correlation with repeated measures

I'm analyzing data from an experiment with a repeated measure design (two conditions). In the experiment, couples eat dinner together. In the control condition the couple had as much time as they usually need to eat dinner. In the experimental condition, they had 50% more time. The independent variable is time (minutes). The dependent variable is quality of social interaction. We found group differences: in the experimental condition (50% more time) social interactions were more positive than in the control condition. Now, I am wondering whether there is a linear relationship between social interaction and time. Can we say: the more the better? A simple regression analysis with time spent eating together as predictor and social interaction quality as outcome is problematic, because due to the repeated measure design, data points are not independent from each other. I am looking for a way to analyze this relationship while taking dependencies into account.