Attitude and motivation to work in autism rehabilitation operators: Algorithms based on reaction times

Assessing the true motivation and attitudes of rehabilitation professionals working in autism care is crucial for ensuring effective interventions and high-quality support. Traditional self-report tools are limited by social desirability bias, underscoring the need for implicit and computational approaches. This exploratory study aims to assess whether reaction-time-based implicit measures, combined with machine learning (ML), can identify latent demotivation and attitudinal inconsistency among rehabilitation professionals. Twenty-eight rehabilitation operators and twenty university students participated in the study. Students were assigned to two simulated profiles: satisfied (N=10) and dissatisfied (N=10). A custom computerized reaction-time task was developed in which participants were presented with statements and asked to respond "Yes" or "No" by pressing a key on the keyboard. Additionally, three paper-and-pencil tests were administered to assess work motivation, social desirability, and to collect sociodemographic information. The significant variables relating to latency and consistency obtained from the student data were used in the ML analysis (Naïve Bayes and Random Forest) to perform an exploratory classification of the rehabilitation professional sample. The computerized task, unlike the paper-and-pencil tests, showed statistically significant differences (p<.005) between the two profiles of the student sample. Based on the significant variables derived by the student sample, the exploratory model classified seven professionals as dissatisfied, achieving an accuracy of 78%. Latency-based implicit measures, when integrated with ML, offer a promising methodological framework for detecting hidden motivational patterns. Results are preliminary and should not be interpreted as diagnostic or predictive; further validation with larger, ecologically valid samples of professionals is required. • The assessment of work attitude is crucial in the field of autism rehabilitation • We explore the application of an algorithm based on implicit measurements • A computerised task identifies false declarations based on reaction times • Explicit self-report measures don’t distinguish satisfied vs dissatisfied profiles • Machine Learning algorithms classify dissatisfied operators with 78% accuracy