As nobody wanted to give me a review copy I went out and bought this book, and I’m glad I did. It is a good book – it is an important book – indeed I think every researcher (whether coming from a positivist or interpretivist standpoint) would benefit from reading it.
The first few chapters provide one of the best explanations of research that I have read – not least because it addresses one of the most problematic aspects of much positivist research, which is that unless your data is meaningful the statistics tell you nothing of value.
Shaffer unpacks the notion that all research is based on descriptions – beautifully explaining the differences between projectable and non-projectable properties, emic and etic observations, and thick and thin descriptions, as well as explaining bias (which exists for both positivists and interpretivists). If you are a researcher and are wondering what any of those things might be then you definitely need to read this book.
He explains the importance of structuring qualitative data systematically and consistently in to data tables – I’m not doing this substantial and critical aspect of the book justice. Those data tables are not only important if you are analysing the data manually, but also make it possible to automate part of the process of analysis. This makes it possible to analyse huge data sets that it would not be possible to manage manually. Again I am skipping over some really important aspects of how you do this.
Shaffer then explains how you can use statistical techniques to provide “warrants for claims of theoretical saturation" (p.391) – in effect to demonstrate the credibility of the qualitative analysis of (a subset) of the data.
The claim implicit in the title that the book squares the circle of qualitative (ethnographic) and quantitative approaches is based on reframing the positivist notion of generalisability. Rather than claiming that the statistical analysis permits you to generalise to the wider population (“beyond the data at hand” (p.392)), Shaffer argues that it allows you to generalise “within” your data (i.e. to the whole of your sample). This is particularly important/powerful where your data set is huge.
However, it is clear that Shaffer’s theoretical stance always remains firmly grounded in an interpretivist theoretical frame. Using the framework for research that I originally included in a paper for Computers & Education, but have since expanded in a halfbaked.education blog post (see Figure below) it seems clear that Shaffer’s approach aligns with a qualitative methodology. His data is both numerical and non-numerical, and he is using both non-numerical and numerical techniques of data analysis (including statistics). At the outcomes level his use of generalisability is different to that used in quantitative approaches as he is not claiming to be able to extrapolate beyond his data set. However, when dealing with huge data sets being able to generalise within the set, but beyond the manual qualitative analysis of the data, is powerful. Perhaps it warrants a third type of Outcome in Figure 1 for extrapolation within a massive data set - though I suspect this would still fit within the Qualitative Methodology column as it is still aimed at answering 'the why question'.
Overall, I still think that quantitative ethnography might be less confusingly be called numerical ethnography – thought that is much less catchy. However, this does not detract from the power of the approach or the excellence of the book. I highly recommend it.