Why are push polls criticized by responsible scientific pollsters?

Typically a push poll is not designed to merely collect information from those polled but instead to attempt modifying the outlook of those polled or at least increase the probability of obtaining a desired response. As such, the way the poll is present inherently displays a bias and in all likelihood skews the results. For example, if a poll stated "Do you consider yourself a vegetarian or vegan?", the results would probably be far more accurate than if presented by "Do you eat foods that require animals be slaughtered?"