If recruited patients in a study had a high attrition rate in one of the treatment arms, leaving only those patients who responded well to treatment, what kind of bias would this represent?