Sheep attacks and harassment: research

3. Methodology

Our research comprised three key components: an initial desk review; a large mixed-method online and telephone survey of sheep farmers; and follow up qualitative research with sheep farmers.

Desk review

The main purpose of the desk review was to provide an overview of the findings of previous related studies, to inform the focus and design of the main stage of fieldwork, including identifying any gaps in the existing literature. The review aimed to cover prior research on the prevalence of attacks on sheep by dogs and wildlife; the impact of attacks on both sheep and on farmers; and on prevention and mitigation measures.

The review was undertaken in a systematic way and followed PRISMA^[38] (Preferred Reporting Items for Systematic Reviews and MetaAnalysis) criteria as far as was possible. However, one key criteria for a successful systematic review strategy is that it must be replicable, hence the requirement for the detailed and precise recording of database name, search terms and scope. By definition a 'systematic review' is not able to include the use of 'grey literature' or expert opinion and they typically take up to eighteen months to complete. Hence as a descriptive and evaluative review, this piece of work was strictly a 'literature review' and not a rigid 'systematic review' as would be appropriate for meta-analysis.^[39]

Primarily NUSEARCH was used; this University of Nottingham resource has access to over one million print books and journals, over 300,000 ebooks and 20,000 ejournals as well as all commonly used databases.

Search Matrix

Concepts -	Sheep	Worrying by Dogs	Predation by Wildlife	Animal Impact	Farmer Impact
Synonyms	Flock, 'Ovis aries'	Dog worrying, dog attack 'dog chasing'	Hunting Predator-attack prey	Trauma Stress, wound, kill, distress, hurt, maul	Cost, 'financial impact', emotional, psychological, stress, worry, concern,
Broader Terms	livestock farm animals	attack, stress, chase, aggression	attack, predation, hunt	damage welfare	Farmer, prevention measures, time, labour, work
Narrower Terms	Lamb ewe	'dog-worry' 'dog attack' 'sheep worrying'	fox, 'white-tailed eagle', 'golden eagle', raven3	Injury, fear, flight, chase, abortion	shepherd, flockmaster, crofter, mitigation,
Word Variations	sheep* flock*	dog* attack* worry*	predat* wild*	wound* trauma*

Searches in NUSEARCH

Search in 'All collections' for 'subject or title' contains sheep and (attack* or worry* or chas*)

91 results of which 14 were considered relevant

Search in 'All collections' for 'subject or title' contains sheep and abort* and stress*

5 results of which none were relevant

For the repeatability of a systematic review, some searches were limited to the individual database Web of Science

Search in Topic for 'All years (1900-2019)' for dog* worry* or dog* attack* and sheep*

114 results of which 14 were considered relevant and read in detail.

Search in Topic for 'All years (1900-2019)' for sheep* injur* and (attack* or worry* or chas*)

28 results of which 2 were considered relevant and fully accessible.

Search in Topic for 'All years (1900-2019)' for sheep* abort* and (attack* or worry* or chas*)

7 results of which none were considered relevant

Search in Topic for 'All years (1900-2019)' for sheep* abort* and stress

27 results of which none were considered relevant

Relevant and cited works are listed in Appendix D.

Quantitative research

Following the initial review of the literature, the main stage of the research involved a large-scale representative mixed-mode survey of sheep farmers. We employed a "push to web" approach, with letters sent out to respondents inviting them to take part online, combined with a telephone survey targeting those who had not responded online.

Sampling

9,148 sheep farmers (based on holdings with sheep) were selected to take part in the survey, identified through a combination of the Scottish Government's Agricultural Census data and other RPID administrative data.

Measures were taken to maximise the representativeness of the sample, with farmers selected to reflect the profile of the Scottish sheep farming population both in terms of regional distribution and the distribution of flock sizes.

We decided against sampling on the basis of sheep distribution as this would have resulted in almost all the farms with large flocks being selected and almost none of those with smaller flocks being selected, which would have precluded analysis by size of flock. Instead, we designed the sample to be representative of sheep farms in Scotland. We banded flock size into four categories (less than 20, 20-149, 150-749 and 750+) and aimed for a roughly equal number of farmers within each band. We then used a stratified sampling approach where (within each band) we ordered the sample by Parish Code and selected 1 in N farmers.

All those selected were sent a letter by post containing a five-digit unique ID code, inviting them to take part online by entering their code at a specific web address. They were also provided with an alternative option, if they were unable to take part online, to opt into the telephone survey by completing and returning a short form provided on the letter to a freepost address.

Questionnaire Development

The survey covered a wide range of topics: the prevalence and time period of attacks on sheep by dogs and wildlife; the impact these attacks have on sheep; the impact of attacks on farmers in terms of the financial impact, the time impact and the emotional impact; the perceived effectiveness of any preventative techniques used by farmers; and farmers' views on potential policy interventions.

In order to ensure that the findings were as current as possible, and that the survey took no longer than 20 minutes on average to complete, farmers were asked to provide details on the circumstances and impacts of incidents in the last year, and specifically of only their most recent attacks by dogs and wildlife.

The survey questions were tested through in-depth telephone interviews with a total of twenty sheep farmers recruited to encompass a mix of geographical areas, a range of flock sizes and a mixture of LFA and Non-LFA land.^[40] Fifteen of the participants were recruited from the mainstage sample, while the remaining five were recruited through informal connections and networks. We split the draft questionnaire into sections, asked participants the questions, then asked how they found each section. We also probed on specific points.

The findings and recommendations for questionnaire changes from the cognitive testing are contained at Appendix F.

Fieldwork

Fieldwork took place between 7 May and 9 June 2019 with online fieldwork conducted throughout the whole period, and the telephone fieldwork taking place between 20 May and 5 June.

Overall, a total of 1931 sheep farmers took part in the survey, including 1346 respondents who took part online and 585 who took part in the telephone survey. The overall response rate was 21% which is high for a survey of this nature.

In addition to providing an opportunity for response amongst farmers unable to participate online, the telephone survey provided an opportunity to target groups underrepresented in the online fieldwork, and to ensure that the final composition of respondents aligned with the profile of the farming population in terms of regional distribution and flock sizes. The final profile of respondents closely matched the population profile in these respects as shown below in Figures 3.1 and 3.2.

Figure 3.1 Sample and respondent profile by flock size

	Flock size
	Achieved	Sample
Base	1930	9165
1 - 19	430	2233
	22%	24%
20 - 149	400	1239
	21%	14%
150 - 749	401	2230
	21%	24%
750+	699	3460
	36%	38%

Figure 3.2 Sample and respondent profile by region^[41]

	Region
	Achieved	Sample
Base	1930	9165
Argyll & Bute	112	521
	6%	6%
Ayrshire	90	480
	5%	5%
Clyde Valley	93	463
	5%	5%
Dumfries & Galloway	211	979
	11%	11%
East Central	57	250
	3%	3%
Eileanan an Iar	194	960
	10%	10%
Fife	37	170
	2%	2%
Highland	399	1827
	21%	20%
Lothian	48	204
	2%	2%
NE Scotland	222	1140
	12%	12%
Orkney	58	334
	3%	4%
Scottish Borders	162	675
	8%	7%
Shetland	98	581
	5%	6%
Tayside	149	568
	8%	6%

For analysis, the data was weighted by region and flock size (according to agricultural census data). This offset the effects of the sampling approach (which over-represented those with smaller flocks) and the small differences in response rate by these variables.

Qualitative research

Following the survey fieldwork, follow up qualitative research was conducted between 18 June and 3 July among survey respondents who had agreed to be re-contacted for further research.

The purpose of the qualitative research was to explore some of the topics covered in the survey in greater depth, including the impact of sheep attacks on farmers and in particular the emotional impact of attacks and views on potential mitigation measures and policy interventions. The qualitative research also provided the opportunity to explore any wider impacts of attacks not covered in the survey, as well as attitudes towards reporting attacks to the police, and to insuring against attacks.

Participants were recruited through an invitation letter, inviting them to opt-in to take part in a focus group. In the event, around one third of participants opted into the research, while around two thirds were recruited through a follow up telephone call. A total of 23 sheep farmers took part in the qualitative research, across five small discussion groups. The research took the form of small focus groups which were held face-to-face in Inverurie, Moffat and Stirling; and by telephone conference with farmers based in Argyll & Bute and the North Western Highlands & Islands, where participants were provided with a free phone number to call. These areas were chosen to achieve a regional spread of participants, and to include areas such as East Central Scotland and Dumfries & Galloway where the reported incidence of attacks was higher than average. Those invited to take part had recent experience of attacks.

The groups lasted around 60-90 minutes and participants were given £30-£35 as a "thank you" for their participation, with a higher incentive offered to those who took part in the longer groups where they attended in person. With participants' consent, the discussions were audio-recorded and transcribed for analysis purposes.