Agricultural Household Survey 2025 Report

Annex 2: Sampling Process and Interpreting Results

Sampling process

Analysis of the Scottish Government database of farms and crofts in Scotland contained 52,621 contacts. After cleaning (for example the removal of non-Scottish businesses and duplicate Business Reference Numbers (BRN)), the sample consisted of 47,360 contacts. Furthermore, data was collected during the survey to identify any out of scope addresses (where postal surveys were undeliverable, email surveys which had bounced back, where the farm or croft was considered out of scope for some other reason such as refusals, where the named contact person was deceased or where the land was not being used for agricultural purposes). It should also be noted that there were a total of 92 questionnaires that were returned or submitted where the respondent noted they were not a farm or croft. In some cases, they answered the survey regardless. These cases have been excluded from this report. A total of 4,284 responses were received from 38,298 in scope contacts, which equates to an 11% response rate. This overall level of return provides robust data, accurate to +1.41% of the Scottish agricultural household population. Analysis of subgroups will be less robust.

Interpreting results

The results of the research are based upon a sample survey, therefore all figures quoted are estimates rather than precise percentages of the whole population. All tables have a descriptive and numerical base, showing the population examined in it. A population is a group targeted for analysis or study, in this case crofters and farmers. A sample is part of a population that researchers and statisticians use to collect data. Samples help represent populations and can be useful for making meaningful conclusions about a population. Samples of population, so long as the sample reflects the same characteristics of the overall population, are used to collect views from the whole population.

Standard statistical tests were used to examine whether differences are likely to be due to chance. The z-test was used to compare the differences between scores to establish if the difference was significant. Only differences that are statistically significant at the 95% confidence level are described as differences in the text of this report, unless explicitly stated otherwise. As such, we can be 95% sure that any differences included in this report reflect a genuine, real-world difference, and are not down to chance of who responded.

All proportions produced in a survey have a degree of error associated with them because they are generated from a sample of the population rather than the population as a whole. Any proportion measured in the survey has an associated confidence interval (within which the 'true' proportion of the whole population is likely to lie), usually expressed as ±x%. It is possible with any survey that the sample achieved produces estimates that are outside this range. The number of times out of 100 surveys when the result achieved would lie within the confidence interval is also quoted; conventionally the level set is 95 out of 100, or 95%. Technically, all results should be quoted in this way. However, it is less cumbersome to simply report the percentage as a single percentage, the convention adopted in this report.

Reporting percentages

Percentages are rounded up or down to the nearest whole percentage for all percentages over 0.5%. Where the percentage is less than 0.5% it has been reported to 1 decimal place. Where no respondents have answered a dash “-“ has been used to show non response.

Not all percentages will sum to 100% due to rounding. Rounding can also cause percentages described in the supporting text or summarising ‘overall agreement’ (i.e. adding strongly agree and agree responses together) to differ from the charts by 1% when two percentages are added together.

Where respondents could select more than one response to a question the percentages will sum to more than 100%.