The quantitative study of linguistic variables has been an integral part of sociolinguistic research since the mid twentieth century, but it was only recently that the use of attitudinal data as potential quantitative correlates of language variation has been advanced, thereby uniting the agentive focus of recent variationist scholarship with quantifiable attitudinal findings. Based on the fact that conation is one of the components of attitudes, this chapter demonstrates how variable analysis can profoundly enrich our knowledge of language attitudes. The key strengths of using variable analysis (e.g. high levels of statistical rigour) are discussed at length, as well as the potential limitations and complications (e.g. how to align ‘big’ attitudinal data with social constructivist frameworks). The chapter discusses practical issues of research design, such as the tasks by means of which phonetic, morphosyntactic, and lexical variables can be elicited. Analytical approaches that are suitable to the analysis of both variationist and attitudinal data are addressed, with an emphasis on mixed-effects linear regression modelling. To illustrate the key points pertaining to variable analysis as a means of investigating language attitudes, the chapter concludes with a case study of Catalan as spoken in southern France, in the region of Northern Catalonia.