People with intellectual and developmental disabilities (IDD) may find it difficult to answer survey questions for a number of reasons, including cognitive capacity and limitations to verbal communication (Scott & Havercamp, 2018). Because of these difficulties, research with people with IDD frequently allows for proxy-responses to survey questions (Scott & Havercamp, 2018). Despite its prevalence in research, psychometric analyses of proxy responses in comparison to self-respondents are lacking. These analyses are particularly needed since some literature has found that proxy-responses may not be valid, especially for subjective constructs (Scott & Havercamp, 2018; Tournier et al., 2020).
Methods
This study used data from the Virginia 2017-2018 and 2018-2019 National Core Indicators In-Person Survey (NCI-IPs). The NCI-IPS is a face-to-face survey of adults with IDD who use at least one state-funded service in addition to case management. Respondents are randomly selected so as to be representative of each state’s service users. Virginia randomly selects approximately 800 people each year, for a total combined sample of 1620 for this study.
Some questions in the NCI-IPS may be answered by either the individual with IDD or a proxy who knows them well. These variables are accompanied by a question about who responded, which were used to classify respondents into one of three response categories: self-response only, proxy-response only, or mixed-response. These three response categories were used in subsequent analyses. A measurement model for personal opportunities consisting of scales for rights, choice, and community participation was developed and validated in a previous study (Authors, 2021). We ran a multigroup confirmatory factor analysis (MG-CFA) to test for measurement invariance between the three response categories.
Results
Bivariate analysis revealed significant differences between the three response categories. Respondents who answered via proxy were more likely to have severe or profound intellectual disabilities, autism, and behavioral support needs. Interestingly, respondents who answered via proxy were also more likely to communicate verbally and ambulate independently. Respondents who answered via proxy were more likely to live in medium (4-6 person) group homes or with family and were more likely to have a legal guardian. Despite these differences, multigroup CFA established configural (X2 = 140.55, p < .001), metric (X2 = 178.88, p < .001), and scalar invariance (X2 = 199.69, p < .001).
Implications
Measurement invariance suggests that people in all three response categories interpret and respond to questions about personal opportunities in similar ways and that, by extension, meaningful comparisons can be made between personal opportunities reported by self- and proxy-respondents. This finding has important implications for research with people with IDD. The NCI-IPS offers important insights into developing research questions, training interviewers, and selecting appropriate proxies for survey research.