Revision of the Historical Series of Divorce Figures

The historical series of Divorce figures (for 1988 onwards) was revised slightly in March 2012 (at the same time as the preliminary figure for 2011 was published), after 810 ‘duplicate’ records were deleted from the National Records of Scotland (NRS) statistical database. The paragraphs that follow describe how the duplicates arose, how they were identified, and the implications of the problem. Information is also available about an earlier (August 2009) revision of the historical series of Divorce figures in our Divorces Time Series data section.

March 2012 revision

With effect from September 2010, the Scottish Court Service has supplied data on divorces (and dissolutions of civil partnerships) in the form of weekly electronic files of data records. Before then, courts provided the information to NRS on paper, using ‘DR1’ forms and photocopies of the extract of the decree of the divorce. NRS keyed the information from the forms, and scanned the images of the extracts of the decrees of divorce, into its computer systems. The procedures we used had some weaknesses, including not checking the newly-received data against the records already in our statistical database. As a result, if a court sent a second form and extract for the same divorce, NRS would process them as if they were data for a completely new divorce, with the result that the statistical database would contain two records for the same divorce.

A new computer system was developed to process the electronically-supplied data, and to create a new consolidated statistical database, which combines the new records with those from the old database. While developing the new database, NRS found a number of cases where it had more than one divorce record which appeared to relate to the same marriage (because each such record had the same marriage reference details). NRS then looked more carefully at its old statistical database, and found 810 records which, although not necessarily exact duplicates of other records in the database, appeared to relate to the same marriages as other records. More information about how these ‘duplicates’ were detected is available in the document ‘Identifying duplicate Divorce records in the data for 1988 to 2009’ (Adobe Acrobat Portable Document Format) (PDF 27 Kb). When the new database was set up, all the ‘duplicates’ that had been identified as such were ‘flagged’ so that they could be excluded from the revised historical series.

PDF files require Acrobat Reader. Download Acrobat Reader free

The 810 ‘duplicates’ represent an average of about 37 such cases per year over the period from 1988 to 2009, inclusive. The scale of the revision varies from year to year (from under 20 to over 60). As there were around 10,000 to 13,000 divorces per year in that period, the exclusion of the ‘duplicates’ from the historical series has made relatively little difference to the levels of the figures for the different years, let alone the overall trends. However, as NRS was setting up a new consolidated statistical database, it had to deal with the problem in the historical data records. All subsequent NRS publications of divorce statistics should use the revised figures. However, material which was produced in 2011 and earlier years will not be revised retrospectively - so, for example, the 2010 Vital Events Reference Tables will continue to show the old figures.

The historical series' figures for 1988 to 2008 (inclusive) have been reduced, as a result of the removal of the duplicates. However, it should be noted that the revised figures for 2009 and 2010 are larger than those that were published in August 2011, due to the inclusion of some ‘very late’ returns for divorces which had been granted in those years. NRS did not receive the data for those divorces in time to count them in the previously-published figures for 2009 and 2010.

August 2009 revision

The historical series of Divorce figures was previously revised in August 2009, when the way in which Divorces were counted by year was changed to use the calendar year in which the decree was granted (rather than the so-called ‘proxy year’). Further information about this revision is available on the August 2009 revision page of this website.