The term “Matthew effect” is frequently used in relation to positive correlations seen between baseline proficiency level and growth, and is an effect frequently seen in the assessment of proficiencies like reading comprehension. Such effects play a role in theories of reading comprehension, as well as the development of interventions that target low proficiency readers. However, the observation of such effects is often questioned from a methodological perspective due to their heavy reliance on interval level properties of score metrics. Little attention has been given to how or why, from a measurement perspective, score metrics might be distorted to produce such effects. In this paper we demonstrate how such correlations can be produced by negative item characteristic curve (ICC) asymmetry. We provide a psychological justification for why such systematic asymmetries may be present and show why true asymmetries can be missed when fitting symmetric IRT models. We ultimately demonstrate the potential for realistic forms of ICC asymmetry to produce correlations of the magnitude reported in literature despite no correlation being present between baseline proficiency and growth.