Unwanted bias or promising intervention effect? Dealing with measurement reactivity in controlled trials of physical activity interventions