Horse people are often told that a practice is “evidence-based” or “supported by research.” Sometimes the research is used to justify training methods, equipment, or management practices. Sometimes horse owners are asked to volunteer their horses for studies.
And yet many people – including veterinarians, trainers, and riders – quietly admit the same thing: they don’t feel confident judging whether a study has really been designed with horses’ welfare at its centre.
That uncertainty is not a failure of interest or intelligence. It reflects how complex welfare research actually is – especially when it involves a training or equipment intervention.
Why equitation science research is different
Studies that investigate training, handling, restraint, or equipment are not neutral interventions. They deliberately alter what horses experience, how they move, how they respond, and what options are available to them in a given situation.
Horses, in particular, present a special challenge. They are highly trainable, strongly motivated to comply, and capable of continuing to perform despite discomfort, conflict, or stress. That means a lack of obvious resistance or dramatic stress responses does not necessarily mean a lack of welfare impact.
Add to this the reality that equine research is difficult and expensive to conduct. Suitable horses are hard to recruit. Conditions are hard to standardise. Measuring welfare — especially subjective experience — is always an inferential task. Even well-intentioned studies can struggle to detect harm, let alone rule it out confidently.
These challenges are precisely why a large international group of researchers has now published the COMPASS Guidelines.
What are the COMPASS Guidelines?
COMPASS is a set of guidelines designed to support welfare-focused research into behaviour modification, including training and equipment use. They sit alongside existing animal research frameworks (ARRIVE and PREPARE), but focus specifically on the areas where welfare risks are most likely to be underestimated or overlooked.
Rather than telling researchers what conclusions to reach, COMPASS focuses on how studies are designed, conducted, and interpreted. It brings together learning theory, welfare science, ethics, and practical experience with horses to highlight what good practice looks like when behaviour — not just biology — is being studied.
Why this matters to horse owners and practitioners
You don’t need to be a researcher to benefit from these guidelines.
COMPASS offers a way of asking grounded, reasonable questions such as:
- Were baseline welfare measures taken for each horse (that is, how the horse was doing before the study began), or only group averages (where individual experiences can be hidden)?
- Were individual horses monitored, or could signs of discomfort in some horses be masked when results were reported as an overall mean (the mathematical average)?
- Were pressure, restriction, or force properly measured and calibrated, using tools that are known to work reliably for horses, rather than assumed?
- Was welfare assessed using multiple indicators, such as behaviour, physiology, and function, rather than a single proxy (a stand-in measure that is assumed to reflect welfare, but may not do so reliably)?
- Were potential conflicts of interest disclosed, such as industry funding, equipment provision, or commercial involvement?
- Were clear stop criteria in place, meaning predefined points at which a horse would be withdrawn if signs of distress, pain, or excessive strain were observed?
These are not “gotcha” questions. They are the kinds of questions animal ethics committees, peer reviewers, and careful researchers already ask — but which are often invisible to the wider equine community.
COMPASS simply makes those – and many other – expectations visible, setting out 124 detailed points that together describe what welfare-focused behaviour research should account for.
A reference, not a verdict
Importantly, the guidelines are not a checklist for labelling research as “good” or “bad.” They are a reference point — something people can return to when deciding whether to trust a claim, participate in a study, or interpret a finding.
They are also protective of good research. Studies that genuinely prioritise horses’ welfare, use robust methods, and acknowledge limitations are strengthened when clear standards exist. COMPASS helps distinguish careful, transparent work from research that may be well-intentioned but poorly equipped to answer welfare questions.
Why this matters now
Public trust in equine research — and in welfare claims more broadly — depends not on assurances, but on transparency. Horses cannot consent to participation in research. The responsibility therefore falls on humans to ensure that the methods used to study them are genuinely capable of detecting harm, not just dismissing it.
The COMPASS Guidelines are one step toward that accountability. They don’t ask horse people to become scientists — only to recognise that it is reasonable to ask whether research has been designed in a way that truly protects horses’ welfare.
If you are asked to volunteer your horse, or you encounter research being used to justify a practice, COMPASS offers something simple but powerful: a place to stand while you ask better questions.
A practical support for asking better questions
For people who are not researchers, one of the hardest parts of engaging with “evidence-based” claims is knowing where to start. Studies often sound technical, confident, and authoritative — especially when they come from experts, institutions, or respected journals.
One simple way to support that process is to use a large language model (LLM), such as ChatGPT, Claude, or Gemini, as a question-generating tool.
This does not replace expert review, ethical approval, or peer review. It does not tell you whether a study is right or wrong. What it can do is help surface welfare-relevant questions that are easy to miss when you are not trained in research methods — particularly when a study is presented as safe, standard, or well established.
Used carefully, LLMs can help people check whether key protections for horses are clearly stated, or quietly assumed.
For example, they can help you notice:
-
whether the welfare of individual horses was assessed, or only group averages reported;
-
whether baseline welfare was measured before an intervention began;
-
whether pressure, restriction, or force was actually measured, or simply presumed to be minimal;
-
whether clear stop points were defined if a horse showed signs of distress or pain;
-
or whether the absence of visible resistance is being treated as evidence that no harm occurred.
These are not specialist concerns. They are basic safeguards — but they are often buried in methods sections, statistical analyses, supplementary material, or not mentioned at all.
How horse people might use this in practice
If you are given a research paper, a study summary, or a call asking for volunteer horses, you can copy the text into an LLM, upload the COMPASS Guidelines, and ask a simple question such as:
“Based on the COMPASS Guidelines, what welfare safeguards are clearly described here, and what information would a horse owner reasonably want before agreeing to participate?”
If the text is a proposal rather than a published paper, you might ask:e
“According to the COMPASS Guidelines, what welfare risks could arise in this study, and how would those risks normally be monitored or managed?”
The aim is not to accuse researchers of bad faith, or to reject research outright. It is to make implicit assumptions visible — especially those that rely on horses’ trainability, compliance, or ability to continue performing despite discomfort.
A way to stand your ground
You do not need to be suspicious of research to ask for clarity. You do not need to understand every technical detail to notice when important information is missing. And you do not need to defer automatically to expertise when the welfare of a horse is at stake.
The COMPASS Guidelines exist because behaviour-focused research carries real welfare risks that are easy to underestimate. Tools that help surface those risks — whether through guidelines, ethics review, or careful questioning — are not obstacles to good science. They are part of doing it responsibly.
Used in this way, LLMs are not arbiters of truth. They are simply a support for thinking — a way to slow down, ask better questions, and take seriously the responsibility that comes with making decisions on behalf of animals who cannot speak for themselves.
References:
- McGreevy, P. D., Mellor, D. J., Freire, R., Fenner, K., Merkies, K., Warren-Smith, A., … & Henshall, C. (2026). COMPASS Guidelines for Conducting Welfare-Focused Research into Behaviour Modification of Animals. Animals, 16(2), 206.
- Percie du Sert, N., Hurst, V., Ahluwalia, A., Alam, S., Avey, M. T., Baker, M., … & Würbel, H. (2020). The ARRIVE guidelines 2.0: Updated guidelines for reporting animal research. Journal of Cerebral Blood Flow & Metabolism, 40(9), 1769-1777.
- Smith, A. J., Clutton, R. E., Lilley, E., Hansen, K. E. A., & Brattelid, T. (2018). PREPARE: guidelines for planning animal research and testing. Laboratory animals, 52(2), 135-141.
