700 people is more than enough to get a very representative sample: https://en.wikipedia.org/wiki/Sample_size_determination
If you're measuring the education quality at a University of a few thousand students, then 700 students (from across different faculties) is ample. If you're attempting to measure the perceived accuracy of stereotypes across society at large, then you need to show that your sample is representative of that society, as personal backgrounds and personal experience are likely to play large factors in determining how others are perceived.
The sample you pick from matters. To give an example, imagine asking 700 people in the ethnic majority of a society about the effects of racism in that society.