What should be the basis for identifying which breeds should be included in Schedule 4?

Only those breeds able to be statistically shown to be more likely to attack than any other attributable factor (EG: Neutered / Chained / Owned by Beneficiary) should be included on Schedule 4.

Furthermore breed specific bans should only be implemented where a breed can be shown to be more likely to cause major harm or death than the highest risk racial group. Otherwise that racial group should have the same restrictions applied to them first.

I would also refer to the message by Associate Professor Kevin Stafford (Institute of Veterinary Animal and Biomedical Sciences) which states that:
In the late 1980s a list of breeds involved in 40 serious dogs attacks on children in Adelaide included German shepherd dogs (10), German shepherd crossbreds (5), rottweilers (7), pitbull terrier-type dogs (4), Siberian huskies (3) and one akita, doberman pinscher, labrador retriever, chow chow and Australian shepherd. Pitbull terrier-type dogs have been involved in many of the recently reported dog attacks and are the target for those promoting breed control legislation. But dog aggression was a public problem in New Zealand before this type of terrier became common and some of the breeds listed above may come under scrutiny in the near future. In a 1995 study of veterinary opinion in New Zealand, rottweilers were considered much more aggressive in the veterinary clinic than any other breed of dog. Intact male dogs are also much more likely to be involved in dog attacks than females or desexed animals.

This makes it critically important that dog breeds are assessed on evidence; not popular opinion. It is the focus on "bad breeds" which leads to the number of attacks by Labrodors and Jack Russels; both rarely focused on in "bad dog breed" lists - but frequently in the news stories of dog attacks.