Animal welfare prosecutions reported by the Scottish SPCA: 2011-2019

Findings of analysis of animal welfare case data collected by the Scottish Society for the Prevention of Cruelty to Animals.


The data

The Scottish SPCA are a specialist reporting agency to the Crown Office and Procurator Fiscal Service (PF). This, alongside the powers of investigation for inspectors appointed through the 2006 Act, mean the Scottish SPCA have a legal authority to both investigate animal welfare cases and submit these directly to the PF. They are the only animal welfare organisation with these powers in the UK.

As part of their investigations, the Scottish SPCA collect a large body of information surrounding these cases, providing a wealth of potential data that could be used to help inform policy and practice related to animal welfare issues. This administrative data routinely collected by the Scottish SPCA contains information on charges submitted to the PF, such as the type of offence, legislation, animals involved, outcome and penalties. Demographic information on persons involved in these charges is also collected, such as age, occupation and location.

Information on cases investigated by the Scottish SPCA date back to around the 1930’s, with the earlier years recorded as physical paper records. From around 1980 the Scottish SPCA records are recorded in digital formats using Microsoft Access or Excel. A new recording system was implemented around 2011, with details on cases before this stored separately. For this project the Scottish SPCA provided the Scottish Government with two extracts of data on charges submitted to the PF with offence dates from around 1980 and up until 23 July 2019 (closed cases only). Only charges submitted to the PF by the Scottish SPCA were included in this data, meaning any charges submitted to the PF during this time from other agencies such as the Police and local authorities are not represented.

The first of these data files were extracted from the old records system, in action from around 1980-2011. The second extract contained cases exported from the current data recording system, implemented around 2011. The data was provided as two separate password protected Microsoft Excel spreadsheets, each with different variables and in a range of formats.[16] This meant it was not possible to merge the two files without considerable work and a loss of detail from each. In addition, although in theory these data should not overlap as they cover different recording systems over two separate time periods, there were some cases that had been transferred between these systems, meaning the two sets of data did not cover two discrete periods of time – although the rationale for which had been transferred and others not was not known.

For the extract containing files from the older system there are a total of 3,214 charges with offence dates ranging from 1986[17] until 2014. In the ‘later’ data file there are a total of 2,085 cases ranging from 2000[18] until July 2019. Although there is some duplication between the files[19] not all charges appear in both data files between the dates they overlap (2000 - 2014). Table 1 below shows the number of charges in each data file where they overlap, and the difference between them.

Table 1 shows that from 2000, and up until 2011, there are more charges in the earlier file than in the later file. From 2011 onwards (shown in italics) there are more charges in the later file than are in the earlier file. It is from this point on that we could be more confident that all charges submitted to the PF are included in the later data.

Because of this we have decided in this report to focus on charges from 2011 onwards, using the later data only. This lessens the likelihood of bias as a result of missing cases, especially where the reason for why certain cases or charges may be missing is unknown. In future it may be worth spending some time making sure that all cases held on the earlier system are transferred over to the most recent systems and therefore available for analysis in a consistent format. If it is desirable that only select charges/cases are transferred (e.g. for certain years but not others), then this should be done so in a way that is consistent and well documented. The transfer of historic charges in such a way would allow for analysis to be carried out before 2011 (e.g. to look at changes around the 2006 Act) without the problematic issue of missing cases.

Table 1. Charges in each data extract by year of offence
Year of offence Number in earlier file Number in later file Difference (number) Difference
(% of file with most charges)
2000 57 12 45 21.1
2001 48 8 40 16.7
2002 43 8 35 18.6
2003 43 16 27 37.2
2004 55 28 27 50.9
2005 115 30 85 26.1
(whole year)
128 34 94 26.6
(after 2006 Act only)
63 17 46 27.0
2007 129 47 82 36.4
2008 165 61 104 37.0
2009 202 145 57 71.8
2010 161 145 16 90.1
2011 157 174 -17 90.2
2012 165 234 -69 70.5
2013 153 256 -103 59.8
2014 50 244 -194 20.5


The data was prepared and analysed using Microsoft Excel and IBM SPSS Statistics v24. Results are presented as tables and graphs where appropriate and where there is sufficient justification and large enough numbers, a Pearson’s Chi square test of association was used to look for any significant relationships between variables. A relationship was considered statistically significant if the probability value of achieving a result as extreme was less than 0.05. Cramer’s V was used to look at the strength of these associations, where 0.1 is considered a small effect size, 0.3 a moderate effect size and 0.5 or above a large effect size (Cohen, 1968). Only where this is explicitly mentioned was a test for statistical significance used.

Average values are presented as a mean and standard deviation for variables which appeared to be normally distributed[20] or as a median and interquartile range where distributions could not be considered normal. Average values are used to establish the central, or ‘typical’ value within a range of data, whilst standard deviation (SD) and interquartile range (IQR) are measures of the spread of data, that is how far away the range of values tend to be from the average value. The lower these values the less spread out the data tends to be.

With the exception of a small amount of analysis carried out in section 4.1 of this report, all results are presented at charge, not person level. To carry out this analysis at person level would require the data to be re-structured from long (one row per charge) to a wide (one row per person) format. This is relatively easy to achieve with sufficient timescales.

This report uses the terminology of both charge and offence to describe the crime a defendant is accused of. The report also refers to charge and offence type, where these represent separate categories describing the types of crimes the defendant is accused of. Charge type represents the main legal description for the offence, referring to specific sections of the relevant legislation, whereas offence type refers to categories of offence that contain a more detailed description of the offence. A particular charge type may have multiple offence types associated with this and although in most cases particular offences are associated with specific sections of legislation (e.g. ‘omit to provide veterinary attention’ is normally covered by Section 19 of the 2006 Act), this is not necessarily true in all cases – although some of these may actually represent issues with data quality.



Back to top