Improving Certainty about Uncertainty in ASD? Confirmatory Factor Analysis of the Intolerance of Uncertainty Scale (12) in ASD Adults

Background: Intolerance of Uncertainty (IU) impairs the capacity to function in ambiguous contexts including situations of unpredictable change. IU is commonly observed in people who have high levels of anxiety, including many people diagnosed with autism spectrum disorder (ASD). The 12-item Intolerance of Uncertainty Scale has been validated with good psychometric properties on normative undergraduate samples. However, people with autism seem to experience anxiety differently than non-ASD samples and new autism-specific measures of anxiety are appearing. We investigated whether IU might similarly differ in ASD samples by examining the factor structure of the IUS-12 in adults diagnosed with ASD.  

Objectives: This study examined the established two-factor IUS-12 using confirmatory factor analysis. These two subscales include: the Prospective IU scale, which reflects a desire for predictability, anxiety about future uncertainties, and active engagement in seeking to increase certainty; and the Inhibitory IU scale, a measure of avoidance and paralysis in response to present uncertainties. We examined evidence for this structure as well as the overall total IUS total score.

Methods: Participants for this study consisted of 74 adults diagnosed with Autism Spectrum Disorder (ASD) recruited from two separate sites in the USA and UK. Multiple CFA analyses were performed using STATA 14 to assess the construct validity of the IUS. We performed hierarchical (second-order) CFA and a bi-factor CFA using the maximum-likelihood estimator to examine the relationship between the total IUS scores and its proposed subdomains. In the event that these models did not converge or fit properly, we then performed first-order CFAs for the total score, covaried sub-scale scores, and individual CFA for each sub-scale.

Results: CFA analyses of the hierarchal and bi-factor models did not converge after 10,000 iterations, indicting that neither model was appropriate given the available data. Though the total score CFA and the covaried sub-scales CFA did converge properly, both models showed inadequate fit. Individual sub-scales CFA showed that the Inhibitory sub-scale fit properly with no modifications. The Prospective sub-scale fit properly after dropping one item (#4; “One should always look ahead so as to avoid surprises”) and covarying one set of error terms. Many participants commented that it was difficult to respond to the questions because a) they included abstract rather than concrete scenarios and b) the respondents’ degree of uncertainty varies across situations.  

Conclusions: Several recent studies have shown that IU is a useful construct for mediating anxiety and repetitive behaviors in ASD samples. However, the current structure of the IUS-12 does not provide a good overall fit for this pooled sample, though the brief Inhibitory subscale works better. As with ongoing work to create autism-specific measures of anxiety, more refinement of the IU construct and measures may be helpful to understand what aspects of uncertainty are most relevant in ASD.