Social Economic Status and the M-CHAT: Unreliable Results for Mothers with Low Education Level

Background: Children with autism from rural areas and disadvantaged backgrounds (low socioeconomic status; SES) tend to receive an ASD diagnosis at later ages (Rhoades, Scarpa, & Salley, 2008). This delay may increase the average age at which intervention is received (Mandell, Listerud, Levy, & Pinto-Martin, 2002; Mandell, Noval, & Zubritzky, 2005). The Modified Checklist for Autism in Toddlers (M-CHAT) remains the most well-studied, validated, and practical tool for early screening in the primary care setting (Robins, Fein, Barton, & Green, 2001). However, the psychometric properties of the M-CHAT need to be further examined in diverse samples in order to determine if this measure needs to be adapted accordingly.

Objectives: This study investigated internal reliability of the M-CHAT completed by mothers of toddlers from a rural area in Southwest Virginia. The sample contained both English and Spanish-speaking mothers. 

Methods: Data were collected from the mothers of 317 toddlers between 17 and 30 months of age. The M-CHAT was completed during children’s 18- or 24- month-well visit. The mother’s report of her highest academic achievement was used as a proxy for SES. For this study, three academic achievement categories were created: (1) less than 12^th grade/GED completed (n = 90), (2) 12^th/GED completed (n = 143), and (3) more than 12^th/GED completed (n = 84).

Results: The M-CHAT’s internal reliability differed based on mothers’ SES. Mothers who had less than or had completed a 12^th grade education/GED had Cronbach's alphas of .428 and .225, respectively, whereas mothers who had greater than a 12^th grade/GED education obtained a Cronbach's alpha of .729. Similarly, these education groups significantly differed on how many items they failed, F(2, 313) = 3.701, p=.026, with less than a 12^th grade education/GED failing the most items, than completed 12^th grade/GED, and mothers who completed more than a 12^th grade/GED education failed the least number of items. Specifically, mothers with less than a 12^th grade education/GED tended to fail items 17, and 22, whereas mothers with a 12^th grade/GED education or higher tended to fail item 15, c2(1, N =  315) = 11.141, p=.004.

Conclusions: This study had two main findings. First, the psychometric properties of the M-CHAT appear to be unreliable when assessing toddlers of mothers with a 12-grade education or less. That is, these results demonstrate that the M-CHAT screening was not acceptable or consistent with what is considered adequate/acceptable internal reliability (alpha > 70). Second, maternal education levels also predicted response patterns of certain items on the M-CHAT. Specifically, toddlers with mothers with lower educational level were more likely to be perceived as developing more atypically. Several factors could have contributed to these unreliable results, including mothers’ lack of understanding of typical/atypical behavior, motivation to respond accurately, or poorly constructed questions (e.g., vague language). Including a questionnaire with pictures depicting behaviors being assessed might help mothers provide more informed responses. In any case, work needs to be done to improve reliability of the M-CHAT for use among less educated families during early pediatric screenings.