The Confounding Effect of Class Size on the Validity of Object-Oriented Metrics

By K. El Emam, Saïda Benlarbi, Nishith Goel, Shesh N. Rai. IEEE Trans. Software Eng., 2001

459 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, Software

View Paper: https://www.semanticscholar.org/paper/eda92ba4a59676349f5954c2f529adc5ae608198

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

In Software

Citations in the corpus

How, and why, process metrics are better (2013, background): “Size is a very important metric by itself and there is a considerable body of evidence that size is highly correlated with most product metrics including McCabe and Halstead metrics [7, 9].”
Coverage is not strongly correlated with test suite effectiveness (2014, background): “For example, previous work suggests that the size of a class can affect the validity of object-oriented metrics [11].”
A large scale study of programming languages and code quality in github (2014, background): “Considering software quality, for example, there are a number of well-known inﬂuential factors, including source code size [11], the number of developers [36, 6], and age/maturity [16].”
The prediction of faulty classes using object-oriented design metrics (2001, methodology): “In constructing our models, we follow previous literature in that we do not present results for interactions nor higher order terms, for example, see (Basili et al., 1996; Benlarbi and Melo, 1999; Briand et al., 1997; Briand et al., 1998a; Briand et al., 1998b; Briand et al., 2000; El-Emam et al., 2”
An Exploratory Study of the Impact of Code Smells on Software Change-proneness (2009, background): “[29] showed that after controlling for the confounding effect of size, the correla tion between metrics and fault-proneness disappeared: man y metrics are correlated with size and, therefore, do not brin g more information to predict fault proneness.”
Empirical Analysis of Object-Oriented Design Metrics for Predicting High and Low Severity Faults (2006, background): “Previous research on OO design metrics has shown some of them to be useful for predicting the fault-proneness of classes [1], [2], [3], [4], [5], [6], [7], [8], [9], yet has also made the point that this usefulness depends upon their being valid whatever the degree of severity of the fault [6].”
Preliminary Guidelines for Empirical Research in Software Engineering (2002, unknown): “To ensure that the description of the research provides enough detail for other researchers and for practitioners.”
Application of neural networks for software quality prediction using object-oriented metrics (2003, background): “The confounding effect of class size is studied in [7].”
Empirical Validation of Three Software Metrics Suites to Predict Fault-Proneness of Object-Oriented Classes Developed Using Highly Iterative or Agile Software Development Processes (2007, background): “[7] are four additional metrics that have been the subject of validation studies in the past: NMO (measuring the number of inherited methods overridden by a subclass), NMA (measuring the number of methods added by a subclass), SIX (a combination of inherited metrics), and NPAVG (measuring the averag”
Don't touch my code!: examining the effects of ownership on software quality (2011, background): “Prior research has shown that when characteristics such as size are not considered, they may affect the validity of other software metrics [13].”
Empirical Analysis of CK Metrics for Object-Oriented Design Complexity: Implications for Software Defects (2003, background): “One of the few prior studies addressing defects in object-oriented software has also discussed the potential confounding effect of size on the relationship between design metrics and defectproneness [25].”
Using the Conceptual Cohesion of Classes for Fault Prediction in Object-Oriented Systems (2008, result): “In order to generalize the results, a large-scale evaluation similar to the one presented by Succi et al. [73] is required which should take into account software systems from different domains written in different programming languages and of varying class sizes in those systems [ 29 ].”