Introduction
Despite the impact non-autistic people have on the lives of autistic people, their attitudes towards people with autism are poorly understood. This is an important line of enquiry, given that these attitudes are likely to underpin behaviour towards autistic people (Kraus, Reference Kraus1995). Strikingly, however, there is only one measure of attitudes to autism, that is, the Societal Attitudes towards Autism Scale (SATAS; Flood et al., Reference Flood, Bulgrin and Morgan2013), which appears to conflate attitudes, intentions, and behaviours towards autistic people. Accordingly, it is unclear if the SATAS has a unidimensional structure, given that some of its items concern affective dimensions of attitudes (e.g., “People with autism should not have children”) whereas others refer to behaviours (e.g., “I would be uncomfortable hugging a person with autism”). This left us wondering if well-established attitude measures unrelated to autism (e.g., Armitage et al., Reference Armitage, Armitage, Conner, Loach and Willetts1999; Haddock et al., Reference Haddock, Zanna and Esses1993) could be adapted to quantify attitudes towards autistic people.
Objective
The present research had two aims. First, using two datasets, we tested whether the SATAS (Flood et al., Reference Flood, Bulgrin and Morgan2013) is unidimensional to address outstanding questions about its factor structure. Second, we developed 1- and 4-item measure of attitudes towards autistic people and explored if they had similar predictive validity to the 16-item SATAS. All things being equal, shorter measures are preferable as they reduce participant fatigue and attrition, which are otherwise a threat to reliability and validity (Rammstedt & Beierlein, Reference Rammstedt and Beierlein2014). All materials, code to reproduce the analyses, detailed results, and the data are openly accessible at https://osf.io/mf5bg
Methods
Participants were 135 undergraduate students, of which six failed two attention checks and were excluded (M age = 19.86, SD = 2.99; 103 women). They completed an online survey in exchange for course credits. Participants completed the SATAS (Flood et al., Reference Flood, Bulgrin and Morgan2013) which had satisfactory internal reliability (α = 0.76); a novel 1-item evaluative thermometer (based on Haddock et al., Reference Haddock, Zanna and Esses1993) ranging between 0 and 100 to indicate their “overall evaluation of an autistic person”; and an adapted version of the 4-item attitude scale originally used by Armitage et al. (Reference Armitage, Armitage, Conner, Loach and Willetts1999). The items all read “I see autistic people in general as...” and were answered on a scale ranging from −3 (bad, unfavourable, negative, unsatisfactory) to +3 (good, favourable, positive, satisfactory; α = 0.92). The order of the three scales was randomised. Finally, we measured intentions to engage with autistic people using a single item on a 1 (certainly not) to 7 (most certainly) point scale.
Results
A confirmatory factor analysis with robust standard errors and a Satorra-Bentler scaled test statistic revealed that the SATAS was not unidimensional, CFI = .78, RMSEA = .065, that is, it was not measuring one construct. Instead, an exploratory factor analysis suggested the presence of three factors, similar to the pattern of results from a re-analysis of Flood et al.’s (Reference Flood, Bulgrin and Morgan2013) data (n = 475; see Supplemental Materials). In contrast, our 4-item measure was unidimensional (CFI = 1.00, RMSEA = 0.00). The SATAS correlated positively with our attitude measures (Table 1) and all three measures were similarly predictive of the intention to engage with autistic people.
Correlations between Attitudes towards and Intentions to Engage with Autistic People.
Note. *p < .05, **p < .01, ***p < .001, 95%-CIs in brackets.
Discussion
We found that a 16-item measure of attitudes towards autistic people (Flood et al., Reference Flood, Bulgrin and Morgan2013) is not unidimensional. Rather, it has several factors, and is therefore unlikely to selectively measure attitudes towards autistic people. It likely conflates measurement of attitudes, intentions, and behaviours towards autistic people, whereas our 1-item Attitudes towards Autism (ATA-1) measure is more parsimonious and robust. It also has greater practical utility and no worse predictive validity than longer measures, including our 4-item measure that was included in the present study to compare against the 1-item scale. Overall, we suggest that the ATA-1 should be used in future attitude-related autism research.
Conclusion and Future Directions
In conclusion, attitudes towards autistic people can be measured more precisely with a single item than a scale with a complex factor structure. A question on engagement with autistic people also validated our attitude measure and permitted comparison with the SATAS validation (Flood et al., Reference Flood, Bulgrin and Morgan2013). It is possible that this question may not be measuring intended engagement with autistic people and instead tapping into general intentions to engage in volunteering activities. Future research will therefore benefit from testing if the ATA-1 is predictive of actual rather than intended engagement with autistic people. Our findings also need to be replicated in larger and non-student samples (Hanel & Vione, Reference Hanel and Vione2016) and, if we are going to develop more comprehensive attitudes to autism measures, they should be grounded in theoretical and psychometric research on both autism and attitude formation (Breckler, Reference Breckler1984; Fazio, Reference Fazio1990; Shah et al., Reference Shah, Livingston, Callan and Player2019). Nonetheless, the current study represents an important development in measurement of attitudes towards autistic people, which will benefit future research on understanding and changing (negative) attitudes towards people with autism.
Acknowledgements
We thank Betsy Morgan for sharing the raw data that was described in their paper (Flood et al., Reference Flood, Bulgrin and Morgan2013).
We also thank Alanna Shand for assistance with checking analyses and proof reading.
Conflict of Interest
The authors declare none.
Author’s note.
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Author Contributions
PH and PS conceived and designed the study, PH conducted data collection, and PH and PS performed statistical analyses and wrote the article.
Data Availability Statement
All materials, R code to reproduce the analyses, detailed results, and the data are openly accessible at https://osf.io/mf5bg/?view_only=9579bd4c1aad4fd995a91b31ac3afae5
Ethics statement.
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Supplementary Materials
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/exp.2020.36.
Open access
Comments
Comments to the Author: This is a well written article that is clearly situated. The results are useful to the Autism research community, and is a welcome methodical improvement on previous measures quantifying attitudes towards autistic people, mostly in terms of saving time. Some minor comments - 1) Were the 3 tests presented in the same order and would counterbalancing have made a difference? 2) Sample size is mentioned in the supplementary material, would be useful to have in the Discussion 3) Could you make clearer what the rationale was for the 4-item scale as they seem to be asking essentially the same question? 4) The engagement question at the end may not necessarily be tapping ‘intentions to engage’. Participant responses could be influenced by general perceptions of volunteering and time rather than their attitude towards autistic people. This can be explored in future extensions of the study, and mentioned in the limitations section.