NOMINAL RECORD LINKAGE AND LETTER CLUSTER SAMPLING
James E. Bradley · 1990
This study of religious influence on political behavior is based upon identifying single individuals by name in two or more discrete sets of vital, popular, or electoral records. In order to examine a person's political behavior over a span of several elections, it is necessary to link a number of originally unrelated documents and merge them into a single file. In those boroughs where a run of poll books or signed petitions have survived, along with the Dissenters' registers of births or burials, it is possible to identify politically active Nonconformists. Unfortunately, in numerous boroughs the data from the non-parochial registers was inadequate to provide any significant analysis. At Warwick, for example, there was a coercive address circulated with 212 signatures, and while three registers from the Dissenting chapels are available, the earliest register began in 1784 and listed only a handful of early entries. Conversely, at Lancaster the non-parochial registers were excellent, but the coercive petition had only thirty signatures. At Southampton, the absence of occupations in the non-parochial registers rendered it impossible to link the registers to the poll books, while at Coventry, the poll books lacked occupations. However, in this study the baptismal and burial recorders of Dissenters in twelve boroughs were sufficiently well kept to create extended panels or linked files combining registers with poll books or signed petitions and addresses. Variations in surname spelling and the physical mobility of people from parish to parish have proved to be the two most intractable problems in computer assisted record linkage.