Is Manual Data Collection Hampered by the Presence of Inner Classes or Class Size?

Abstract

In this paper, we present an empirical study in which we hypothesize that the existence of Java ‘inner classes’ and class size are strong impediments to the data collector during manual data collection. We collected inner class and class size data from the classes of four Java open-source systems - first manually and then automatically (after the manual collection had finished) using a bespoke software tool. The data collected by the tool provided the benchmark against which errors and oversights in the manual data collection of these two features could be recorded. Results showed our initial hypotheses to be refuted – manual errors in data collection from the four Java systems arose not from the presence of inner classes or from class size but from variations in coding style, lack of standards, class layout and disparateness of class feature declarations.