100 people of all ages. If they pick any number of people at the city park, that's considered a bias, because what if the typical city park goers are in their 20s-30s? what if they're made up of ONLY dog owners? these are irrelevant details that shadow away from the true, useful, unbiased results.
Secondly, if they choose a smaller amount of people, they'll get a smaller number of responses, which could lead their study astray.