A comparison of content analysis usage and text mining in CSR corporate disclosure

This paper investigates content analysis and text mining as two different research techniques largely used by scholars to perform the text analysis of company social and environmental reports. Its aim is to demonstrate that these techniques are not so irreconcilable as one might suppose, but can be applied to the same problem being addressed in certain circumstances. The paper starts with a presentation of the two research techniques, providing information about their origin, diffusion and fields of application in academia. It examines them with respect to the assumption each one holds about the nature of knowledge and continues with a discussion of the elements that display the differences and similarities between them. Subsequently, the paper presents an empirical application of the two techniques to the same research problem: the identification of possible changes in the amount of company disclosure after an industrial disaster. The focus is on a company’s communication of economic, social and environmental goals and impacts usually included in sustainability reports. The two techniques are applied to the same set of company reports published by four large multinationals that went through the industrial disaster characterised by strong negative social and environmental consequences. Results from the trend analysis obtained by the application of the two techniques indicate that they are not such irreconcilable methods, but they may lead to different conclusions about a company’s behaviour in trying to restore its corporate reputation damaged by the disaster. Thus, the two techniques should not be used to crosscheck results, although they provide similar output data.