The scholastic cheating of early adolescents: from exploratory research towards a structural model. A pilot study.

The aim of this work is to investigate, which behaviour at school is perceived by early adolescents (age average 14-16 years) as cheating, whether it is possible to classify this behaviour into specific types (factors) and whether these can be linked with individual student variables (sex, GPA, number of missed classes, student goal orientation, worry, self-efficacy, and achievement value), as well as with the contextual variables (parents goal orientation, teachers goal orientation, satisfaction with the school, neutralization, peer cheating behaviour, and cheating punishment). The obtained data were statistically evaluated (N = 401). Using exploratory factor analysis based on students self-reporting the frequency of the behaviour classified as cheating, two types of dishonest behaviour were extracted: cheating (copying, hinting) and falsification (forgery of signatures, absences). Two student factors, school approach (motivation) and to school avoidance (amotivation) orientation, and two context factors, teachers and parents goal orientation, were extracted as predictors of cheating. Results showed that neither parents nor cheating punishment had significant influence on cheating and falsification. The structural model confirmed that a positive relationship with teachers goal orientation and GPA had the highest influence on falsification. Student avoidance to school, student approach to school, and teachers goal orientation had the highest influence on cheating. Cheating and falsification were well correlated. The values of regression coefficients remained similar after removing the parent factor from the model. The structural equation model explaining the relationship of latent endogenous cheating and falsification variables with the three latent exogenous variables (avoidance to school, approach to school and teachers goal orientation) and the three manifest variables (sex, GPA and peer cheating behaviour), fitted the data well. The model explained 42% of cheating variance and 39% of falsification varianc