Random forest as an imputation method for education and psychology research: its impact on item fit and difficulty of the Rasch model

2021

Molecular Autism

Background Alexithymia, a personality trait characterized by difficulties interpreting one’s own emotional states, is commonly elevated in autistic adults, and a growing body of literature suggests that this trait underlies a number of cognitive and emotional differences previously attributed to autism, such as difficulties in facial emotion recognition and reduced empathy. Although questionnaires such as the twenty-item Toronto Alexithymia Scale (TAS-20) are frequently used to measure alexithymia in the autistic population, few studies have attempted to determine the psychometric properties of these questionnaires in autistic adults, including whether differential item functioning (I-DIF) exists between autistic and general population adults. Methods We conducted an in-depth psychometric analysis of the TAS-20 in a large sample of 743 verbal autistic adults recruited from the Simons Foundation SPARK participant pool and 721 general population controls enrolled in a large international psychological study (the Human Penguin Project). The factor structure of the TAS-20 was examined using confirmatory factor analysis, and item response theory was used to further refine the scale based on local model misfit and I-DIF between the groups. Correlations between alexithymia and other clinical outcomes such as autistic traits, anxiety, and quality-of-life were used to assess the nomological validity of the revised alexithymia scale in the SPARK sample. Results The TAS-20 did not exhibit adequate global model fit in either the autistic or general population samples. Empirically driven item reduction was undertaken, resulting in an eight-item unidimensional scale (TAS-8) with sound psychometric properties and practically ignorable I-DIF between diagnostic groups. Correlational analyses indicated that TAS-8 scores meaningfully predict autistic trait levels, anxiety and depression symptoms, and quality of life, even after controlling for trait neuroticism. Limitations Limitations of the current study include a sample of autistic adults that was overwhelmingly female, later-diagnosed, and well-educated; clinical and control groups drawn from different studies with variable measures; and an inability to test several other important psychometric characteristics of the TAS-8, including sensitivity to change and I-DIF across multiple administrations. Conclusions These results indicate the potential of the TAS-8 as a psychometrically robust tool to measure alexithymia in both autistic and non-autistic adults. A free online score calculator has been created to facilitate the use of norm-referenced TAS-8 latent trait scores in research applications (available at http://asdmeasures.shinyapps.io/TAS8_Score).

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

RETRACTED ARTICLE: Improving the measurement of alexithymia in autistic adults: a psychometric investigation and refinement of the twenty-item Toronto Alexithymia Scale

2021

Molecular Autism

“…As noted earlier, neurotypical participants in the HPP sample lled out the TAS-16, a version of the TAS-20 in which four problematic items have been removed from the scale (64). However, as we wished to compare total scores from the TAS-20 between HPP and SPARK samples, we conducted single imputation for missing items in both groups using a random-forest algorithm implemented in the R missForest package (93)(94)(95). Such item-level imputation allowed for us to approximate the TAS-20 score distribution of the HPP participants, including the proportion of individuals exceeding the "high alexithymia" cutoff of 61.…”

Section: Measures Toronto Alexithymia Scale (Tas)mentioning

confidence: 99%

“…We t the model using a diagonally weighted least squares estimator ( 109) with a mean-and variance-corrected test statistic (i.e., "WLSMV" estimation), as implemented in the R package lavaan (110). Very few of the item responses in our dataset contained missing values (0.16% missing item responses in the SPARK sample, no missing TAS-16 data in HPP sample), and missing values were singly imputed using missForest (93)(94)(95).…”

Section: Con Rmatory Factor Analysis and Model-based Bifactor Coe Cientsmentioning

confidence: 99%

WITHDRAWN: Improving the Measurement of Alexithymia in Autistic Adults: A Psychometric Investigation and Refinement of the Twenty-item Toronto Alexithymia Scale

2021

Preprint

Background: Alexithymia, a personality trait characterized by difficulties interpreting one’s own emotional states, is commonly elevated in autistic adults, and a growing body of literature suggests that this trait underlies a number of cognitive and emotional differences previously attributed to autism, such as difficulties in facial emotion recognition and reduced empathy. Although questionnaires such as the twenty-item Toronto Alexithymia Scale (TAS-20) are frequently used to measure alexithymia in the autistic population, few studies have attempted to determine the psychometric properties of these questionnaires in autistic adults, including whether differential item functioning (I-DIF) exists between autistic and general population adults. Methods: We conducted an in-depth psychometric analysis of the TAS-20 in a large sample of 743 verbal autistic adults recruited from the Simons Foundation SPARK participant pool and 721 general population controls enrolled in a large international psychological study (the Human Penguin Project). The factor structure of the TAS-20 was examined using confirmatory factor analysis, and item response theory was used to further refine the scale based on local model misfit and I-DIF between the groups. Correlations between alexithymia and other clinical outcomes such as autistic traits, anxiety, and quality-of-life were used to assess the nomological validity of the revised alexithymia scale in the SPARK sample. Results: The TAS-20 did not exhibit adequate global model fit in either the autistic or general population samples. Empirically driven item reduction was undertaken, resulting in an eight-item unidimensional scale (TAS-8) with sound psychometric properties and practically ignorable I-DIF between diagnostic groups. Correlational analyses indicated that TAS-8 scores meaningfully predict autistic trait levels, anxiety and depression symptoms, and quality of life, even after controlling for trait neuroticism. Limitations: Limitations of the current study include a sample of autistic adults that was overwhelmingly female, later-diagnosed, and well-educated; clinical and control groups drawn from different studies with variable measures; and an inability to test several other important psychometric characteristics of the TAS-8, including sensitivity to change and I-DIF across multiple administrations. Conclusions: These results indicate the potential of the TAS-8 as a psychometrically robust tool to measure alexithymia in both autistic and non-autistic adults. A free online score calculator has been created to facilitate the use of norm-referenced TAS-8 latent trait scores in research applications (available at http://asdmeasures.shinyapps.io/TAS8_Score).

“…As noted earlier, neurotypical participants in the HPP sample filled out the TAS-16, a version of the TAS-20 in which four problematic items have been removed from the scale [65]. However, as we wished to compare total scores from the TAS-20 between HPP and SPARK samples, we conducted single imputation for missing items in both groups using a random-forest algorithm implemented in the R missForest package [96][97][98]. Such item-level imputation allowed for us to approximate the TAS-20 score distribution of the HPP participants, including the proportion of individuals exceeding the "high alexithymia" cutoff of 61.…”

Section: Toronto Alexithymia Scale (Tas)mentioning

confidence: 99%

WITHDRAWN: Improving the Measurement of Alexithymia in Autistic Adults: A Psychometric Investigation and Refinement of the Twenty-item Toronto Alexithymia Scale

2020

Preprint

Background: Alexithymia, a personality trait characterized by difficulties interpreting emotional states, is commonly elevated in autistic adults, and a growing body of literature suggests that this trait underlies a number of cognitive and emotional differences previously attributed to autism. Although questionnaires such as the twenty-item Toronto Alexithymia Scale (TAS-20) are frequently used to measure alexithymia in the autistic population, there have been few studies attempting to determine the psychometric properties of these questionnaires in autistic adults, including whether differential item functioning (I-DIF) exists between autistic and general population adults. Methods: We conducted an in-depth psychometric analysis of the TAS-20 in a large sample of 743 cognitively able autistic adults recruited from the Simons Foundation SPARK participant pool and 721 general population controls enrolled in a large international psychological study. The factor structure of the TAS-20 was examined using confirmatory factor analysis, and item response theory was used to further refine the scale based on local model misfit and I-DIF between the groups. Correlations between alexithymia and other clinical outcomes were used to assess the nomological validity of the revised alexithymia scale in the SPARK sample. Results: The TAS-20 did not exhibit adequate model fit in either the autistic or general population samples. Empirically driven item reduction was undertaken, resulting in an eight-item unidimensional scale (TAS-8) with sound psychometric properties and practically ignorable I-DIF between diagnostic groups. Correlational analyses indicated that TAS-8 scores meaningfully predict autistic trait levels, anxiety and depression symptoms, and quality of life, even after controlling for trait neuroticism. Limitations: Limitations of the current study include a sample of autistic adults that was overwhelmingly female, later-diagnosed, and well-educated; clinical and control groups drawn from different studies with variable measures; and an inability to test several other important psychometric characteristics of the TAS-8, including sensitivity to change and I-DIF across multiple administrations. Conclusions: These results indicate the potential of the TAS-8 as a psychometrically robust tool to measure alexithymia in both autistic and non-autistic adults. A free online score calculator has been created to facilitate the use of norm-referenced TAS-8 latent trait scores in research applications (available at http://asdmeasures.shinyapps.io/TAS8_Score).