The more people you survey, the higher the confidence that the questionnaire is representative of the entire population, or the 'confidence interval'. Since the population was randomly sampled & non-responses don't correlate to the topic questions, you can assume that the 90 students are representative of the entire population with +- some confidence interval.
