Participants

Our sample was comprised of forty-eight siblings. Forty-three (89.6%) of the participants were women; five (10.4%) were men. The majority of the participants (n = 41, 85.4%) identified as nondisabled, however a number (n = 7, 14.6%) were disabled. Participants were White (n = 38, 79.2%), Asian or Pacific Islander (n = 3, 6.3%), Black (n = 3, 6.3%), Hispanic or Latino/a (n = 2, 4.2%), Middle Eastern (n = 1, 2.1%), and interracial (n = 1, 2.1%). Participants ranged in age from 18 to 68, with a mean age of 38.23 years (SD = 13.28).

Measures

The Disability Attitudes Implicit Association Test. The Disability Attitude Implicit Association Test (DA-IAT) was utilized to examine participants' implicit disability attitudes (Greenwald et al., 1998). The DA-IAT presents participants with word and symbol stimuli and asks them to sort items as quickly as possible into two target-concept discriminations, 'disabled persons' and 'abled persons,' and the two attribute dimensions, 'good' and 'bad,' with one target-concept discrimination and one attribute dimension on each side of the computer screen at a time. For example, 'disabled persons' and 'good' on the left side of the screen and 'abled persons' and 'bad' on the right side of the screen. The target-concept discriminations and attribute dimensions switch sides so they are both stereotype consistent and inconsistent at different times. The DA-IAT measures the differences in reaction time when participants sort items into stereotype consistent and inconsistent categories.

The DA-IAT is the most prominent implicit measure of disability attitudes, having been administered to over 38,500 participants through the open-access Project Implicit study website alone (Nosek et al., 2007). The DA-IAT has been found to have construct validity (Aaberg, 2012; Pruett, 2004; Pruett & Chan, 2006), discriminant validity (White, Jackson, & Gordon, 2006), and reliability (Pruett, 2004; Pruett & Chan, 2006; Thomas, Vaughn, Doyle, & Bubb, 2013).

The Symbolic Ableism Scale. The Symbolic Ableism Scale (SAS) (Friedman & Awsumb, in preparation), which is an adapted version of the Symbolic Racism Scale 2000 (Henry & Sears, 2002), was utilized to measure participants' explicit disability attitudes. The SAS administers thirteen statements about disability to participants and asks them their opinion on a seven point Likert scale from strongly disagree to strongly agree. For example, one statement participants are presented with is: 'Most disabled people who don't get ahead should not blame the system; they really have only themselves to blame.' The Symbolic Racism Scale 2000 has good construct validity, discriminant validity, and predictive validity; it is also internally consistent (Henry & Sears, 2002).

Other items. In addition to the use of the two scales above, participants were also asked the open-ended question: "What is your definition of disability?" We also asked participants a number of other questions about their demographics, political orientation (sliding scale), relationships with disabled people, and involvement in disability advocacy.

Procedure

After completing an informed consent via the study website, participants were instructed that in order to participate in the DA-IAT they would push the 'E' and 'I' keys on the keyboard to sort stimuli to on the left and right side of the screen respectively. The instructions told them to do so as quickly as possible but with the least amount of errors. If a mistake is made and a stimulus is placed in the wrong category a red 'X' appeared in the middle of the screen until participants correct their answers. After viewing the instructions participants completed the DA-IAT, which is comprised of seven blocks (rounds) of categorization tasks. In the first block participants practice sorting only target-concept discrimination ('disabled persons' and 'abled persons') related stimuli (see table 1). (It should be noted that the terms disabled and abled persons are used to measure disability broadly rather than at an individual level.)

Table 1: Example of DA-IAT Block Presentation. Block one lists 'disabled persons' on the left and 'abled persons' on the right. In the middle there is the wheelchair symbol serving as the stimuli. Block 2 lists 'good' on the left and 'bad' on the right. The word bomb in the middle serves as the stimuli. Block 3 and 4 lists both 'disabled persons' and 'good' on the left and both 'abled persons' and 'bad' on the right. The stimuli in the middle is a stick figure of a person walking with a cane. Block 5 lists 'bad' on the left' and 'good' on the right. The stimuli in the middle is the word 'love'. Block 6 and 7 list 'disabled persons' and 'bad' on the left and 'abled persons' and 'good' on the right. The stimuli word, 'rotten', is in the middle.

Table 1

Example of DA-IAT Block Presentation
Block 1
Disabled persons		Abled persons
Block 2
Good	Bomb	Bad
Blocks 3 and 4
Disabled persons		Abled persons
Good		Bad
Block 5
Bad	Love	Good
Blocks 6 and 7
Disabled persons	Rotten	Abled persons
Bad		Good
Note. The stimuli presented in the middle are examples.

The second block, which is also practice, presents only the attribute-dimensions ('good' and 'bad') related stimuli for sorting. During blocks three and four participants are presented with both the target-concept discriminations and attribute-dimensions and all the stimuli. The computer randomizes if they are presented with these items in a stereotype congruent or incongruent manner. In block five, only the attribute dimensions are presented, however they are now on opposite sides so that the participants can practice with the new location. Finally, in blocks six and seven participants are again presented with both the target-concept discriminations and attribute dimensions and all stimuli; however, these items are presented opposite to blocks three and four so if they were originally stereotype congruent they are now presented as stereotype incongruent and visa versa.

After the DA-IAT participants responded to the thirteen SAS disability statements from strongly disagree to strongly agree. Next participants completed the questionnaire items about their definitions of disability, and about their demographics and other factors.

Analysis

Participants' definitions of disability were analyzed using thematic analysis, which is a useful method for examining similarities and differences both within a single participant's response and across participants (Braun & Clarke, 2006; Manhire et al., 2007). A theory-driven approach, which involved engaging with relevant literature (Kafer, 2013; Kittay, 1999; Siebers, 2008), was used. Prior to examining the qualitative data, the authors discussed how the theoretical frameworks would inform data analysis, focusing on three aspects:

1. The unique, but multiple experiences of individuals who may not identify as disabled themselves but may be in close relationship with disabled people, such as parents, partners and siblings (Arnold et al., 2012; Heller & Kramer, 2009; Kafer, 2013);
2. Longstanding discrimination against disabled people, which may be related to certain understandings of disability (Charlton, 1998; Daruwalla & Darcy, 2003);
3. Theoretical concepts related to disability, including disability as social movement (Hahn, 1985), disability as identity (Gill, 1997); disability as relational (Thomas, 2004; Kafer, 2013), and disability as complex embodiment (Siebers, 2008).

Then, the open-ended responses were independently coded by each author, using the themes described above. Each author then used the "in situ" codes to create themes. The two authors then met to discuss the codes and themes, combining themes together to reflect a cohesive analysis. We then returned to the raw data to verify the themes accurately represented the data, enacting the qualitative value of trustworthiness (Krefting, 1991).

Participants' implicit disability attitudes (DA-IAT) were calculated using Greenwald, Nosek, & Banaji's (2003) updated IAT scoring procedure, which compares participants' own stereotype congruent and incongruent blocks using response latencies. DA-IAT scores reveal participants' implicit preference for nondisabled or disabled people (Aaberg, 2012; Greenwald et al., 2003). Participants' mean scores on the SAS items served as their explicit disability attitude scores. The rest of the quantitative data was analyzed using descriptive statistics.

In addition to this qualitative analysis, we also quantified the qualitative data as a form of across-methods triangulation in order to enhance understandings of the data (Bernard, 1996; Chi, 1997; Fielding, 2012; Jick, 1979; Sandelowski, Volis, & Knafl, 2009; Ward, 2007; Young, 1981). Utilizing this mixed method allows one to "incorporat[e] multiple viewpoints in defining the theoretical boundaries of assessment practice, allowing numbers to 'speak' in order to enhance our understanding of data" (Ward, 2007, p. 10). Moreover, this quantified analysis was a beneficial additive to the "narrative layering and textual meaning" produced by the qualitative analysis (Ward, 2007, p. 10). Qualitative themes regarding participants' definitions of disability were quantitatively coded using indicator coding so relationships between their understandings of disability with disability attitudes, and other factors could be determined. For example, participant would receive a 'one' where their response reflected a theme, and a 'zero' where it did not. Themes were not mutually exclusive, that is participants could receive ones for multiple themes. After these analyses were completed, multiple regression models were run to determine statistically significant relationships between definitions of disability, disability attitudes, and other factors.

Descriptive Statistics and Prejudice Summary Data

On a sliding scale that ranged from very liberal (1) to very conservative (100), participants' mean political orientation was 30.63 (SD = 22.3). Two-thirds of participants (n = 32) have worked in a disability related industry. Of those participants, they had worked in a disability related industry for 16 years (SD = 14.21) on average. Fifty-four percent of participants (n = 26) responded they were active in disability advocacy. Nine participants (16.1%) were guardians to siblings with disabilities.

On average participants had low explicit disability prejudice scores (mean = 0.28, SD = 0.13). Yet, participants in this study had a moderate implicit preference for nondisabled people (mean = 0.53, SD = 0.44) on average. In fact, 83% (n = 39) of participants preferred nondisabled people, while 8% (n = 4) preferred disabled people, and 8% (n = 4) had no preference. Friedman (2017) explores the relationship between participants' explicit and implicit disability prejudice more in depth.

Disability Definitions from Siblings of Disabled People

In conjunction with the DA-IAT, participants were asked to define 'disability.' Their responses were coded using thematic analysis (Braun and Clarke, 2006), making use of the guiding theoretical concepts mentioned above. Several strong themes were present, including disability as preventing or slowing action, atypical function, lack of independence, socially constructed, and as general difference. These themes, taken together, reveal complex understandings of disability that relate to one another.

For instance, disability as preventing or slowing action is closely aligned with atypical function. The focus on action as an immutable good typifies an aspect of Siebers' (2008) ideology of ability in that disabled bodies are required to fit into broad social parameters of functioning, while the difference in mode of action is often ascribed to an individual difference in functioning. One participant defined disability as "someone who is incapable of doing something for themselves or has limited access to do something" while another participant identified disability as "something that makes you do things differently than the general population." These two exemplars relate to each other in that while the first emphasizes that being incapable of doing something for oneself is the exception, and therefore worthy of mention, the second example reinforces that the individual difference is made significant by the backdrop of what everyone else is doing. Disability is thus recognizable as that which makes one different from others, but applying the ideology of ability, it is clear that the referent for disability is the nondisabled "norm" or normate (Garland-Thomson, 1997). Defining disability, then, must draw upon the logics of other "deep structures" such as race, class, and gender.

Meanings of disability, as depicted within the present study, are not only represented by categorizations. The theme lack of independence emphasizes the perception that disability is related to reliance on others. This theme shifts the focus from a function-oriented focus to a relationally-centered definition. Disability, seen through the lens of dependency, forms the basis for relationships that may emerge from necessity, at least identified by these respondents, but it is a relationship nonetheless. As Eva Kittay (1999) reminds us, "interdependence begins with dependence" (p. xii). Disability as a relationship, however, is not limited to individuals, but rather serves as way to think about how disability functions as a social structure. The relationality of disability means that the presence of impairment does not only have meaningful effects on a disabled body, but on other bodies (Kafer, 2013). For instance, children who do not meet 'normal' height requirements are impacted by how they do or do not fit into compulsory norms (Kafer, 2013). Disability, then, is both embodied, but also spills over (Shildrick, 1997) into other bodies, the space between bodies, and the things that connect or separate bodies. The relationality of disability also, however, holds promise for further theorizing how particular roles related to disabled people might be distinctive and how those roles might illuminate something about how disability operates as a social structure (Thomas, 2004). For instance, even if parents and siblings of disabled people do not themselves identify as disabled, their particular knowledges, perspectives, attitudes, and understandings about disability may offer something to Disability Studies.

In addition to Kafer's work, Meltzer and Kramer (2016) and Mauldin (2014; 2016) have explored how conceptualizations of disability may be extended by focusing on relational spaces as sites where we might learn something about how the concepts of disability and ableism are socially and culturally constructed. Mauldin (2014), for instance, demonstrates how in the context of pediatric cochlear implant therapy, deafness is recast from sensory deficit to neurological issue, placing further responsibility onto mothers. These mothers, in response to this turn, come to view their children's deafness in complex ways, showing how technology, disability and familial relationships intersect and act upon one another. Such complexity, identified within our study as understanding disability as a social construction and general difference, may offer insight into how to diminish the negative effects of ableism. Close relationality to disability may serve to interrupt surface-level tropes of disabled people that so often find their way into how individuals value disability.

While Kafer's (2013) emphasis on the relational nature of disability is useful, equally important is her emphasis on the political nature of disability. For Kafer (2013), disability has to do with how certain relationships are ordered. Some respondents identified disability as socially constructed, including one participant who said disability is "a consequence of a society's lack of accommodations for people who are different." Another participant said that disability is "differences in the body or mind that meet socially constructed criteria to be deemed a disorder or disability. It is socially undesirable…" This definition resonates with the theme of disability as socially constructed but also invokes the theme of disability as a general difference . The theme of general difference acknowledges that disability is distinctive in some ways, but does not necessarily connote negative or deficit-based ideas about disability. In addition, both themes of socially constructed and general difference recognize that disability has been built and reproduced in certain ways. This connects with other themes including the emphasis on preventing or slowing action and atypical function. Put side by side, the emphasis on the 'negative' aspects of disability are often recognized as being a part of the larger social forces at work sustaining certain meanings about disability.

Relationships Between Disability Definitions and Disability Attitudes

In order to explore how definitions can relate to disability attitudes, binary logistic regressions were run to determine what definitions of disability (i.e., preventing/slowing actions/tasks; lacking independence; challenges and difficulty; general difference; social construction; atypical function; limitation; impairment; lack of participation) related to implicit disability attitudes when combined with explicit disability attitudes. Defining disability as lacking independence, impairment, and in relation to the norm, and explicit attitudes significantly predicted implicit attitudes (F(4, 42) = 2.63, p = .048, R² = .20). This finding suggests peoples' unconscious (implicit) disability attitudes can be predicted by their conscious (explicit) disability attitudes, and their understandings of disability as lacking independence, impairment, and/or in relation to the norm. Figure 1 shows expected implicit prejudice based on the participants' explicit prejudice and these definitions of disability.

According to this model, those who define disability in relation to the norm are expected to have higher implicit disability prejudice than those who define disability as lacking independence, or impairment. However, most people are still expected to implicitly prefer nondisabled people for all three of these themes.

No other definition themes from this study significantly related to implicit prejudice when combined with explicit prejudice.

Factors that Impact Meanings of Disability

To further tease out the relationship between understandings of and attitudes towards disability one aim of this study was to examine factors that impact how participants define disability. As such, regression models were run between the qualitative themes (i.e., preventing/slowing actions/tasks; lacking independence; challenges and difficulty; general difference; social construction; atypical function; limitation; impairment; lack of participation) and factors including political orientation, disability identity, participant in disability advocacy, etc. A binary logistic regression between years worked in a disability industry (zero for not applicable) significantly predicted defining disability as preventing/slowing action/tasks (-2LL = 54.13, Chi-square (1) = 7.39, p = 0.007. 55.3% of the cases are correctly predicted in the current data set.) Univariate analyses indicated that those who have not worked in a disability industry were significantly more likely to define disability as preventing/slowing action/tasks than those who have worked in a disability industry. According to the model, as the number of years worked in a disability industry increases, the odds of defining disability as preventing/slowing actions/tasks decreases. Figure 2 details the odds of defining disability in this way depending on if and how long people have worked in a disability industry.

Years worked in a disability industry also significantly predicted defining disability as a general difference in another binary logistic regression model (-2LL = 41.95, Chi-square (1) = 9.20, p = 0.002. 83% of the cases are correctly predicted in the current data set.) Univariate analyses indicated those who have not worked in a disability industry were significantly less likely to define disability as a general difference than those who have worked in a disability industry. As the number of years worked in a disability industry increases the odds of defining disability as a general difference increases. See figure 2.

All other factors (i.e., political orientation, disability status, participation in disability advocacy, gender, guardianship) were not significant with any other qualitative themes.