Agile and conventional methodologies: an empirical investigation of their impact on software quality parameters

Abstract

The advent of agile methodologies has brought about an illuminating debate in Software Engineering, particularly with regard to software quality. Some studies have reported that agile methodologies do improve software quality when compared to traditional methodologies; other studies have been inconclusive or contradictory, while others have argued that empirical evidence is limited. This study sought to investigate the correlation between agile methodologies when compared to traditional methodologies for selected software quality parameters. The research design was causal comparative, as well as correlational. The approach was quantitative, using a survey as the data collection method. SPSS was used to conduct descriptive and correlational analysis for 106 responses received. The main findings were that there was a statistically significant relation between traditional methodology use and ease of system testing (p=0.014); a statistically significant relation between traditional methodology use and timeliness (p=0.02); a statistically significant relation between software quality standards used and ease of system testing (p=0.017); a statistically significant relation between active stakeholder participation on projects and ease of system interactivity (p=0.047); and a statistically significant relation between mandatory workshop interactivity (p=0.047); and a statistically significant relation between mandatory workshop attendance or training and ease of system navigation (p=0.031).Claims that agile methodology use leads to improved software quality for selected quality parameters could not be empirically validated. The association between most of the selected software quality criteria in relation to methodology use in general was not apparent. Agile methodologies are suitable in small environments. Scrum was the most widely used agile methodology by far. The popularity and adoption state of XP showed a significantly decreasing trend. Traditional and agile methodologies combined are being used (47%) more than any other methodology. Agile methodology use (28%) surpassed traditional methodology use (19%). A suitable consensus definition for agile methodologies did not emerge from the data collected. The most suitable project life cycle model was evolutionary, incremental and iterative. ‘Other’ methodologies, meaning customised agile or SDLC, are suitable, as the environment becomes increasingly large and complex. Only 13% of organisations surveyed have an agile experience of six years and beyond. Based on these findings and gaps in the literature, implications and recommendations for further research areas are proposed, where the findings and contributions of this study are found to be relevant to practice for application and to academia for further research