This depends on whether your initial topic setting is reasonable. If it is reasonable, it is not a question of the topic, and the topic should not be deleted.

If it is determined that it is a problem with topic setting, you can delete the topic. It depends on which part of your investigation has gone wrong, just correct the part with the problem.

If you want to study the influence relationship, such as the influence relationship between cognition and attitude on behavior, you can consider using regression analysis at this time. Linear regression is to fit a series of influencing factors and results into an equation (there is no equation in nonlinear regression, and the human brain is like a nonlinear regression), and then we can make predictions by applying this equation to other similar events. The so-called return is for an ideal.