I'll soon start to take some data (i.e. making measurements) and from the little I've seen from a physicist doing research, I believe he trashes some data. In other words he doesn't publish all the data he extracts from the laboratory. He applies some non rigorous/mathematical criterion to the data to trash and keep only what seems reasonable to him, be it for his data to agree with previous research or because the data is so unexpected that it doesn't fit in any theoretical model and it may mean he has goofed the experiment. I was wondering whether this behavior was common between physicists and if there are some non obvious cases where trashing some data is acceptable.