This article corrects:

  1. A new symptom model for autism cross-validated in an independent sample Volume 49, Issue 8, 809–816, Article first published online: 19 May 2008

In the article of Boomsma, A., van Lang, N.D.J., de Jonge, M.V., de Bildt, A.A., van Engeland, H., & Minderaa, R.B. (2008). A new symptom model for autism cross-validated in an independent sample. Journal of Child Psychology and Psychiatry, 49, 809–816, to our regret, an error was made in the analyses.

In the analyses, items of the original research version of the Autism Diagnostic Interview Revised (ADI-R) that apply to ratings of the current status for youth 10 years and older were included as well as the youth in that age group (= 356). These items and youth should have been deleted according to the scoring rules, and therefore analyses were redone, resulting in a new Table 2 (= 161). Replacement tables for Tables 3 and 4 were not possible given the reduced N.

The corrections for Table 2. on page 4: Goodness-of-fit values of Model B2 with data from Utrecht and Groningen of children aged 4–9 years (= 161; ADI-R current status)

 Global fit measuresMLM-based statistics
ADI-R current status
Model B25182.0.00484.

In regard to the text on page 4, column two, under Cross-validation, Model B2, the following correction should replace the current text: The size of standardized factor loadings of Model B2 with ADI-R current ratings is good to excellent (all indicators ≥.64 in F1; ≥.42 in F2; and ≥.74 in F3). The estimated standardized correlations between the factors are r(F1, F2) = .38, r(F1, F3) = .88, and r(F2, F3) = .33. Because the standardized correlation between F1 and F3 is quite large (.88), the model fit of a two-factor model, where factor (F3) impaired play skills was incorporated in factor (F1) impaired social communication, was also calculated. Although the goodness-of-fit values of this two-factor model were also not bad (df = 53; χ² = 91.3; = .001; CFI = .94; TLI = .92; AIC = 6787; RMSEA = .07; SRMR = .06), the appropriate chi-square difference test indicated a significantly better fit (Δχ² = 7.2; Δdf = 2, = .03) for the three-factor model as compared with that of the two-factor model.

These errors do not result in any changes in the conclusions drawn in the article.