Dr. Harrison wants to test the effectiveness of the support group he runs for undergraduates who have drinking problems. He recruits a group of students who have been referred to the counseling center. He randomizes them to two groups: a structured-session group and an unstructured-session group. The structured-session group attends 10 structured support-group sessions at 8:00 a. m. on Fridays that are facilitated by Dr. Harrison. The unstructured-session group attends 10 unstructured meetings at 3:00 p. m. on Wednesdays that are facilitated by members of the group. Several of the participants in the structured-session group stop attending after just a couple of sessions. All the members of the unstructured-session group attend their meeting regularly. At the end of the 10 weeks, Dr. Harrison measures drinking behavior of the people who are still attending the groups. He finds the drinking behavior of people who attended all 10 structured-group sessions is less than the drinking behavior of people who attended all 10 unstructured-group sessions. Which kind of threat to internal validity should Dr. Harrison be concerned about?