Proper experimental design is an important part of the planning phase of any research project. Let's assume that your investigation into the domestic division of labor will be largely survey-based. You will interview family members about their participation in household duties in order to collect your data. Therefore, you should design your survey to collect data which is as accurate as possible.
Three problems you are likely to face are confirmation bias, recalled frequency error, and complications from differences in cultural beliefs.
Confirmation bias is when the researcher or team conducting the survey already has a deep set belief about the issue. These beliefs result in the researcher only collecting data which confirm his or her preconceived beliefs. This can go so far as to result in leading questions.
Recalled frequency error occurs when a person is asked to recall how many times they performed a particular task within a time frame. For example, 'How many times did you wash the dishes last week?' The larger the time frame and the more insignificant the task, the less likely the subject is to remember the number accurately.
With respect to cultural beliefs, it will be important to ask questions which gather objective data in addition to the subject's perceptions. Different cultures will have different definitions of what qualifies as household labor that will need to be taken into account during the analysis.