In 2003, The National Research Council estimated that obesity occurs in 25% of dogs and cats in Westernized societies. (Obesity can be determined by squeezing an animal in the lower chest area and seeing if the ribs can be felt.) Perhaps the 25% estimate was obtained by calling homeowners and, if they had a cat or dog, asking them to see if the animal's ribs could be felt by squeezing. Alternatively, the estimate could have been obtained by asking veterinarians to report what percentage of the animals they treat could be classified as obese, using the above method. Keeping in mind that bias may arise either from a nonrepresentative sample or from inaccurately assessing a variable,
a. Which approach is more likely to lead to a nonrepresentative sample: calling homeowners or asking veterinarians?
b. Which approach is more likely to lead to inaccurate assessments: calling homeowners or asking veterinarians?