This paper examines how Rasch analysis and Principal Components Analysis have been used complementarily in the evaluation of an astronomy course. Twelve dichotomous items were presented to 134 secondary school students in Greece. The items evaluated a STEM project, with emphasis on its Astronomy component. Nonlinear Principal Components analysis and Rasch analysis are compared regarding the issue of unidimensionality. It is argued that non-linear Principal Component Analysis offers safer criteria for suggesting unidimensionality but at the same time Rasch analysis provides more information regarding items’ difficulty and participants’ ability.