed their child with ASD reported significantly more positive experiences overall on the FEASD Scale as compared to those families who had a psychologist diagnose their child with ASD. it is vital to examine settings in which families of children with ASD most frequently navigate (Bellin, et al., 2011). Previous research suggests that families of children with ASD have a more difficult experience in schools, health care settings, and in their family lives (Hagner et al., 2012;Hendricks & Wehman, 2009;Howlin et al., 2004).
However, no scales have been developed to measure the experiences of families of children with ASD in all three areas (Bailey et al., 2011;Hoffman et al., 2006;Kontos & Diamond, 2002;Maijala et al., 2009;Seid et al., 2001;Summers et al., 2005;Thompson & Mazer, 2012). Therefore, there were two main objectives of the current research study. The first purpose of this study was to develop a scale that measured the experiences of families with children with ASD. This involved creating the items used in the scale and calculating the psychometric properties of the scale including the factor structure and internal consistency of the scale to create a new validated instrument, The Family Experiences with Autism Spectrum Disorders (FEASD) Scale.
The second purpose of the research study was to use the validated FEASD to assess families of children with ASD across five personal background questions.

Psychometric Characteristics of the FEASD scale
The researcher investigated psychometric properties of the FEASD scale. The 39 items were examined using principal components analysis. It was hypothesized that the items would load onto 3 factors: Community/Family Support, School Quality, and Heath Care Quality. The initial loadings were mostly consistent with this hypothesis.
The first factor, "Community/Family Support," was hypothesized to have 10 items, but after the analysis, the subscale was reduced to eight items. Two items that were below the recommended loading of .4 were discarded from the final instrument.
The decision to include items that focused on family support was based on the literature that has shown the importance of family support for caregivers of children with disabilities (Bayat, 2007;Ekas et al., 2009). Additionally, the guidelines of including subscales with three or more items and using items with correlations .4 or better were used (Stevens, 2002).
The second factor, "School Quality," had 18 items, the same as hypothesized prior to the principal components analysis. There were no items that were eliminated from the scale based on the psychometric properties from the "School Quality" scale.
The third factor, "Health Care Quality," had 11 items prior to the principal components analysis. After the psychometric properties were examined, all items were retained in the final scale. See Table 12 for the list of the final items on the FEASD scale. The FEASD scale had a satisfactory internal consistency and proved to be a valid measure. Family-centered care provided a strong theoretical framework for the items developed in the scale and fit well within the three factors on the scale. Familycentered care is a theory that is cited in educational and health care research as a best practice (Beatson, 2008;Dunst, 2002;Epley et al., 2010;King et al., 2004;Kuo et al., 2011;Moore et al., 2009;Rosenbaum et al., 1998;Tomasello et al., 2010;Trute, 2007). The FEASD scale is an important measurement tool of family support, and school, and health care quality for families of children with ASD, as it is the first instrument that aims to measure experiences in all three settings.

Effects of Family and Health Care Provider Variables on FEASD Scale
The second purpose of the study was to use the newly created scale to examine various demographic variables in the sample. Research questions three, four, five, six, and seven are discussed below.
The third research question examined the impact of race on the three factors in the FEASD. It was hypothesized that families of color would have a more negative experience in each of the three subscales of the FEASD, as documented by previous literature (De Valenzuela et al., 2006;Fierros & Conroy, 2002;Knapp et al., 2010;Mandell et al., 2010;Montes & Halterman, 2011;Ngui & Flores, 2006;Thomas et al., 2007;Zoints et al., 2003). The MANOVAs performed on the data showed that race does not predict the experiences of families on any of the FEASD scales.
To better understand the sample from the current study, the income levels of the families by race were examined more closely and compared to the US Census data income by race. This information may explain the reason that the data did not show significant differences. In the current study, 22.2% of Black families reported earning $100,000 or more annually, compared with 9.3% of Black families from the US Census data (2013). This suggests that the Black families who participated in this research study were more affluent than the Black population in the United States.
Likewise, 14.3% of Hispanic families reported an annual household income of $100,000 or more compared with the 11.7% reported by the US Census (2013). These data suggest that the sample in this study was not representative of the US population by race and annual household income. See Table 13 for the annual household income by race/ethnicity from the US Census (2012) compared with the sample included in this study. Note. The FEASD data sum for each race/ethnicity does not equal 100% because of respondents who did not answer the question. items concern the level of family support that caregivers of children with ASD feel.
Families who reported earning $100,000-$124,999 a year had a statistically significant higher mean score on the Family Support scale than families who earn less than $25,000 a year. These results are similar to Kogan et al. (2008) Cooper-Patrick et al., 1999;Saha et al., 2003).
The sixth research question examined the impact of the type of specialist that diagnosed their child with ASD on the three components in the FEASD and the overall scale mean. It was hypothesized that families who had a pediatrician diagnose their child with ASD would have more positive experiences, as documented by previous literature (American Academy of Pediatrics, 2003). The items included in the scale were written using family-centered care components. Since the American Academy of Pediatrics has endorsed family-centered care as a best practice, the researcher hypothesized that families who worked with a pediatrician would have more positive experiences.
The MANOVAs performed on the data showed that having a Pediatrician diagnose a child does predict the experiences of families on the Total FEASD Scale.
The mean scores were significantly more positive for families who were cared for by a pediatrician than those families who were cared for by a psychologist. The results corroborate with Rhoades et al. (2007) who found pediatricians were more likely to contribute additional information about education-related support services, such as occupational therapy or social skills training, that help their children with ASD than other health care professionals.
Interestingly, the mean scores of the Health Care Quality Scale were not statistically different. Families who reported a psychologist as the diagnosing professional had the lowest mean score on the Health Care Quality Scale. One might hypothesize that the Health Care Quality scale would be significantly different, since the physician most often impacts health care quality. This finding suggests that the type of professional that diagnosed the child with ASD impacts a family's experiences overall in each of the three areas. Family-centered care stresses the importance of collaboration, communication, and respect; all of which are part of the FEASD Scale.
The seventh research question examined the impact of the education of the caregiver on the three components in the FEASD and the overall scale mean. It was hypothesized that caregivers with more education would have more positive experiences in each of the three subscales of the FEASD, as documented by previous literature (Thomas et al., 2007). The MANOVAs performed on the data showed that education does not predict the experiences of families.

Limitations
There are several limitations with the current study. First, the results should be cautiously generalizable since the household income levels and number of respondents from a racial/ethnic minority group were not representative of the United States population.
Second, due to the small number of some of the racial/ethnic groups, all families of color were collapsed into one racial/ethnic minority group. This reduced the ability to detect differences at more specified levels of the IVs. For example, there may be significant differences in all three scales between Asian families and Black families, but due to the low number of Asian families who participated in the study (n = 7) this could not be explored. Fourth, snowball sampling was employed to recruit respondents in this study.
This method of recruitment is less desirable than random sampling, which would have randomly selected families of children with ASD to participate in the study.
Fifth, the survey was accessible to families who spoke English. Families with linguistic backgrounds other than English were excluded from participating in this study. While there are many challenges to translating an instrument into another language, the fact that this survey was only accessible to families who spoke English narrows the particular sample who participated in the study.

Experiences to Obtain ASD Diagnosis
This study aimed to examine various demographic factors in regard to experiences in schools, health care settings, and in their families. These three settings are connected to one another, as a diagnosis by a physician is often required for a child to be provided special education services. While this information offered insight into families who currently have a child diagnosed with ASD, it does not explain the challenges that families face while trying to obtain a diagnosis of ASD. Mandel et al.
(2002) indicates a later diagnosis for families of color, thus examining the experiences of these families would be an important contribution to research. Either a quantitative or qualitative examination of the racial, cultural, and/or socioeconomic experiences that impede families from obtaining an ASD diagnosis would help in understanding how to provide support to these families.

Use of the FEASD
As previously mentioned, differences between racial/ethnic minority groups could not be measured due to the sample size. Another study using the FEASD Scale should focus on obtaining responses from more families of color so that disaggregated analyses can be conducted. Also, it would be important to obtain a sample more representative in terms of income by race. Furthermore, adding a personal background question that inquired the age that the child was diagnosed with ASD, the types of services families receive, the annual out of pocket expenses to pay for services, and/or the severity of their child's disability could be examined further. It would also be important to develop and validate the scale in other languages.

Concluding Remarks
The FEASD Scale is a validated scale to measure the experiences of families of children with ASD based on the results of principal components analysis and internal consistency reliability. The scale appears to measure the School Quality, Health Care Quality, and Family Support of families of children with ASD. Two of the findings were consistent with previous research (American Academy of Pediatrics, 2003, Knapp et al., 2010Kogan et al., 2008;Mandell et al., 2010;Montes & Halterman, 2011;Morrier et al., 2008, Rhoades, et al., 2007  Read each item. Use the scales below to determine the relevance and clarity of each item. Please feel free to comment on items or suggest revisions as you feel are necessary. Adapted from Waltz and Bausell (1983)  I invite you to code the following items on the Family Experiences with Autism Spectrum Disorder (FEASD) Scale. In table 1, you will find 5 pre-determined codes These codes have been developed from extensive review of research on familycentered care, a best-practice for working with families of children with special needs.

ITEMS Relevance Clarity COMMENTS/SUGGESTIONS
For each item answer the following question, "What is the underlying theme of this statement?" Each of the pre-determined codes has been assigned a letter to make the process easier for you.
For example, if you feel that item 1, should be coded with "Level of Information Sharing &/or Seeking," write the letter, "A" in the box marked "Code." If you feel that the item does not fit in any of the codes, please write in the code you feel best represents that particular item. In addition, if you have any additional comments that you feel would be helpful, please provide these comments in the box provided to the right of each item. The purpose of this study is to examine the experiences of primary caregivers with children with Autism Spectrum Disorder (ASD). The research will be studying these experiences from families across the United States. Responses to these items will involve filling out a survey about your experiences in raising a child with ASD, and accessing services, as well as some personal background questions. All of the anonymous data collected in this study will be kept on a password-protected computer in a locked office at the University of Rhode Island.

YOU MUST BE AT LEAST 18 YEARS OLD to be in this research project.
If you decide to take part in this study, your participation will involve filling out a survey pertaining to your experiences in raising a child with ASD, and accessing services, as well as some personal background questions. The survey should take you approximate 15 minutes to complete.
The possible risks or discomforts of the study are minimal, although you may feel some embarrassment answering questions about private matters.
Although there are no direct benefits of the study, your answers will help increase the knowledge regarding how to best meet the needs of families of children with ASD in education, health care and organizational settings.
Your part in this study is anonymous. That means that your answers to all questions are private. No one else can know if you participated in this study and no one else can find out what your answers were. Scientific reports will be based on group data and will not identify you or any individual as being in this project.
The decision to participate in this research project is up to you. You do not have to participate and you can refuse to answer any question.
Participation in this study is not expected to be harmful or injurious to you. However, if this study causes you any injury, you should write or call Mr. Adam Moore at (401)  Please use the following form to provide your feedback about any item or demographic question on the survey. Your input will help make final adjustments to the scale before it is sent out for families to complete.
If you have questions or would like to talk to me, I can be reached at (401)  A few days from now you will receive a request via email to post a short questionnaire for a dissertation study being conducted at the University of Rhode Island. The email asks your assistance in sending a message on your ASD listserve to families of children with ASD. This questionnaire aims to better understand the experiences of families with children with Autism Spectrum Disorder (ASD).
I would appreciate your help in reaching out to as many families as possible to take part in this short survey.
Organizations who agree to post the link to the survey on their website or contact families about this survey will be entered into a drawing for a $100 donation.
Additionally, individual families who participate in the survey will be entered into a separate drawing for $25.
If your organization supports families who do not have access to the Internet, paper surveys are available. The instructions sent in the email will ask families to call, tollfree, 1-866-733-4190 and provide their name, address, and a number so that study materials may be sent to them. Alternately, I can provide your organization with an electronic version of the survey that can be printed out and distributed to families who wish to complete paper surveys and either mailed or faxed back to me.
If you have questions or would like to talk to me, I can be reached at (401)  A few days ago you received an email from me contacting you about posting a link to a survey that aims to better understand the experiences of families with children with ASD.
Below is information that I would appreciate you post on your website for families so that they may participate in the survey.
If you have questions or would like to talk to me, I can be reached at (401) 874-4200 or at the email this message is being generated from. Additionally if you would like to speak to my supervisor, Dr. Joanne Eichinger, she can be reached at (401)  THANK YOU for your willingness to post the link to the Family Experiences with Autism Spectrum Disorder (FEASD) survey to your organization's website! This project could not be possible without the dedication of organizations like yours-your generosity is greatly appreciated! Your organization is being entered into a drawing for $100 cash donation as a small token of appreciation! Would you also be willing to provide me with the names of other organizations in your area that support families with children with ASD who may be interested in completing this survey?
If you have questions or would like to talk to me, I can be reached at (401)  Thank you for your consideration to post the link to the Family Experiences with Autism Spectrum Disorder (FEASD) Scale. The work that your organization does for families with children with ASD is extraordinary and should not be overlooked. As a passionate researcher in the field, I commend your efforts and understand that you cannot assist in all research projects.
If you reconsider posting information about this survey to your website or within your organization, please feel free to contact me at the email address or at (401)