Literature review of the use of Qualitative Behaviour Assessment with a fixed list of terms
Irena Czycholl, Cecilie Ravn Skovlund, Björn Forkman
TL;DR
This review summarizes how a fixed-term method for assessing animal behavior is used across different species and settings.
Contribution
The paper provides the first comprehensive overview of FL QBA applications, methods, and species.
Findings
FL QBA is most commonly used for on-farm welfare assessment and behavioral profiling.
The method has been applied to farmed, working, companion, and exotic animals but not laboratory animals.
Methodological approaches and reporting quality vary significantly across studies.
Abstract
Qualitative Behaviour Assessment (QBA) is a method that is used to assess emotional states in animals, either based on a list of pre-established terms (fixed list; FL) or developed through Free-Choice-Profiling. Although FL QBA was originally developed for welfare assessment of farm animals, it is nowadays also used for various species and other sectors. This is, amongst others, because QBA contributes a unique ‘whole-animal’ insight into animal experience that complements other measures and its high feasibility along with a general lack of available indicators of positive emotional state. This has led to a number of different usages and applications of FL QBA of which an overview (e.g., exact methodology used, statistical analysis, purpose and aim) so far does not exist. The aim of this review is to provide an overview of the studies that have applied FL QBA, the species it has been…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
| Reference | Species | Setting | Life stage | Aim of QBA1 |
|---|---|---|---|---|
| Adamie et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Andreasen et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Andreasen et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Andric et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Armbrecht et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Barry et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Bokkers et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Brscic et al. ( | Cattle | Farm (dairy production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Bugueiro et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Bugueiro et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Ceballos et al. ( | Cattle | Farm (dairy production) | Adult | Emotional/behavioural response or change to event |
| Chen et al. ( | Cattle | Farm (experimental setting, meat production) | Juvenile | Temperament/behavioural profile |
| Coignard et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Coignard et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Collins et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Collins et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Cooke et al. ( | Cattle | Farm (meat production) | Adult | General welfare assessment/emotional state (WQ) |
| Cooke et al. ( | Cattle | Farm (meat production) | Juvenile and adult | General welfare assessment/emotional state (WQ) |
| de Andrade Kogima et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| des Roches ( | Cattle | Farm (experimental setting, dairy production) | Adult | General welfare assessment/emotional state (WQ)/ emotional/behavioural response or change to event |
| de Graaf et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| de Rosa et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| de Vries et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| de Vries et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| de Vries et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| des Roches et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Dos Santos et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Ebinghaus et al. ( | Cattle | Farm (dairy production) | Adult | Emotional/behavioural response or change to event |
| Ebinghaus et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state |
| Ebinghaus et al. ( | Cattle | Farm (dairy production) | Adult | Emotional/behavioural response or change to event |
| Ebinghaus et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state |
| Ebinghaus et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Ellingsen et al. ( | Cattle | Farm (dairy production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Garro-Aguilar et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Gieseke et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Gois et al. ( | Cattle | Farm (meat production) | Juvenile | Temperament/behavioural profile |
| Grimard et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Gutmann et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Hernandez et al. ( | Cattle | Farm (dual purpose) | Adult | General welfare assessment/emotional state (WQ) |
| Hernandez et al. ( | Cattle | Farm (dual purpose) | Adult | General welfare assessment/emotional state (WQ) |
| Hulsmann et al. ( | Cattle | Farm (meat production) | Juvenile | Temperament/behavioural profile |
| Kaurivi et al. ( | Cattle | Farm (meat production) | Adult | General welfare assessment/emotional state (WQ) |
| Kirchner et al. ( | Cattle | Farm (meat production) | Adult | General welfare assessment/emotional state (WQ) |
| Kirchner et al. ( | Cattle | Farm (meat production) | Adult | General welfare assessment/emotional state (WQ) |
| Krug et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Lutz et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Molina et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Popescu et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Popescu et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Rizzuto et al. ( | Cattle | Rodeo | Juvenile | Emotional/behavioural response or change to event |
| Russell et al. ( | Cattle | Farm (experimental, dairy production) | Adult | Emotional/behavioural response or change to event |
| Sant’Anna and da Costa ( | Cattle | Farn (meat production) | Juvenile | Temperament/behavioural profile |
| Schmitz et al. ( | Cattle | Farm (dairy production) | Adult | Emotional/behavioural response or change to event |
| Schulz et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Thomann et al. ( | Cattle | Farm (dairy production) | Not reported | General welfare assessment/emotional state (WQ) |
| Tremetsberger et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Tremetsberger et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Valente and Stilwell ( | Cattle | Farm (various) | Not reported | General welfare assessment/emotional state (WQ) |
| van Eerdenburg et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Vucemilo et al. ( | Cattle | Farm (dairy) production | Adult | General welfare assessment/emotional state (WQ) |
| Wagner et al. ( | Cattle | Farm (dairy) production | Adult | General welfare assessment/emotional state (WQ) |
| Wagner et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Zhitia et al. ( | Cattle | Farm (dairy) production | Adult | General welfare assessment/emotional state (WQ) |
| Zuliani et al. ( | Cattle | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| da Silva et al. ( | Cattle (Zebu) | Farm (dairy production) | Juvenile | General welfare assessment/emotional state |
| de Rosa et al. ( | Buffalo | Farm (dairy production) | Adult | General welfare assessment/emotional state (WQ) |
| Napolitano et al. ( | Buffalo | Farm (dairy production) | Adult | Other (c-QBA: evaluation of changes in animal behaviour during an observation/response to an event) |
| Serrapica et al. ( | Buffalo | Farm | Adult | Other (c-QBA: evaluation of changes in animal behaviour during an observation/response to an event) |
| Brandt et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Camerlink et al. ( | Pig | Experimental setting | Juvenile | General welfare assessment/emotional state (WQ) |
| Cardona et al. ( | Pig | Farm (experimental, meat production) | Juvenile | General welfare assessment/emotional state/ Emotional/behavioural response/change to event |
| Cardona et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state/ Emotional/behavioural response/change to event |
| Carreras et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Carroll et al. ( | Pig | Farm (experimental, meat production) | Juvenile | General welfare assessment/emotional state |
| Clarke et al. ( | Pig | Farm (meat production) | Adult | General welfare assessment/emotional state (WQ) |
| Czycholl et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Czycholl et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Czycholl et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Czycholl et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Czycholl et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Duijvesteijn et al. ( | Pig | Farm (meat production) | (Juvenile) | General welfare assessment/emotional state (WQ) |
| Friedrich et al. ( | Pig | Farm (meat production) | Adult | General welfare assessment/emotional state (WQ) |
| Friedrich et al. ( | Pig | Farm (meat production) | Juvenile and adult | General welfare assessment/emotional state (WQ) |
| Friedrich et al. ( | Pig | Farm (meat production) | Juvenile and adult | General welfare assessment/emotional state (WQ) |
| Friedrich et al. ( | Pig | Farm (meat production) | Juvenile and adult | General welfare assessment/emotional state (WQ) |
| Friedrich et al. ( | Pig | Farm (meat production) | Adult | General welfare assessment/emotional state (WQ) |
| Hubbard and Scott ( | Pig | Farm (meat production) | Adult | General welfare assessment/emotional state (WQ) |
| Kang et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Losada-Espinosa et al. ( | Pig | Farm (meat production) | All | General welfare assessment/emotional state (WQ) |
| Martin et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Martinez et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Meyer-Hamme et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Munsterhjelm et al. ( | Pig | Farm (meat production) | Juvenile and adult | General welfare assessment/emotional state (WQ) |
| Munsterhjelm et al. ( | Pig | Farm (meat production) | Juvenile and adult | General welfare assessment/emotional state (WQ) |
| Oldham et al. ( | Pig | Farm (experimental, meat production) | Juvenile | Emotional/behavioural response/change to event |
| Rocha et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Schmitt et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Schmitt et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Temple et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Temple et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Temple et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Termatzidou et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Thomann et al. ( | Pig | Farm (meat production) | Juvenile and adult | General welfare assessment/emotional state (WQ) |
| Vitali et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Vitali et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Wiseman-Orr et al. ( | Pig | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Bassler et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Buijs et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Chen et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| de Jong et al. ( | Chickens | Farm/slaughter plant (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| di Marcantonio et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Federici et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Granquist et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| He et al. ( | Chickens | Farm (egg production) | Adult | General welfare assessment/emotional state (WQ) |
| Iannetti et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Iannetti et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Li et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Muri et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Nenadovic et al. ( | Chickens | Farm (egg production) | Adult | General welfare assessment/emotional state (WQ) |
| Plitman et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Sans et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Sans et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Sans et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Sans et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Souza et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Souza et al. ( | Chickens | Farm (experimental and commercial, meat production) | Juvenile | General welfare assessment/emotional state |
| Tuyttens et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Vasdal et al. ( | Chickens | Farm (meat production) | Juvenile | General welfare assessment/emotional state (WQ) |
| Vasdal et al. ( | Chickens | Farm (egg production) | Adult | General welfare assessment/emotional and/or behavioural change/response |
| Bodas et al. ( | Sheep | Farm (meat production) | Juvenile | General welfare assessment/emotional state (AWIN) |
| Collins et al. ( | Sheep | Farm/pre-export (wool production) | Adult | General welfare assessment/emotional state |
| Diaz-Lundahl et al. ( | Sheep | Farm (various) | Adult | General welfare assessment/emotional state |
| Hernandez et al. ( | Sheep | Farm (grazing, wool production) | Not reported | General welfare assessment/emotional state |
| Mialon et al. ( | Sheep | Farm (experimental, meat production) | Juvenile | General welfare assessment/emotional state (AWIN) |
| Muri and Stubsjøen ( | Sheep | Farm | Adult | General welfare assessment/emotional state |
| Phythian et al. ( | Sheep | Farm (various settings) | All | General welfare assessment/emotional state |
| Phythian et al. ( | Sheep | Farm | Juvenile and adult | General welfare assessment/emotional state |
| Stubsjøen et al. ( | Sheep | Farm | Adult | General welfare assessment/emotional state (FåreBygg project) |
| Willis et al. ( | Sheep | Sea transport (wool production) | All | General welfare assessment/emotional state |
| Battini et al. ( | Goat | Farm (dairy production) | Adult | General welfare assessment/emotional state (AWIN) |
| Battini et al. ( | Goat | Farm (dairy production) | Adult | General welfare assessment/emotional state (AWIN) |
| Battini et al. ( | Goat | Farm (dairy production) | Adult | General welfare assessment/emotional state |
| Battini et al. ( | Goat | Farm (dairy production) | Adult | General welfare assessment/emotional state (AWIN) |
| Can et al. ( | Goat | Farm (dairy production) | Adult | General welfare assessment/emotional state (AWIN) |
| Can et al. ( | Goat | Farm (dairy production) | Adult | General welfare assessment/emotional state (AWIN) |
| Costa et al. ( | Goat | Farm (experimental, feedlot system) | (Adult) | General welfare assessment/emotional state |
| Grosso et al. ( | Goat | Farm (dairy production) | Adult | General welfare assessment/emotional state |
| Muri et al. ( | Goat | Farm (dairy production) | Adult | General welfare assessment/emotional state |
| Muri et al. ( | Goat | Farm (dairy production) | Adult | General welfare assessment/emotional state |
| Napolitano et al. ( | Goat | (Farm) | Juvenile | Other (c-QBA: evaluation of changes in animal behaviour during an observation/response to an event) |
| Czycholl et al. ( | Horse | Farm/boarding stable | Adult and senior | General welfare assessment/emotional state (AWIN) |
| Czycholl et al. ( | Horse | Farm (various stables) | Adult and senior | General welfare assessment/emotional state (AWIN) |
| Czycholl et al. ( | Horse | Farm/boarding stable (various) | Adult | General welfare assessment/emotional state (AWIN) |
| Dai et al. ( | Horse | Farm/boarding stable | Adult | Emotional/behavioural response to an event |
| Gronqvist et al. ( | Horse | Not reported | Adult | Other (description of expressive behaviour; valence and arousal) |
| Harvey et al. ( | Horse | Wild | All | General welfare assessment/emotional state |
| Jaramillo et al. ( | Horse | Racing | Juvenile and adult | Emotional/behavioural response to an event |
| Minero et al. ( | Horse | Farm (various facilities) | Adult | General welfare assessment/emotional state |
| Mullan et al. ( | Horse | Public grazing land | All | General welfare assessment/emotional state |
| Popescu et al. ( | Horse | Farm (privately owned stallions) | Adult | General welfare assessment/emotional state (AWIN) |
| Rowland et al. ( | Horse | Traveller and gypsy owned horses | Adult | General welfare assessment/emotional state |
| Ruet et al. ( | Horse | Riding school | Adult | Emotional/behavioural response to an event (AWIN) |
| Ruet et al. ( | Horse | Riding school | Adult | Emotional/behavioural response to an event (AWIN) |
| Dai et al. ( | Donkey | Farm | Adult | General welfare assessment/emotional state (AWIN) |
| Dai et al. ( | Donkey | Farm (dairy production) | Adult and senior | General welfare assessment/emotional state (AWIN) |
| Gonzalez et al. ( | Donkey | Not reported | Juvenile and adult | Emotional/behavioural response to an event |
| Minero et al. ( | Donkey | Farm (various) | Juvenile and adult | General welfare assessment/emotional state |
| Arena et al. ( | Dog | Shelter | Adult | General welfare assessment/emotional state |
| Arena et al. ( | Dog | Shelter | Adult | General welfare assessment/emotional state (SQP) |
| Barnard et al. ( | Dog | Shelter | Adult | General welfare assessment/emotional state |
| Berteselli et al. ( | Dog | Shelter | All | General welfare assessment/emotional state (SQP) |
| Berteselli et al. ( | Dog | Shelter | Not applicable | General welfare assessment/emotional state (SQP) |
| Cuglovici and Amaral ( | Dog | Shelter | Adult | General welfare assessment/emotional state (SQP) |
| Harvey et al. ( | Dog | Shelter | Adult and senior | Emotional/behavioural response to an event |
| Menchetti et al. ( | Dog | Shelter | Not reported | Emotional/behavioural response or change to event |
| Pedersen and Malm ( | Dog | Pedagogical school dogs | Adult | Emotional/behavioural response or change to event |
| Raudies et al. ( | Dog | Shelter | Adult | General welfare assessment/emotional state |
| Shaw et al. ( | Dog | Privately owned companion dogs (test facility) | Adult and senior | Emotional and/or behavioural change/response |
| Stubsjøen et al. ( | Dog | Shelter | Not reported | General welfare assessment/emotional state (SQP) |
| Stubsjøen et al. ( | Dog | Shelter | Not reported | General welfare assessment/emotional state (SQP) |
| Heritier et al. ( | Dog | Shelter | Not applicable | General welfare assessment/emotional state (SQP) |
| Travnik and Sant’Anna ( | Cat | Shelter | Not reported | Temperament/behavioural profile |
| Travnik et al. ( | Cat | Shelter | Not reported | Temperament/behavioural profile |
| Jarvis et al. ( | Atlantic salmon | Farm (hatchery and rearing unit) | Juvenile | General welfare assessment/emotional state |
| Wiese et al. ( | Atlantic salmon | Experimental setting | Juvenile | Emotional/behavioural response or change to event |
| Stagni et al. ( | Brown bear | Sanctuary | Various | General welfare assessment/emotional state |
| Delfour et al. ( | Dolphin | Not reported | Adult and juvenile | Other (qualitative behavioural scoring; emotional/behavioural response or change to event) |
| Yon et al. ( | Elephant (Asian and African) | Zoo | Various | General welfare assessment/emotional state |
| Dobrikj et al. ( | Elephant (Asian and African) | Zoo | Not reported | General welfare assessment/emotional state |
| Gartland et al. ( | Gorilla | Zoo | Adult | Other (qualitative behavioural scoring) |
| Munerato et al. ( | Pampas deer | Wild | Adult | Emotional/behavioural response or change to event |
| Skovlund et al. ( | Polar bear | Zoo | Various | General welfare assessment/emotional state |
| Nogueira et al. ( | White lipped pecari and collared pecari | Farm (commercial hunting) | Adult | Temperament/behavioural profile |
| Reference | Origin QBA list of terms | Observer training | Live or video-based observation | No. of individuals per assessment | Time per assessment | Length of VAS (mm) | Analysis level |
|---|---|---|---|---|---|---|---|
| Adamie et al. ( | WQ | No training reported | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Andreasen et al. ( | WQ; own creation | Official training (WQ); no training | Live | Not reported | 20 min | 125 | PC level |
| Andreasen et al. ( | WQ; own creation | Official training (WQ); no training | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Andric et al. ( | WQ | No training reported | Live | Not reported (according to WQ) | Not reported (approx. 2.5–10 min) | 125 | WQ aggregation (PC level) |
| Armbrecht et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Barry et al. ( | WQ; own creation | Official training (WQ) | Live | Whole group | Not reported | 125 | WQ aggregation (PC level) |
| Bokkers et al. ( | WQ | Other training; no training | Video | Not reported | 2 min | 125 | Term and PC level |
| Brscic et al. ( | WQ | Other training | Live | Not reported | 20 min | 125 | PC level |
| Bugueiro et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Bugueiro et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Ceballos et al. ( | Gois et al. ( | Other training | Video | 1 | Not reported | 125 | PC level |
| Chen et al. ( | Sant’Anna and da Costa ( | No training reported | Live | 1 | 0.5 min | 136 | Term level |
| Coignard et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Coignard et al. ( | WQ | Other training | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Collins et al. ( | WQ | No training reported | Live | Not reported | 2.5–10 min | 125 | Not reported |
| Collins et al. ( | WQ | Official training (WQ) | Live | Not reported (according to WQ) | 2.5–10 min | 125 | PC level |
| Cooke et al. ( | WQ | Other training | Live; Video | Not reported | 10 min | 125 | Term and PC level |
| Cooke et al. ( | WQ | Other training | Live | Whole group | 10 min | 125 | PC level |
| de Andrade Kogima et al. ( | WQ | No training reported | Live | Not reported | 20 min | 125 | Term level |
| des Roches ( | WQ; own creation | Other training | Live | Individual | 5 min | 125 | PC level |
| de Graaf et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| de Rosa et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| de Vries et al. ( | WQ | Official training (WQ) | Live | Not reported (according to WQ) | 20 min | 125 | WQ aggregation Term level |
| de Vries et al. ( | WQ | Official training (WQ) | Live | Whole group | 20 min | 125 | WQ aggregation (PC level) |
| de Vries et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| des Roches et al. ( | WQ; own creation | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Dos Santos et al. ( | WQ | Other training | Live | Not reported | 60 min | 125 | PC level |
| Ebinghaus et al. ( | Focus group; own creation | Other training | Live | 1 | Not reported | 125 | PC level |
| Ebinghaus et al. ( | Ebinghaus et al. ( | Other training | Live | 1 | Not reported | 125 | PC level |
| Ebinghaus et al. ( | Ebinghaus et al. ( | Other training | Live | 1 | Not reported | 125 (QBA App) | PC level |
| Ebinghaus et al. ( | Ebinghaus et al. ( | Other training | Live | Not reported | Not reported | 125 | PC level |
| Ebinghaus et al. ( | WQ | No training reported | Live | Not reported | Not reported | Not reported | PC level |
| Ellingsen et al. ( | WQ | No training reported | Live | 1 | 10–20 min | 125 | PC level |
| Garro-Aguilar et al. ( | WQ | Official training (WQ) | Live | Not reported (according to WQ) | 2.5–10 min | 125 | WQ aggregation (PC level) |
| Gieseke et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Gois et al. ( | Sant’Anna and da Costa ( | Other training | Live | 1 | 5 s | Not reported | Term and PC level |
| Grimard et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Gutmann et al. ( | WQ | Other training | Video | Not reported | 4 min | 125 | Term and PC level |
| Hernandez et al. ( | WQ | No training reported | Live | Whole group | 20 min | 125 | WQ aggregation (PC level) |
| Hernandez et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Hulsmann et al. ( | Sant’Anna and da Costa ( | No training reported | Live | 1 | Not reported | 136 | Term level |
| Kaurivi et al. ( | WQ; own creation | No training reported | Live | Whole group | 20 min | 125 | Not reported |
| Kirchner et al. ( | WQ | Official training (WQ) | Live | Whole group | 20 min | 125 | WQ aggregation (PC level) |
| Kirchner et al. ( | WQ | Official training (WQ) | Live | Whole group | 20 min | 125 | WQ aggregation (PC level) |
| Krug et al. ( | WQ | No training reported | Live | Whole group | 20 min | 125 | WQ aggregation (PC level) |
| Lutz et al. ( | WQ | Official training (WQ) | Live | Not reported (according to WQ) | 3.5–10 min | 125 | Not reported (According to WQ) |
| Molina et al. ( | WQ | No training reported | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Popescu et al. ( | WQ | No training reported | Live | Whole group | 20 min | 125 | WQ aggregation (PC level) |
| Popescu et al. ( | WQ | Other training | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Rizzuto et al. ( | WQ; own creation | Other training | Video | 1 | Not reported | Survey software | Term and PC level |
| Russell et al. ( | WQ | Other training | Live | 48 | 20 min | 125 | PC level |
| Sant’Anna and da Costa ( | WQ; own creation | No training reported | Live | 1 | 30 s | Not reported | PC level |
| Schmitz et al. ( | Ebinghaus et al. ( | Other training | Live | 1 | Not reported | 125 | PC level |
| Schulz et al. ( | WQ | Official training (WQ) | Live | Whole group | 20 min | 125 | WQ aggregation (PC level) |
| Thomann et al. ( | WQ | No training reported | Live | Not reported (according to WQ) | Not reported (according to WQ) | Not reported (according to WQ) | PC level |
| Tremetsberger et al. ( | WQ | No training reported | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Tremetsberger et al. ( | WQ | No training reported | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Valente and Stilwell ( | WQ | No training reported | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| van Eerdenburg et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Vucemilo et al. ( | WQ | No training reported | Live | 17–20 | 20 min | Not reported | Term level |
| Wagner et al. ( | WQ | Official training (WQ) | Live; Video | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Wagner et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Zhitia et al. ( | WQ | Official training (WQ) | Live | Not reported | 20 min | 125 | WQ aggregation (PC level) |
| Zuliani et al. ( | WQ | Official training (WQ) | Live | Whole group | 20 min | 125 | WQ aggregation (PC level) |
| da Silva et al. ( | WQ; literature; own creation | No training reported | Video | 1 | 3 min | 125 | PC level |
| De Rosa et al. ( | WQ (for cattle) | Official training (WQ) | Live | 25–64 | 2.5–20 min | 125 | Term and PC level |
| Napolitano et al. ( | Napolitano et al. ( | Other training | Video | 1 | 150 s | 100 | (Term level) (c-QBA) |
| Serrapica et al. ( | Napolitano et al. ( | Other training | Video | 1 | 2 min | 100 | (Term level) (c-QBA) |
| Brandt et al. ( | WQ | Not reported | Live | Sample of herd | 3.5–10 min | 125 | WQ aggregation (PC level) |
| Camerlink et al. ( | WQ; Duijvesteijn et al. ( | Other training | Video | 1 | 1 min | 125 | PC level |
| Cardona et al. ( | WQ | Other training | Video | 5 | 3–5 min | 125 | Term and PC level |
| Cardona et al. ( | WQ | Other training | Video | 10–12 | 1–5 min | 125 | Term and PC level |
| Carreras et al. ( | WQ | Other training | Live | 11 | 10 min | 125 | WQ aggregation (PC level) |
| Carroll et al. ( | Not reported | Not reported | Live; Video | 1 | Not reported | 100 | PC level |
| Clarke et al. ( | WQ | Other training | Video | 15–18 | 1 min | 100 | Term and PC level |
| Czycholl et al. ( | WQ | Official training (WQ) | Live | 80–240 | 3.5–5 min | 125 | Term level |
| Czycholl et al. ( | WQ | Official training (WQ) | Live | 100–200 | 3.5–5 min | 125 | Term level and WQ aggregation (PC level) |
| Czycholl et al. ( | WQ | Official training (WQ) | Live; Video | Not reported | 3.5–20 min | 125 | Term and PC level |
| Czycholl et al. ( | WQ | Official training (WQ) | Live | Sample of herd | 3.5–10 min | 125 | WQ aggregation (PC level) |
| Czycholl et al. ( | WQ | Official training (WQ) | Live | Sample of herd | 3.5–5 min | 125 | WQ aggregation (PC level) |
| Duijvesteijn et al. ( | WQ; own creation | No training reported | Video | 1 | 1–2 min | 125 | PC level |
| Friedrich et al. ( | WQ | Official training (WQ) | Live | Sample of herd | 3.5–5 min | 125 | Term and PC level |
| Friedrich et al. ( | WQ | Official training (WQ) | Live | Sample of herd | 2.5–10 min | 125 | Term and PC level |
| Friedrich et al. ( | WQ | Official training (WQ) | Live | Not reported | Not reported | Not reported | Not reported |
| Friedrich et al. ( | WQ | Official training (WQ) | Live | Sample of herd | 2.5–10 min | 125 | WQ aggregation (PC level) |
| Friedrich et al. ( | WQ | Official training (WQ) | Live | Sample of herd | 2.5–10 min | 125 | Term and PC level |
| Hubbard and Scott ( | WQ | No training reported | Live | Not reported | 2.5–10 min | 125 | Term |
| Kang et al. ( | WQ | No training reported | Live | Sample of herd | 2.5–10 min | 125 | Term level; WQ aggregation (PC level) |
| Losada-Espinosa et al. ( | WQ | No training reported | Live | Sample of herd | 2.5–10 min; 3.5–10 min | 125 | WQ aggregation (PC level) |
| Martin et al. ( | WQ | Official training (WQ) | Live | Sample of herd | 2.5–10 min | 125 | WQ aggregation (PC level) |
| Martinez et al. ( | WQ | Official training (WQ) | Live | Sample of herd | 2.5–10 min | 125 | WQ aggregation (PC level) |
| Meyer-Hamme et al. ( | WQ | Official training (WQ) | Live | 100–200 | 50 | 125 | WQ aggregation (PC level) |
| Munsterhjelm et al. ( | WQ | Official training (WQ) | Live | Sample of herd | 2.5–10 min | 125 | PC level |
| Munsterhjelm et al. ( | WQ | Other training | Live | Sample of herd | 2.5–10 min | 125 | PC level |
| Oldham et al. ( | WQ; Duijvesteijn et al. ( | Other training | Video | 2 | 30 s | 125 | PC level |
| Rocha et al. ( | WQ | Other training | Live | Sample of herd | 2.5 | 125 | WQ aggregation (PC level) |
| Schmitt et al. ( | WQ | No training reported | Live | 12 | 20 | 125 | Term and PC level; WQ aggregation (PC level) |
| Schmitt et al. ( | WQ | No training reported | Live | Not reported | 2 min | 125 | PC level |
| Temple et al. ( | WQ | Other training | Live | Sample of herd | 2.5 min | 125 | Term level |
| Temple et al. ( | WQ | Other training | Live | Sample of herd | 2.5 min | 125 | PC level |
| Temple et al. ( | WQ | No training reported | Live | 100–200 | 2.5 min | 125 | PC level |
| Termatzidou et al. ( | WQ | No training reported | Live | 10 | 2 | / | (Term level) |
| Thomann et al. ( | Not reported | Not reported | Not reported | Not reported | Not reported | Not reported (according to WQ) | Not reported |
| Vitali et al. ( | WQ | Official training (WQ) | Live | Not reported | 5 | 125 | WQ aggregation (PC level) |
| Vitali et al. ( | WQ | Other training | Live | Not reported | 5 | 125 | Term and PC level |
| Wiseman-Orr et al. ( | / | No training reported | / | / | / | 100; absence/presence | Not reported |
| Bassler et al. ( | WQ | Official training (WQ) | Live | Whole group | 20 | 125 | WQ aggregation (PC level) |
| Buijs et al. ( | WQ | Other training | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| Chen et al. ( | WQ | Not reported | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| de Jong et al. ( | WQ | Other training | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| di Marcantonio et al. ( | WQ | No training reported | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| Federici et al. ( | WQ | Other training | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| Granquist et al. ( | WQ | No training reported | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| He et al. ( | WQ | Other training | Live | Whole group | 20 min | Not reported | WQ aggregation (PC level) |
| Iannetti et al. ( | WQ | No training reported | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| Iannetti et al. ( | WQ | No training reported | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| Li et al. ( | WQ | No training reported | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| Muri et al. ( | WQ | Other training | Live | Whole group | Not reported | Not reported | PC level |
| Nenadovic et al. ( | WQ | No training reported | Live | Whole group | Not reported | Not reported | Term level |
| Plitman et al. ( | WQ | No training reported | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| Sans et al. ( | WQ; own creation | No training reported | Live | 100 | 5 min | 125 | Term level |
| Sans et al. ( | WQ; Souza et al. ( | Official training (WQ) | Live | Whole group | 10 min | 125 | PC level |
| Sans et al. ( | WQ; Souza et al. ( | Official training (WQ) | Live | Whole group | 10 min | 125 | PC level |
| Sans et al. ( | WQ; own creation | No training reported | Live | Whole group | 10 min | 125 | PC level |
| Souza et al. ( | WQ | Other training | Live | Whole group | 10 | 125 | WQ aggregation (PC level) |
| Souza et al. ( | Focus group; own creation | Other training | Video | Group | 1 min | 125 | Term and PC level |
| Tuyttens et al. ( | WQ | Official training (WQ) | Live | Whole group | 20 min | 125 | WQ aggregation (PC level) |
| Vasdal et al. ( | WQ | Official training (WQ) | Live | Whole group | Not reported | Not reported | WQ aggregation (PC level) |
| Vasdal et al. ( | WQ; own creation | Official training (WQ)/Other training | Live | Whole group | 20 min | 125 | PC level |
| Bodas et al. ( | AWIN | No training reported | Live | Not reported | Not reported (according to AWIN) | Not reported according to Grosso et al. ( | PC level |
| Collins et al. ( | Literature; own creation | Other training | Video | (20) | 30–45 s | 100 | PC level |
| Diaz-Lundahl et al. ( | Literature Muri and Stubsjøen ( | Other training | Video | 111 | 2 min | 125 | PC level |
| Hernandez et al. ( | AWIN; focus group; own creation | Other training | Live | 9–2,000 | 20 min | Not reported | PC level |
| Mialon et al. ( | AWIN | Other training | Video | (7; pen) | 45 s | 125 | PC level |
| Muri and Stubsjøen ( | Literature; focus group; own creation | Other training | Live; Video | 3–26; 43–109 | 2 min; 20 min | 125 | Term and PC level |
| Phythian et al. ( | Previous study (FCP); focus group; own creation | Official training | Video | 1—whole group | 1 min | 125 | PC level |
| Phythian et al. ( | Phythian et al. ( | Other training | Live | 77 | 30 | 125 | PC level |
| Stubsjøen et al. ( | Literature Muri and Stubsjøen ( | Other training | Live | Whole group | 20 min | Not reported | PC level |
| Willis et al. ( | Literature; own creation | Not reported | Live | Whole group | 5–8 min | 100 | (PC level) |
| Battini et al. ( | Grosso et al. ( | Other training | Live | 72.28 ± 7.40 | 10–20 | 125 | PC level |
| Battini et al. ( | AWIN | Other training | Live | Not reported | Not reported | Not reported | Not reported |
| Battini et al. ( | AWIN | Official training (AWIN) | Live | 7–192 | Not reported | Not reported | PC level |
| Battini et al. ( | AWIN | Official training (AWIN) | Live | Whole group | 10 | Not reported (according to AWIN) | PC level |
| Can et al. ( | AWIN | Other training | Live | Whole group | 10–20 | Not reported | Term and PC level |
| Can et al. ( | AWIN | Other training | Live | 12.5–167 | Not reported (according to AWIN) | Not reported (according to AWIN) | / |
| Costa et al. ( | Grosso et al. ( | Not reported | Live | 8 | Not reported | Not reported | Term level |
| Grosso et al. ( | Literature; focus group; own creation | Other training | Live | Whole herd | 10–20 | 125 | Term and PC level |
| Muri et al. ( | WQ (dairy cows); own creation | No training reported | Live | 11–173 | 20 | Not reported | Term level |
| Muri et al. ( | Not reported; own creation | Other training | Live | Whole group | Not reported | Not reported | Term level |
| Napolitano et al. ( | Focus group; own creation | Other training | Video | 9–10 | 90 s | 100 | (Term level) (c-QBA) |
| Czycholl et al. ( | AWIN | Official training (AWIN) | Live | 1 | 1 min | 125 | / |
| Czycholl et al. ( | AWIN | Official training (AWIN) | Live | 1 | 1 min | 125 | Term and PC level |
| Czycholl et al. ( | AWIN | Official training (AWIN) | Live | 1 | 1 min | 125 | PC level |
| Dai et al. ( | FCP; own creation | Other training | Video | 1 | 39–110 s | 125 | Term and PC level |
| Gronqvist et al. ( | Minero et al. ( | No training reported | Video | 1 | 10 s | 5-point scale | Term level |
| Harvey et al. ( | Not reported | Other training | Video | 1 | 1–252 s | Not reported | / |
| Jaramillo et al. ( | FCP; own creation | Other training | Video | 1 | 3 s | Not reported | Term and PC level |
| Minero et al. ( | Literature; focus group; own creation | Other training | Live | 1 | 1 min | 125 | PC level |
| Mullan et al. ( | Not reported | No training reported | Live | 1 | Not reported | Not reported | PC level |
| Popescu et al. ( | AWIN | Official training (AWIN) | Live | 1 | 30–60 s | 125 | PC level |
| Rowland et al. ( | Focus group; own creation | No training reported | Live | 1 | 2–3 min | Not reported | PC level |
| Ruet et al. ( | AWIN | Other training | Video | 1 | 8 min | 100 | PC level |
| Ruet et al. ( | AWIN | Other training | Live | 1 | 1 min | 125 | Term level |
| Dai et al. ( | AWIN | Other training | Live | Whole herd | 2.5–10 min | 125 | PC level |
| Dai et al. ( | Minero et al. ( | Other training | Live | 1 | 2.5–10 min | 125 | PC level |
| Gonzalez et al. ( | Minero et al. ( | Other training | Video | 1 | 1–450 s | Mercalli scale | Other |
| Minero et al. ( | Literature; focus group; own creation | Other training | Live; Video | Whole group | 7.5–15 min | 125 | PC level |
| Arena et al. ( | Arena et al. ( | Other training | Video | Whole group | 1.5 min | 125 | PC level |
| Arena et al. ( | SQP | No training reported | Live | 1–5 | 1 min | Not reported | / |
| Barnard et al. ( | Arena et al. ( | Other training | Video | 1 - whole group | 1.5 | 125 | Term level |
| Berteselli et al. ( | Arena et al. ( | Official training (SQP) | Live | Whole group | 1 | 125 | Term level |
| Berteselli et al. ( | SQP | No training reported | Not reported | Not reported | Not reported | Not reported | (Term and PC level) |
| Cuglovici and Amaral ( | SQP | No training reported | Live | Whole group | 1 | 125 | Term level |
| Harvey et al. ( | Arena et al. ( | Other training | Video | 1 | 30 s – 2 min | 125 | PC level |
| Menchetti et al. ( | Literature; own creation | Official training (SQP) | Live | 1 | Approx. 80 s | 5-point scale | Term and PC level |
| Pedersen and Malm ( | Own creation (consulted QBA experts) | Other training | Live | 1 | 15–45 min | 125 | Term level |
| Raudies et al. ( | Not reported; SQP | Not reported | Live; Video | Not reported | Not reported | Not reported | / |
| Shaw et al. ( | Arena et al. ( | Other training | Video | 1 | 2 min | 125 | PC level |
| Stubsjøen et al. ( | Literature; focus group; own creation | Other training | Video | 1–5 | Not reported | 125 | Term and PC level |
| Stubsjøen et al. ( | Stubsjøen et al. ( | Other training | Live; Video | 1 - whole group | 2 | 125 | Term and PC level |
| Heritier et al. ( | SQP; literature; from FCP; focus group; own creation | / | Not reported | Not reported | Not reported | Not reported | Term level |
| Travnik and Sant’Anna ( | Literature; own creation | Other training | Video | 1 | 12 min | 126 | PC level |
| Travnik et al. ( | Travnik et al. ( | Other training | Video | 1 | 12 min | Not reported | PC level |
| Jarvis et al. ( | Field observations; focus group; own creation | No training reported | Live | 1 | 60 min | 125 | PC level |
| Wiese et al. ( | Literature; field observations; focus group; own creation | Other training | Video | (80) | 1 min | 100 | PC level |
| Stagni et al. ( | Literature; field observations; own creation | Other training | Live | 1 | 20 min | 125 | Term and PC level |
| Delfour et al. ( | Own creation | No training reported | Live | 1 - group | Not reported | Not reported | / |
| Yon et al. ( | Literature; field observations; own creation | No training reported | Live | 1 | 1 min | 125 | Term and PC level |
| Dobrikj et al. ( | Yon et al. ( | No training reported | Live | 1 - group | 1 min | Likert scale | Term level |
| Gartland et al. ( | Literature; own creation | No training reported | Live | 1 | Not reported | 5-point scale | Term level |
| Munerato et al. ( | Not reported | No training reported | Live | 1 | Not reported | 125 | Term and PC level |
| Skovlund et al. ( | Literature; field observations; own creation | Other training | Live | 1 | 5 min | 125 | PC level |
| Nogueira et al. ( | Literature; own creation | No training reported | Video | Not reported | 20 s | 125 | Term and PC level |
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAnimal Behavior and Welfare Studies · Human-Animal Interaction Studies · Meat and Animal Product Quality
Introduction
1
Qualitative Behaviour Assessment (QBA) is currently one of the relatively few available indicators of positive welfare [see, e.g., Boissy et al. (1) and Keeling et al. (2)] and one of the few methods currently thought as directly inferring an emotional state (3). The key characteristic of QBA is that it addresses the whole dynamic animal, describing and quantifying the emotionally expressive qualities that emerge from the animal’s way of moving around its environment. Qualitative descriptors such as fearful, joyful, or energetic integrate different aspects of an animal’s demeanour and are presumed to reflect an animal’s experience of its surroundings. Thus, QBA postulates that behaviour has observable dynamic expressive qualities open to formal analysis (4).
QBA was first mentioned in literature in 2000. Based on the argument that traditional, quantitative (ethogram-based) behavioural observation methodologies may not capture information on how an animal carries out behaviour (i.e., demeanour), a qualitative approach was explored as a novel methodology for integrative animal welfare assessment (5). Qualitative approaches have been used before to identify personality traits in animal personality research, inferring on underlying constructs that are not only based on which behaviours are performed, but also on how they are performed (6). This means that instead of only quantifying certain behaviours like in traditional ethograms, QBA specifically aims to capture the quality of behaviour, i.e., the style or expressive quality. A first link between this approach and the emotional state of animals was proposed by Wemelsfelder et al. (5). In the original article, as well as in the following years to come [e.g., Wemelsfelder et al. (4, 5) and Rousing and Wemelsfelder (7)], QBA was based on Free-Choice-Profiling (FCP) methodology, in which multiple observers freely generate terms to describe animals’ behavioural expressivity, usually based on video clips. In short, in Free Choice Profiling, observers use their own words to describe the expressive quality of the behaviours they see. A group of observers observes animals (usually from video clips) and then each observer writes down descriptive terms that in his or her opinion describe best the expressive quality of behaviours observed (e.g., descriptors like curious, relaxed, fearful). Then the same observers, using their self-generated descriptor list, rate the expressivity of observed animals on a Visual Analogue Scale (VAS) ranging from ‘minimum’ (expression absent) to ‘maximum’ (expression strongly dominant). Because everyone uses different words, the data is analysed using a statistical method called Generalised Procrustes Analysis (GPA). This technique finds common patterns in the ratings, despite the differences in vocabulary (4, 5, 8).
In a widely cited literature review on measuring positive emotions in animals (1), QBA is mentioned as one potential methodology to measure positive emotional state of animals and for potential inclusion in welfare assessment protocols, although the authors also highlight the general problem of validating such indicators of positive affect. Moreover, as studies suggested good reliability [e.g., (4, 9–11)] along with the fact that not many (feasible) indicators for the positive emotional state had been described [e.g., (1)], QBA was included as a measure of positive emotion within the Welfare Quality project (WQ) (development of feasible on-farm welfare assessment protocols). In order to enhance the feasibility of the QBA method, here, for the first time, the development of QBA fixed lists (FL), as ready-to-use lists of terms, is described (9–11). In the FL approach, the list of terms is pre-established based on existing research (sometimes further developed and refined in further studies), or on consultation with suitable species experts and stakeholders [e.g., (12)], and is not, as with FCP, chosen freely by the observer(s) who end up using the list. This standardisation means that a FL QBA can be carried out by a single observer.
After inclusion in the WQ, QBA has also been extended to other welfare assessment frameworks and protocols, especially as measure for the criterion ‘positive emotional state’ [e.g., (13–16)]. Because FL QBA is now a part of various welfare assessment schemes [e.g., WQ, Animal Welfare Indicators (AWIN), Shelter Quality (SQP)] and thus the use of FL QBA is sometimes not obvious in these studies, the exact number of studies that have used FL QBA is unclear. The methods used to develop the QBA lists vary, as does the context in which it has been used. Although some literature reviews on QBA already exist, these have so far focused on its potential use in welfare assessment protocols (17–21) and were focused on a specific group of animal species and/or on the usefulness of QBA as a tool for specific contexts [e.g., inclusion in Australian livestock industry (18) or for zoo animal welfare assessment (19)]. Moreover, these were not systematic reviews. The aim of the present review is to provide a structured overview of the application of QBA in studies using the FL approach, covering all species that a FL QBA has been developed for, as well as the uses (aims) of the method. The focus on FL QBA was chosen because this approach is most relevant for welfare assessment tools (i.e., on-farm/on-site use) due to its higher feasibility compared to FCP. Therein, we did not limit on specific purposes of use of FL QBA, but aimed to provide an overview of use of FL QBA in all areas of current research. The specific research questions we aim to answer are: (1) On which animal species has the FL QBA been carried out so far? (2) How were the FLs developed? (3) What was the aim of studies using the FL QBA? (4) How was the FL QBA applied? (5) How were QBA results analysed statistically? Providing such an overview is useful for guiding further developments of the method; the review’s focus will be on identifying methodological concerns for further discussion and research. However, as this is not a comprehensive review of QBA research, it will not address whether the listed QBA FL studies have used QBA successfully or not.
Materials and methods
2
Search methods
2.1
The electronic database Web of Science (WoS) was searched for relevant publications on QBA. This was carried out between October 2023 and February 2024. After initial scoping to detect the best possible search word combination, different searches were carried out for specific species, or groups of animals, to ensure covering the most common species within farmed, companion, experimental and wild animals. Specifically, this included cattle, buffaloes, pigs, poultry, sheep, goats, horses, dogs, cats, fish, experimental animals, as well as wild and exotic animals. For each species or animal group, two searches were applied. The first search was specified by the keywords ‘qualitative + behav + assessment’* and/or ‘QBA’, supplemented by (i.e., also including) relevant species-specific terms (for example, species-specific terms for horse consisted of ‘horse OR equ* OR pon* OR foal OR filly OR mare OR stallion OR gelding’). The keywords of the second search included ‘welfare + assessment’ and/or ‘Welfare + Quality’* and/or ‘AWIN’, along with the species-specific terms of the first search. The second search was added because QBA commonly is part of existing welfare assessment schemes such as WQ or AWIN and related publications, which in some cases, were missed by the first search. All search strings were specified to search in ‘topic’ (includes title, abstract and author keywords) with no limitation on publication year. For experimental and wild animals, one broad search was made for each category owing to the large number of species belonging to these categories (for experimental animals, specific searches for rats, mice, hamsters, rabbits, guinea pigs were also included). Finally, the species-specific searches were supplemented by a broad search without species specific terms with the search string ‘qualitative + behav + assessment* OR QBA OR welfare + assessment OR Welfare + Quality* OR AWIN’ to ensure all relevant publications were identified.
Inclusion and exclusion criteria
2.2
Title and abstract of all publications appearing in the searches were initially screened against inclusion and exclusion criteria (if the information could not be obtained from the title or abstract, the full text was screened). Publications that met all the following inclusion criteria were included in the review: (i) applied QBA as part of the study’s methodology (either focused entirely on QBA or included it as part of a larger objective), (ii) used the FL approach (either exclusively or as a second step to FCP for, e.g., term list development), (iii) published in a peer-reviewed journal, (iv) available in English, and (v) available in full. Any duplicate publications (i.e., publications that were already included) were excluded. Consequently, only original research publications utilising QBA based on the FL approach as defined by Wemelsfelder et al. (9), Wemelsfelder (10), Wemelsfelder et al. (11) (based on the respective authors’ claim and interpretation) were included. In addition publications that reported using an existing welfare assessment protocol of which QBA is an established part of (e.g., WQ, AWIN, SQP) were also included in the review, even if QBA was not specifically mentioned in the text.
Extraction of information
2.3
Selected parameters related to the studies’ methodology and results were extracted from the included publications. These parameters included information on the aim of the QBA, the animals used (e.g., species assessed, number of individuals, and life stages), information on the assessors (e.g., experience with species, QBA training received), the QBA method [e.g., term list development, time spent observing the animals, length of the visual analogue scale (VAS)], statistical methods (e.g., whether data suitability criteria were met, whether principal component analysis (PCA) was carried out and number of extracted principal components). Database searches, initial review of publications against inclusion and exclusion criteria, and extraction of parameters on publication-level were carried out approximately evenly distributed by the three authors. Fourteen randomly selected papers were reviewed independently beforehand to assure sufficient agreement in extraction (100%) between the three authors.
Results
3
The searches resulted in 193 included articles which ranged from the years 2011–2023. The last search without species-specific terms did not result in any additional articles. The studies and key results are presented in two tables: Table 1 presents the species, the setting, the life stage and the aim for which the QBA was used. Table 2 presents the experimental procedure of the same studies, i.e., origin of the QBA list of terms used, observer training, observation method and time, length of the VAS and whether QBA scores are analysed at PC or term level. Supplementary Table 1 contains the terms used in the studies on cattle, pigs and poultry, Supplementary Table 2 contains terms for sheep and goats, Supplementary Table 3 contains the same for horses and donkeys, and finally Supplementary Table 4 contains the same for dogs. The majority of the studies were done on production animals. More than half (54.4%) of the studies were on either cattle (34.7%, mainly dairy cattle) or pigs (19.7%). 11.9% of the studies were done on poultry and another 10.9% on small ruminants. On equids, encompassing working, farmed and companion animals alike, 8.8% of the studies were carried out. 8.2% of the studies were carried out on dogs (7.2%) and cats (1.0%). The remaining studies (5.2%) were done on zoo and aquaria animals or fish. No studies were found on experimental animals.
Aim of the studies and origin of QBA term lists
3.1
The aim of the studies was in most cases welfare assessment (144 papers), and QBA was often done as part of the WQ (112 papers) or AWIN protocols (14 papers) (Table 1). Seven studies did not use QBA as an indicator in the area of general welfare assessment but rather as a measure of temperament [e.g., Gois et al. (22)]. The remaining studies’ aims can be summed up as assessment of emotional state independent of general welfare assessment, for example as an evaluation of an animal’s emotional response to specific events or contexts (e.g., disease, sport events etc.).
The greater representation of studies using QBA as part of a welfare assessment protocol, is also reflected in the origin of the FL, since the term lists came from either the WQ or the AWIN protocol in 141 out of the 193 studies (see Table 2). However, in many cases the protocols were modified by adding one or several new terms [e.g., Andreasen et al. (23) and Sans et al. (24)], by reducing the number of terms overall [e.g., des Roches et al. (25) and da Silva et al. (26)] or, e.g., by exclusively using negative valenced terms (27). When the FL was not part of an existing welfare assessment protocol (identified as WQ, AWIN or SQP), list development can be categorised as being either based on the literature (terms are collected in the literature to form a list), FCP (a new list was created based on the FCP approach) or using focus groups (terms were generated in a focus group), and was often based on a combination of these. While the studies on cattle, pigs and poultry most often used standardised lists (typically from WQ, Supplementary Table 1), the case is different for goats and especially sheep (Supplementary Table 2). Although AWIN has developed lists for these species (13, 15), several of the identified studies reported using self-developed lists, and it was not always clearly described how the lists had been developed. In the studies providing details on how the lists were provided, the authors often specifically highlight the need for developing alternative lists for specific purposes [e.g., (28)] or with regard to translation issues when used in different geographical regions [e.g., (12)]. However, in general, another notable fact concerning studies using alternative lists is that altogether, these lists vary greatly in the number of terms, for example in small ruminants, some consist of just six (29) and others of 21 (30) terms. Considering the details of the FL used, please see Supplementary Tables 1–4.
Experimental procedure of use of QBA
3.2
Again, most of the studies applied the QBA according to the methodology described in the respective welfare assessment protocols. However, some differences can be detected, for example regarding the length of the VAS. Eight studies reported using a VAS of only 100 mm (instead of 125 mm as originally described in WQ, AWIN and SQP), and two studies reported a VAS of more than 125 mm. Further adaptations of the VAS were also found in the form of using, e.g., survey software formats, categorical or Mercalli scales [e.g., Menchetti et al. (31), Delfour et al. (32), and Gartland et al. (33)]. Noteworthy, three studies (34–36) used a novel method they termed continuous (c) QBA (c-QBA). C-QBA is a combination of QBA with the “Temporal Dominant Behavioural Expression” methodology (34) and enables recording shifts in individual QBA descriptors over time, i.e., the description of changes in animal behavioural expression during the observation session.
For the production animals, whole groups of animals were typically observed (following the WQ approach for these species), with only a few exceptions [e.g., Ebinghaus et al. (37, 38) and Gois et al. (22)]. For companion animals (including horses, but not donkeys) as well as for the zoo animals the reverse is true; most of the studies observed the animals at an individual level. The total number of animals included in the studies differed widely, with larger numbers of animals observed in production animals, with studies on hens and broilers including the highest numbers. The time frame observed per animal group was in most studies determined by the respective welfare assessment protocols, although different time frames also can be found (see Table 2).
Most of the studies observed the animals directly, while 29 studies used indirect (video) observation, and 11 studies used both direct and indirect observations. The level of observer training was found to vary greatly across the studies and was often not reported or was poorly described. Most studies provided no information on the level of experience with the relevant species (results not included in tables), while 89 studies reported their observers as experienced, however with large variation in provided details and in level of experience. Concerning the number of observers that performed QBA, 29 studies did not provide details, 49 studies were based on observations by one observer, 24 on observations by two observers and the remaining studies on observations by multiple observers. However, of the studies using more than one observer, only 43 reported that observer agreement was checked before data collection. A large variation is found in how observers were trained and how agreement was reached, checked and reported. This ranges from reporting of simple discussions about terms among observers, reaching an overall consensus of the whole WQ, AWIN or SQP (of which QBA is part), to utilising a few videos or spending up to multiple days or weeks on on-site training. Likewise, the analysis of observer agreement varied from descriptive evaluations to different statistical analyses, in which the level of interpretation also varied.
Reported statistical analysis
3.3
As shown in Table 2, 20 studies analysed the results of QBA outcomes solely on term level, 56 studies used the aggregation system of WQ, and 93 studies used a PCA for analysis (these are reported on in more detail in Supplementary Table 5). Fifteen of these 93 studies provided information on data suitability criteria. Fifty-two of the studies retained two PCs to explain the outcomes of the QBA (as in the WQ protocol), the other studies either retained one component (two studies), three components (20 studies, with 17 studies interpreting the third extracted component further) or four components (eight studies, with five studies interpreting the third and fourth extracted principal component further). In less than half of these 93 studies, information on cut-off values for factor loadings that were used for interpretation of the respective components could be extracted, i.e., most of the included studies did not state what was interpreted as loading highly on a PC and thus which values were used for interpreting/naming a PC. In some cases, this information was not clearly reported in the material and methods section, but could be extracted from the results tables. In about a quarter of the studies, principal component loadings of above 0.4 and below −0.4 were reported as used as cut-offs for this interpretation.
Discussion
4
Overall, FL QBA has been used in a variety of species, in many different settings and contexts, with various approaches to its methodology and analysis. The majority of FL QBA studies were carried out as part of a welfare assessment protocol for farm animals. While a large variation in species is evident, the literature search yielded no results on experimental animals. This is somewhat surprising, as the method aligns with other qualitative approaches used in experimental animals, such as those included in some forms of pain grimace assessments [e.g., in rabbits: Benato et al. (39)]. Moreover, using the method in experimental animals may aid in substantiating the validity and reliability of QBA, as laboratory settings typically offer more controlled environments [e.g., Calisi and Bentley (40) compared to, e.g., on farm or in zoos]. Overall, there is a variation in a number of factors that are likely to affect the outcome of a QBA and its meaning. Differences in the conditions under which they were observed (e.g., filmed or live), choice of terms in FL and statistical analysis, makes it difficult to compare the results of the current studies even on the species-level.
Aim of the studies
4.1
The original development of QBA was aimed at the evaluation of welfare (4, 5, 20), arguing that its whole-animal expressive information could make a unique contribution to scientific welfare assessment. Multiple validity and reliability studies on the FCP approach were carried out resulting in a generally proven efficacy of the methodology [reviewed by Wemelsfelder (8)]. Likewise, the high feasibility owing to its rapid assessment and ease of implementation [e.g., (9–11)], compared to other methods of assessing the emotional state [such as the cognitive bias test; Crump et al. (41)] is a clear advantage of QBA. These advantages likely contribute to QBA being included as a customary part of various welfare assessment protocols. The first FL developments were specifically carried out for inclusion in welfare assessment protocols for farm animals (9–11). With this development, it is not surprising that the far most common use of FL QBA in the included studies was identified as general welfare assessment, and predominantly as part of the established frameworks of WQ, AWIN and SQP, belonging to the welfare criterion ‘positive emotional state’ [e.g., Botreau et al. (42)]. Also outside such larger protocols, and following some concern for QBA being at risk for subjectivity due to the reliance on human observers (43), it is generally recommended not to use Fl QBA as a stand-alone indicator for welfare assessment but to combine and cross-validate it with other indicators, as for example Andreasen et al. (23) could not validate QBA as stand-alone indicator for welfare assessment.
Despite this focus on general welfare assessment, the FL QBA has by now been used for a variety of aims. In fact, the second-most common use was its application in specific contexts, mainly to assess emotional reactions to certain events, such as intrusive sampling and capture of, e.g., salmon (44) and pampas deer (45), calf-roping events during rodeo (46), agonistic social encounters in pigs (47), dogs’ interaction with humans during canine-assisted interventions (48) and sport events (28, 49). In these studies, FL QBA was mainly applied to investigate potential impacts of such events on animals’ emotional states and how this might affect their welfare. Further aims included temperament assessment. In general, the various aims showcase a broad and flexible usage of the method within the context of assessment of emotional state as also suggested by Boissy et al. (1). Therein, it should specifically be noted that in comparison to other methods of assessment of emotional state, QBA takes the whole-body language into account (4) instead of relying on separately measuring specific mimics, gestures or body postures [e.g., ear position, play and all grooming in cattle (2)].
Origin of QBA lists
4.2
Because the descriptors that constituted the lists developed for WQ and AWIN were not necessarily appropriate or optimal for other types of situation and contexts, alternative lists were developed for other purposes such as the study of sick animals (25), human-animal relationship tests (37, 50), mother-young interactions (51) and sport competitions (in contrast to the evaluation in the normal husbandry environment) (28). Moreover, it should be noted that translation and cultural interpretation issues might arise concerning the descriptors which might make development of lists for use in specific geographical regions necessary as highlighted by Souza et al. (12). These different circumstances, as well as different aims other than welfare assessment, justify the use of different lists. There was however, some variation in how the FL were developed. However, not all these lists were developed as originally described by a first step of FCP, and the creation of a validated FL as well as the process of FL development or justification of selection of terms used was not always clearly described [e.g., Carroll et al. (52), Kaurivi et al. (27), and Harvey et al. (53)]. It should be noted that many of these studies detailing on development and validation of FL specifically pointed out that the terms included should cover many different affective states and the reduction of terms without any further validation is thus not recommended [e.g., Arena et al. (54, 55) and Souza et al. (12)]. Therewith, there is in principle also a minimum number of terms that should be used in QBA. This study presents an overview of all FL that have been used to date, however, as pointed out, the level of validation of these lists varies.
Experimental procedure of use of QBA
4.3
Most of the studies used a VAS of 125 mm in length. In the first mention in Wemelsfelder et al. (4) of a VAS in the context with QBA, a length of 12.5 cm was described. Since then, and especially in the first developments of FL QBA for welfare assessment protocols (9–11), this length was most commonly used. The authors of this study are not aware of any justification for using 12.5 cm in QBA to have been reported in literature. In human medicine, the most common length of used VAS is 10 cm (56), which also is the second most common VAS length identified in the present literature review. In a controlled trial on patients’ VAS preferences, Sriwatanakul et al. (239) found 10 cm to be the length of the most preferred type of VAS. Another study by Seymour et al. (58) on specifically comparing different VAS lengths, reported a 10 cm continuous scale as the most appropriate, and in general, that lengths from 10 to 15 cm were suitable. Consequently, it is not clear whether 100 mm and 125 mm differ in suitability. The authors of the present study are not aware of any studies that investigate potential effects of VAS length on QBA outcomes. Such knowledge might be beneficial in order to unify the QBA methodology and for comparability across QBA studies.
In addition to the differences in VAS lengths, a few alternative measurement techniques used for QBA were identified: for example, Gartland et al. (33) rated various gorilla expressions and activity patterns on qualitative descriptors such as anxiety, curiosity, irritability, cooperation and dominance, using a categorical 1–5 scale (ranging from ‘very low’ to ‘very high’). Moreover, in some articles, the method referred to as ‘continuous QBA’ (c-QBA) was introduced. C-QBA works with individual descriptors and is based on temporal dominance of sensations (TDS) procedure, which allow raters to detect behavioural fluctuations during sessions, as opposed to the classical approach of QBA where the sum of behavioural expression is considered and rated after a session. C-QBA hence provides information on variation over time in discrete emotions, i.e., shifts in behavioural expressivity over time can be captured. C-QBA was developed for goats (35) and buffaloes (34, 36). These approaches use the same type of qualitative descriptors as QBA, based on whole-animal expressive demeanour, and so require the same type of observational assessment and therefore were included in this review. However, in contrast to the original QBA, the format in which such assessments are subsequently processed and analysed differs.
Large variation was identified in the experimental setups across studies, which is not surprising given the general variation of the purpose and context of the studies. Hence, some studies observed groups of animals while others focused on individuals. Moreover, the size of the group under observation varied largely and was not always clearly reported on. This depends naturally on the species being studied and feasibility in the settings (i.e., groups are more likely to be observed in production animals, explainable by the husbandry environment on farm). However, at this stage, it remains unclear as to what effect individual vs. group-level observations has on FL QBA outcomes. To the authors’ knowledge, this has not yet been investigated. Likewise, in the majority of the studies, direct observations, rather than video-based observations, were used. It is plausible that video observations may yield more accurate results and improved observer focus, since there may be less external disturbances (59–61). On the other hand, assessors may be less involved, meaning that their actual ability to integrate perceived details of behaviours and context and transfer that into descriptors may be limited due to not all information being transferred and the observers are also less able to react to, e.g., sudden changes on-site (which they might not even be aware of) (62–64). The question such arises on whether observers should be informed about the context or background or not when using video observations. A general disadvantage of video observation is moreover the additional costs and time involved (64) that should be taken into account with regard to feasibility of the assessment. Cooke et al. (65) investigated the difference between direct and video observations of beef cattle, and found no difference between the methods for PC1, whereas the response was less pronounced for PC2 for the video observations. Consequently, the authors of the study did not recommend using video observations for QBA. In contrast, Czycholl et al. (62) found good reliability for QBA when based on video observations, but not for on-farm assessments. A possible explanation for the difference in results is that in the study by Cooke et al. (65), the observers were in both cases looking at the same animals, whereas in Czycholl et al. (62), the live observations were carried out in the same section of the farms, however not necessarily on the same animals.
The results of this review further show a large variability in the level of observer training and experience. Tuyttens et al. (66) focused their study on observer bias and effects of observer training and proved an influence on both quantitative and qualitative methods (specifically also the QBA). Likewise, in a QBA-like study Meyer et al. (67) suggested that there were possible interactions between observer experience with dogs, and interpretation of dog behaviour (amongst others). Furthermore, Gronqvist et al. (68) highlighted the importance of experience with a species to correctly interpret potentially dangerous situations and Broom and Johnson (69) emphasised that knowledge about the behaviour of a species is important to avoid misinterpretations. Likewise, in the initial introduction of FL QBAs for welfare assessment for cattle, pigs and poultry alike, it was mentioned that for use of FL QBA, observers need to be trained and experienced (9–11). Accordingly, the most common welfare assessment protocols all highlight the need for a sufficient training level of observers before using welfare assessment protocols (of which QBA is part of). On the contrary, the first introductory publication on FCP QBA worked with observers naive to the species, relying on the general ability of humans to assess the qualitative body language signals (5). Although in principle, QBA methodology thus can work with naive observers, overall, like in quantitative behavioural observations, species experience and training of observers can improve the reliability. Guidelines regarding training level and requirements of species experience when using FL QBA could be helpful with regard to a unification of the literature and thus an enhanced comparability and therewith enable the possibility of drawing better conclusions about the reliability and validity of the methodology and potential influence on results by specific settings.
Reported statistical analysis
4.4
Three main statistical approaches for retrieving FL QBA outcomes were identified: (1) utilising mm values on term level, (2) subjecting the QBA scores to a PCA and (3) aggregation following the WQ approach, based on expert opinion and pre-existing data. The latter was usually applied when the FL QBA was used as part of an existing WQ protocol. It should be noted that in the very first publications on QBA (which are carried out as FCP), statistical analysis was carried out by Generalised Procrustis Analysis (GPA) prior to PCA (4, 5). In the first publications concerning the development of a FL approach in the bounds of the WQ project, results were analysed on term level as well as by PCA (9–11). However, the respective authors argued that a PCA may be the most suitable approach to analysing QBA. Identifying principal components (PCs) on which the terms have a certain loading may help in a more valid interpretation with regard to, e.g., observer agreement. Thus, although the analysis based on term level was presented in those studies (9–11), the authors argued that conducting a PCA would provide more reliable results. This would mean that the analysis based on term level, which occurred in 20 studies cannot be seen as the most appropriate analysis.
Regarding PCA, it should be noted that the data needs to meet certain prerequisites such as certain sample size requirements and interval-level measurement for this statistical method to be applied. In textbooks, it is described that each PC should at least have an eigenvalue of >1 (Kaiser-Guttman criterion), a clear break in eigenvalues is seen between the PCs (scree-test) and a certain amount of variance of the data set is actually explained by the extracted PCs. Additionally, there is the interpretability criterion with regard to variables loading highly on the extracted PCs (70, 71). A further useful parameter to assess the data suitability is the Kaiser-Meyer-Olkin criterion, which was actually invented for factor analysis (72, 73). Looking at the results of this review, it becomes clear that only 15 of 93 studies using PCA as analysis actually tested their data for data suitability criteria beforehand. In textbooks regarding the PCA, factor loadings to be interpreted as meaningful are named as >0.4 (70) or even higher (>0.6–0.7) (74, 75). Looking at the results of this review, 19 of the studies using PCA interpreting factor loadings used cut-off values of >0.4 and three studies used cut-off values of >0.6. It is probably a matter of study aims whether the use of clear cut-off values of factor loadings or the pattern of loadings showing which descriptors contribute most to the identified PCs (70) is most suitable and how informative relatively low-value loadings then are. That said, the use of term-loadings close to zero for interpreting a PC should always be treated with caution. However, using clear cut-off values might be impossible, as, depending on the exact model used (and different statistic programmes use different models by default), exact values may differ. This is further complicated by the fact that some authors [and in some cases it may be justifiable (70)] also use a factor analysis, but interpret it as a PCA. Another noteworthy result of the present literature review is that most of the studies that conducted a PCA extracted two PCs and interpreted those further without explicit reference to the use of general rules for extraction such as scree plot or Guttman criterion (70). This may be due to the fact that studies started adapting their methodology to that of other studies without adjusting it to their own data, which is a known risk and phenomenon in science (76). Not meeting the prerequisites of data for statistical data analysis, incorrect extraction or over interpretation of relatively low values includes quite clearly the risk of misinterpretations (70). This all being said, PCA is in general a relative flexible method and over the years has been adapted to a variety of disciplines (77), so it seems well-suited also for analysis of QBA. However, the findings described highlight the need for more advice on how to correctly use and interpret multi-variate statistical techniques such as PCA for the analysis of QBA data.
The third way of statistical analysis found is via the aggregation system suggested and published by WQ (in the case of FL QBA being a part of a larger welfare assessment protocol). This aggregation system has the general aim of aggregating all the different welfare indicators (of which the FL QBA is only one) into one final welfare score of 0–100 (78). While in total, many different methods for aggregation are used, for the FL QBA, basically weighted sums are used which were obtained by expert opinion and PCA on—by nature of the studies—limited data sets that had conducted FL QBAs (9–11, 42). Those data sets were also limited to certain regions [e.g., 17 farms in Germany, three assessors: (9)], the results obtained from the PCAs on these limited data sets may not be generalizable. It is a well-known fact that small study populations easily lead to over- and under-estimations (79). A solution to overcome this in the future, due to the points raised above, would be a revision of the existing aggregation system, e.g., by joint use of the now available larger data sets of, e.g., WQ data from different working groups and countries, which also aligns with the general aim of many welfare assessment protocols (e.g., WQ) to enhance and revise the existing protocol as new knowledge arises (80).
Quality of the literature search
4.5
The high interobserver reliability between the extractors, along with the fact that only studies after 2011 were extracted [whereby the first developments of FL QBA are described in the Welfare Quality Reports in 2009 (9–11)] and that the last search without species specific terms did not result in any further articles, demonstrates the quality of this literature review.
Conclusion
5
In conclusion, the FL QBA approach has been used across many species, primarily farm animals, but also companion animals and more recently also zoo animals. However, there was during this time no QBA developed for experimental animals. FL QBA has been used for a variety of aims, however mainly for evaluating emotional state and most often as part of welfare assessment, which is also what the FL QBA was originally developed for. Different aims and settings will call for specifically tailored FL in order to strengthen reliability and validity of QBA in those settings. However, if a FL must be standardised as part of larger welfare assessment protocols, then it is advisable to clarify the context in which that list of terms can be used. A number of methodological aspects of FL QBA vary in the identified studies, ranging from using different lengths of VAS, to the evaluation of animals at group/individual level, time used for observation, training and experience level of observers and several other factors. Moreover, different statistical analyses are used, and it is identified that not always respective prerequisites for the use of those methods exist. These are aspects to consider when gathering knowledge of the current level of reliability and validity of QBA. Future studies should thus address the question whether or not there are certain conditions that must be met when applying QBA and what conditions these are, taking into account that studies have different aims and are applied in different settings, which may require a certain flexibility in using and interpreting QBA. This could answer the question whether clearer guidelines on the construction, use and statistical analysis of FL QBA are necessary and allow the potential development of such. This could then also include guidance on how and which results on FL QBA need be presented to encourage cross-study comparison.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Boissy A Manteuffel G Jensen MB Moe RO Spruijt B Keeling LJ . Assessment of positive emotions in animals to improve their welfare. Physiol Behav. (2007) 92:375–97. doi: 10.1016/j.physbeh.2007.02.003, 17428510 · doi ↗ · pubmed ↗
- 2Keeling LJ Winckler C Hintze S Forkman B. Towards a positive welfare protocol for cattle: a critical review of indicators and suggestion of how we might proceed. Front Anim Sci. (2021) 2:753080. doi: 10.3389/fanim.2021.753080 · doi ↗
- 3Rutherford KM Donald RD Lawrence AB Wemelsfelder F. Qualitative behavioural assessment of emotionality in pigs. Appl Anim Behav Sci. (2012) 139:218–24. doi: 10.1016/j.applanim.2012.04.00422915833 PMC 3417235 · doi ↗ · pubmed ↗
- 4Wemelsfelder F Hunter TE Mendl MT Lawrence AB. Assessing the ‘whole animal’: a free choice profiling approach. Anim Behav. (2001) 62:209–20. doi: 10.1006/anbe.2001.1741 · doi ↗
- 5Wemelsfelder F Hunter EA Mendl MT Lawrence AB. The spontaneous qualitative assessment of behavioural expressions in pigs: first explorations of a novel methodology for integrative animal welfare measurement. Appl Anim Behav Sci. (2000) 67:193–215. doi: 10.1016/S 0168-1591(99)00093-310736529 · doi ↗ · pubmed ↗
- 6Uher J Asendorpf JB. Personality assessment in the great apes: comparing ecologically valid behavior measures, behavior ratings, and adjective ratings. J Res Pers. (2008) 42:821–38. doi: 10.1016/j.jrp.2007.10.004 · doi ↗
- 7Rousing T Wemelsfelder F. Qualitative assessment of social behaviour of dairy cows housed in loose housing systems. Appl Anim Behav Sci. (2006) 101:40–53. doi: 10.1016/j.applanim.2005.12.009 · doi ↗
- 8Wemelsfelder F. How animals communicate quality of life: the qualitative assessment of behaviour. Anim Welf. (2007) 16:25–31. doi: 10.1017/S 0962728600031699 · doi ↗