Premium
Retrospective Self‐Report of Alcohol Consumption: Test‐Retest Reliability by Telephone
Author(s) -
Cohen B. Beth,
Vinson Daniel C.
Publication year - 1995
Publication title -
alcoholism: clinical and experimental research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.267
H-Index - 153
eISSN - 1530-0277
pISSN - 0145-6008
DOI - 10.1111/j.1530-0277.1995.tb01595.x
Subject(s) - intraclass correlation , interview , telephone interview , reliability (semiconductor) , test (biology) , psychology , alcohol consumption , clinical psychology , statistics , psychometrics , mathematics , alcohol , social science , power (physics) , paleontology , physics , biochemistry , chemistry , quantum mechanics , sociology , political science , law , biology
The Timeline Follow‐Back (TLFB) is an interview technique for obtaining detailed retrospective self‐reports of alcohol consumption with excellent reliability for various composite variables when both administrations are in person. Because the telephone offers practical advantages over face‐to‐face interviewing for follow‐up assessments in longitudinal studies of problem drinkers, this study was undertaken to compare the test‐retest reliability of a 12‐week TLFB interview when the second administration was by telephone to that when the second interview was in person. In addition, because the reliability of the TLFB has been previously assessed using composite variables, we examined the reliability of the TLFB at the item level. Research participants were 30 adult medical patients who drank frequently, and 75 college students who were problem drinkers. Test‐retest reliability as measured by intraclass correlation was generally high, 0.79 or greater for the number of days of drinking > 6 standard drinks, 0.90 or greater for the number of abstinent days, and 0.80 or greater for the greatest number of drinks consumed on any 1 day, in both the most recent 4‐week interval and in the entire 12‐week interval. Test‐retest correlation coefficients for composite variables derived from the interview data were not systematically affected by whether the second interview was in person or by telephone. Furthermore, item‐level correlations were also substantial. Findings support the use of the telephone for follow‐up interviews, potentially reducing costs of longitudinal studies and facilitating multisite studies with centralized data collection, and lend further general support to the reliability of the TLFB.