Evidence suggests that nasality carries social information in the UK, but existing sociolinguistic studies are mostly impressionistic (Laver 1972, Trudgill 1974). The majority of existing phonetic methods for the analysis of nasality range from inconvenient to intrusive. Recently, promising acoustic methods have been developed for the quantification of nasality, offering potential for middle ground sociophonetic research into nasality. This paper evaluates one of those approaches, the NAF method (Nasalization from Acoustic Features; Carignan 2021), with application to sociophonetic data in mind. NAF involves predicting the degree of nasality from principal component regression on a range of acoustic correlates of nasality and mel-frequency cepstral coefficients. This method was found to perform well, despite the differences in data structure between the original implementation and this study. It also raised some questions about model interpretability and stability, which were addressed with additional model refitting.