First, the purpose of induction is to gather specific facts and examine them in order to arrive at a new conclusion which is not immediately apparent before you start your research process; thus determining the conclusion in advance is somewhat problematic.
If you want to investigate a claim that, e.g. "all dogs are quadrupeds", you do so inductively by observing a sufficiently large quantity of dogs and counting their legs. For an inductive argument to be completely certain, as opposed to probabilistic, you could attempt what is known as "complete induction" (looking at all possible examples of a class) -- but that is rarely done. Instead, scientists (including social scientists) develop statistical tests to determine "random samples" to avoid the logistical issues of complete induction (how could you actually observe every dog alive, much less all that have ever lived?)