Treatment structure: 2 supplements for pregnant dogs(treatment(trt) A and B)

response variable is the number of pups born dead

The design is a randomized block design (block was weight of pregnant dogs)

I have tested for normality and homogenous variance using SAS software. The data turned out not satisfy both assumptions.

I have tried transforming the data but no improvements.

Thus, I am thingking of using nonparametric methods.

Would a chi square or kruskal wallis test be appropiate if not what type of analysis would be correct. I am using the SAS software

Data looks like this:

block trt borndead

1 A 1

1 A 0

2 A 0

2 A 3

3 A 4

3 A 5

4 A 0

4 A 0

5 A 1

5 A 0

1 B 3

1 B 2

2 B 1

2 B 2

3 B 1

3 B 1

4 B 1

4 B 0

5 B 0

5 B 0

But there are more data

I hope someone can show me the correct way

thank you in advance and God bless