This paper has two major goals. The first is to examine how people with differing demographic characteristics (education, income, sex, age) may exhibit differences in a variety of respondent behaviors, including response non-differentiation, response order effects, and response style (extreme, central tendency) in web-based surveys. We also explored the consequences of these respondent behaviors on data quality. The second goal is to advance our understanding on how psychological constructs (need for cognition, need for evaluation, and social-orientation variables) might mediate the effects of demographics on non-differentiation and response style. For the first goal, in four large datasets based on web-based studies we analyzed how demographic variables were related to response behavior (including completion time for two of the studies). In these studies we found some indication that non-differentiation was more likely to occur with respondents with less, rather than more education. We also found that the extreme response style was somewhat more associated with less education as well. Sex and age of respondent were also somewhat related to response style, but not as clearly with non-differentiation. The fifth study had 3933 respondents. Respondents completed a series of items designed to measure a variety of psychological constructs and then were presented with a series of tasks involving ratings across a variety of topics. With regard to the psychological measures, we found that the effects of demographics on response effects were moderated by the need for cognition – those who had more education expressed a higher need for cognition, and were less likely to engage in non-differentiation. We also found a number of effects for some very specific variables on speed of completion and response style.