Behavioral Testing & Temperament

The Predictive Validity Problem

The scientific literature on working dog temperament testing is actually quite self-critical. Multiple systematic reviews flag a fundamental problem: the field has many tests, but remarkably little evidence that most of them reliably predict operational success.

A 2018 systematic review in Frontiers in Veterinary Science by Brady et al. examined 16 papers on behavioral tests used in working dogs and found a widespread lack of information on reliability and validity. Furthermore, there was no agreed-upon terminology, and no consensus on what actually constitutes "success" in the field.

The most thorough recent synthesis, a 2021 review by Bray et al., summarized best practices across assistance, detection, and protection dog programs. The conclusion was stark: selection efficiency remains a major bottleneck. Most structured programs wash out 50-70% of candidates, often very late in expensive training pipelines, highlighting the inadequacy of early temperament tests.

Why Pet Tests Fail Working Dogs

The C-BARQ (Canine Behavioral Assessment and Research Questionnaire) is the most widely validated general behavioral instrument. Developed by Hsu & Serpell (2003), it is a 68-item questionnaire covering 11 reliable factors like trainability, aggression, and fear.

While used extensively, C-BARQ was designed for pet dogs. A 2020 comparison by Hunt et al. found it less predictive for working dogs than purpose-built instruments. Specifically, program-specific questionnaires (like the r-PWQ for guide dogs) dramatically outperformed C-BARQ on crucial traits like distractibility—which are critical for working roles but poorly captured by pet-dog scales.

Working Scales: DIAS & PANAS

For high-drive working roles (police, military, protection), psychometric tools like the DIAS (Dog Impulsivity Assessment Scale) and PANAS (Positive and Negative Activation Scale) are dominant.

Research comparing active UK police dogs with withdrawn dogs found that active dogs scored significantly higher on "energy and interest" (PANAS) and "responsiveness" (DIAS). Critically, this held across UK police, UK military, and Dutch police dogs, suggesting these traits generalize across rigorous service fields. However, high responsiveness should not be confused with mere impulsivity; a successful working dog must channel energy into directed, functional behavior.

The Cognitive Revolution in Selection

Temperament testing traditionally measures "personality," but cognitive tests represent the newest, most promising direction in working dog selection. Instead of asking "how does the dog feel?", cognitive tests ask "how does the dog think?"

MacLean & Hare (2018) tested a 25-item Dog Cognition Test Battery across hundreds of working dogs. They found that social cognition—specifically eye contact with a human during an unsolvable task—predicted assistance dog success with 86% accuracy in the top quartile. Detection dog success was heavily predicted by sensitivity to human communicative cues and short-term memory.

This is a major paradigm shift. For decades, programs evaluated temperament almost exclusively, largely ignoring the underlying cognitive architecture required to parse complex environmental variables.

Implications for Livestock Guardians

What does this mean for evaluating Livestock Guardian Dogs? Unlike detection or police dogs, LGDs are selected for high independence and low reliance on human direction. However, the core lessons from the scientific literature still apply:

• Context is Everything: A test's validity is population- and program-specific. A behavioral test designed for a suburban pet will penalize an LGD's natural, appropriate territoriality.
• Working Lines Matter: The assumption in most working dog research is that candidate dogs come from proven working lines. Test frameworks rarely account for show vs. working line divergence—they assume the genetic baseline of a working dog is already present.
• Beware Novice Assessments: Observer expertise matters substantially. Research shows novices agree with each other, but frequently disagree with experts when assessing what constitutes a "good" behavioral response in a working context.