Successive drafts of this paper have been presented over the last four years and I wish to acknowledge the feedback received. An initial presentation was made on 23 August 2012 to the Geography, Environmental and Development Studies Seminar series at Victoria University of Wellington, a year later on 10th October, 2013 to the National Institute of Demographic and Economic Analysis (NIDEA) at the University of Waikato. A later version was presented to the Department of Geography, University of Otago on 2nd October, 2014. The case study was also used as an application of multilevel modelling in a keynote address to the Oceania Stata User Group Meeting in Sydney, on 28 September, 2016 under the title `Multilevel estimation of contextual effects'.

Several people have commented on earlier drafts. Professor Jacques Poot, University of Waikato, made constructive comments on an earlier manuscript. Dr Michael Thomas, Faculty of Spatial Sciences, University of Groningen, made a number of valuable suggestions on the model, some of which I have adopted and others of which I will address in subsequent work. Dr Tom Collins, School of Geography, University of Leeds whose work on civic pride I drew on in the literature review was kind enough to read the penultimate manuscript and he made several points which I've now included in the paper. I also wish to acknowledge the literature review on this topic initially undertaken by Robert Nairn as part of his honours research essay in 2010. Finally, I wish to thank the three anonymous referees whose comments have strengthened the paper. As usual the responsibility for any errors remain mine.