Your hypothesis is going to come from what you are interested in learning. Maybe you could investigate the role of digital technology in the development of young people. For example, it seems to me that young people are less interested (or able?) to have face to face conversations than they were before the introduction of the smart phone. So your agent of socialization could be family and/or peers. Your hypothesis could be that the introduction of the smart phone has caused teen aged youths to speak less in person (using text messaging more) with family and friends than the teen aged youths who grew up before the smart phone. It might be hard to do a survey using a hypothesis like that, unless your survey asked the age of the respondents.
Here is a simpler one. Hypothesis - Younger people are more likely to end a romantic relationship by text message or ghosting (no formal end, they just discontinue contact) than older people are due to their use of digital technology during their formative years. You could then do a survey asking whether or not the person thinks it is ok to end a relationship by texting. Then find out their age. I would think that older people who didn't grow up in the age of texting might be more in favor of a face to face break-up than younger people would. By collecting survey data, you could get some evidence on this. What do you think of these ideas?
B.
randomization is often hard to achieve.
The answer to this question is <span>when two variables are correlated, we cannot be sure what is causing the correlation.
For example, let's there is a study that found an increase in consumption in tofu lead to an increase in breast cancer.
Even if it's true that those two really correlated (let's just assume it), we wouldn't be able to know why it is correlated without further researches.</span>
