Original Articles
Comparison of Diary and Retrospective Measures for Recording Alcohol Consumption and Sexual Activity

https://doi.org/10.1016/S0895-4356(97)00262-XGet rights and content

Abstract

This study examines issues in the measurement of HIV risk factors, using daily diaries to collect data on both alcohol use and sexual behavior. Seventy-nine adolescents and young adults recruited from an STD clinic and from a university campus were studied. Participants gave daily reports of their drinking, drug use, and sexual activities for 4 weeks. Respondents then completed a retrospective questionnaire asking about the frequencies of these behaviors during the preceding period. Diary reports of behavior were strongly correlated with retrospective reports. More frequent drinking was reported on the diary measure than the retrospective measure, and this discrepancy was larger for more frequent drinkers. Frequency of sexual activity was overreported on the retrospective measure only among adolescents. Errors in the measurement of alcohol use, sexual behavior, or their co-occurrence could affect estimations of the relationship of alcohol use to sexual behavior. The types of error inherent in these measures may differ, resulting in different tendencies toward over- or underreporting of alcohol use and sexual behavior, depending on frequency of the behavior and the characteristics of the subject population.

Introduction

As the second decade of the AIDS epidemic continues, it is evident that the primary defense against the spread of this disease is prevention of the behaviors that result in HIV transmission. One factor proposed to contribute to sexual risk-taking is the use of alcohol or other drugs with sex. Some research has suggested that using alcohol or drugs with sex increases the likelihood of unprotected intercourse; however, contradictory findings exist (see [1]for review).

Research on alcohol use and high-risk sex has used a wide variety of measurement strategies. In the absence of noninvasive and ecologically valid biological or observational techniques, measurement of both behaviors relies on self-reports. Investigating whether alcohol use with sex increases the probability of high-risk sexual activities requires accurate measures not only of alcohol use and of sexual behavior, but of the co-occurrence of the two.

One promising approach to the measurement of alcohol use, sexual activity, and their co-occurrence is a diary methodology, in which drinking occasions, sexual encounters, and the presence of alcohol in sexual encounters are recorded daily. This article describes a study in which daily reports of both alcohol use and sexual behavior were collected, and agreement of diary reports to retrospective reports of alcohol use and sexual behavior were assessed.

Diaries have been used to gather reports of general health-related behaviors 2, 3, alcohol consumption 4, 5, sexual behavior 6, 7, 8, mood states [9], and dietary intake [10]. Due to their expense and the difficulty of implementation, the use of diaries as a method of data collection in epidemiology has been limited [11]. However, because diaries are assumed to be more accurate than retrospective reports, they are not uncommonly used to validate retrospective questionnaires [11]. As a data collection technique, the diary method has both advantages and disadvantages when compared to retrospective questionnaires 3, 12, 13, 14. Diaries generally result in greater frequencies of reported behavior, especially of relatively commonplace behaviors 3, 12, and these higher levels are often assumed to be more accurate [3]. Because participants record events on the day they occur, a diary method can reduce forgetting and “telescoping,” which plague retrospective interviews and may reduce reliability of frequency estimates 11, 15. Retrospective interviews often ask respondents to abstract their behavior (for example, by stating the usual frequency of a certain behavior), rather than eliciting specific events; this strategy can lead to error (see 11, 16, 17, 18). Moreover, “usual” or “average” behavior may not correspond well to actual behavior for any particular interval 19, 20, especially for behaviors that are irregular or infrequent. Summary questions may elicit modal behavior rather than a true average 5, 21, 22. In contrast, a diary method, if implemented for a sufficiently long period, can capture variations in behavior.

Despite these advantages, a diary method has its drawbacks. The investment of time and effort required can lower participant recruitment, retention, and compliance [13], resulting in a subject sample that is likely to be unrepresentative of the population of interest 17, 19, 23. A diary is subject to longitudinal confounds such as fatigue (in which reports of activities drop as subjects tire of the procedure) and reactivity (in which the behavior changes as subjects become more aware of it; see 3, 24, 25, 26, 27but see 19, 28, 29for contrary findings). In most diary studies, participants are provided with forms to complete themselves, and there is no guarantee that they are actually completing them each day. If participants instead fill out the diary forms at a later date, the advantage of this method in terms of accurate recall is compromised 17, 19, 20. Diary studies are also expensive, due to the training and monitoring of respondents and the complexity of the coding procedures [11].

Although accurate measurement of sexual behavior is crucial for research on fertility, sexually transmitted diseases, and HIV transmission, there have been few attempts to assess the reliability and validity of measures of sexual behavior (largely retrospective questionnaires). The AIDS epidemic has highlighted the importance of these measurement issues, and recent articles have described some of the methodological problems of research on sexual behavior 30, 31, 32.

Diaries have been used to measure sexual behavior in a number of studies 33, 34, 35, 36, 37, 38, 39, 40, but their use in studying reliability and validity of self-report measures has been rare. The findings of the studies comparing retrospective reports to diary reports are equivocal 7, 8, 41, 42. Correlations between diary measures and retrospective measures of frequency of sexual behavior appear to be high for general frequency of intercourse in men [8], but extremely variable for specific sexual practices [41]. Moreover, the demonstration of both high correlations between measures and comparable mean levels is necessary to provide complete information on the correspondence of diary and retrospective measures, and only one study [8]reported both correlations and mean differences for all variables. These studies suggest that a diary procedure might yield more accurate information about sexual behavior than retrospective measures, but several issues remain to be addressed, including gender differences and the reliability of reports of specific sexual practices such as condom use.

In some ways, measurement issues in studying alcohol use are similar to those involved in studying sexuality. Questions about both activities are highly personal and are assumed to result in underreporting rather than overreporting 32, 43, 44. For many adults, both alcohol use and sexual activity are common enough that individual instances may become indistinguishable in memory. Frequency reports are then likely to be constructed not by counting discrete events but by using some kind of inference rule 18, 32. For some, drinking and sexual activity are much rarer events. This variety in frequency implies that some methods of asking questions about these behaviors may have different error for different people.

The use of diaries in measuring drinking behavior has a tradition dating back to the 1940s [45]. Correlations between retrospective measures and diary measures are quite high for general volume of drinking 4, 5, 46, 47, average quantity consumed on drinking days 4, 29, and number of drinking days 4, 5, 29. With some exceptions 25, 26, 48, diary measures yield a greater number of reported drinking occasions, and higher estimated consumption, than do retrospective measures. Average quantity consumed per occasion or per day does not differ between diary and retrospective measures 19, 20, 25, 26, 49, 50, 51. Finally, the degree of over- or underestimation of consumption appears to vary with usual drinking frequency. On retrospective measures, compared to diary measures, more frequent drinkers are more likely to underestimate their consumption 4, 17, 50, 51(but see [5]for contrary findings), and less frequent drinkers tend to overestimate their consumption 17, 50.

Measurement issues have been the focus of much discussion among researchers studying sexual behavior 30, 32and alcohol use 16, 22, 27, 52, 53, 54, 55, and both fields have seen ongoing calls for more research on these issues 22, 30, 32, 54. A daily diary method might illuminate potential sources of error in the more commonly used, and easier to implement, retrospective measures of behavior.

In this article, we describe a study that used both diary and retrospective methods to assess alcohol use and sexual behavior. We address issues in the measurement of alcohol use and sexual behavior by examining the agreement of diary reports and retrospective reports of alcohol use and sexual behavior.

Section snippets

Subjects

Subjects were 35 males and 44 females aged 16 to 38 (mean age = 24; median age 22). Twenty-four percent of subjects were aged 16–20, 49.4% were 21–25, 8.9% were 26–30, and 17.8% were 31 or older. The majority of the sample were white (64%), with 25% African American, 5% Asian American, 1% Hispanic, and 5% other racial/ethnic backgrounds. Sixty-one participants (77.2%) had never been married, 11 (13.9%) were married, six (7.6%) were separated or divorced, and one (1.3%) was widowed.

Subjects were

Sexual Intercourse

Number of occasions of sexual intercourse, as assessed by the diary measure, ranged from 0–23.2 Table 1 shows the mean and standard deviation of the number of times sexual intercourse was reported on the diary and retrospective measures, the mean difference between retrospective and diary measures, the confidence interval for this mean difference, and the correlation of the two measures. Men and women reported sexual activity with equivalent frequency, and did not differ in the agreement of

Study Findings on Drinking Behavior

Our results showed that diary reports yielded 39% more drinking occasions than retrospective reports, and that the magnitude of this difference was larger for more frequent drinkers than for less frequent drinkers—findings consistent with most previous research 4, 17, 50. This greater discrepancy among heavier drinkers may arise from the greater error range for more frequent drinkers, state-dependent learning processes that interfere with recall of drinking occasions, social desirability [51],

Conclusion

The findings from the present study suggest that the types of error inherent in these measures may differ, resulting in different tendencies toward over- or underreporting of alcohol use and sexual behavior. Moreover, degree of overreporting or underreporting varies by frequency of the behavior, such that error is likely to be different in different types of subject samples. In studying the role of alcohol in contributing to high-risk sex, researchers should attend to the characteristics of

Acknowledgements

Data collection was supported by a grant from the Alcohol and Drug Abuse Institute, University of Washington, to Mary L. Gillmore. Preparation of this article was supported by a Research Scientist Development Award (K02 AA00183) from the National Institute on Alcohol Abuse and Alcoholism to Barbara C. Leigh.

References (65)

  • BC Leigh et al.

    Substance use and risky sexual behavior for exposure to HIVIssues in methodology, interpretation, and prevention

    Am Psychologist

    (1993)
  • S Sudman et al.

    Health Care Surveys Using Diaries

    (1980)
  • LM Verbrugge

    Health diaries

    Medical Care

    (1980)
  • ME Hilton

    A comparison of a prospective diary and two summary recall techniques for recording alcohol consumption

    Br J Addiction

    (1989)
  • SM Harvey et al.

    Alcohol consumption, female sexual behavior and contraceptive use

    J Studies Alcohol

    (1986)
  • PP Hornsby et al.

    Validity of questionnaire information on frequency of coitus

    Am J Epidemiol

    (1989)
  • AE Reading

    A comparison of the accuracy and reactivity of methods of monitoring male sexual behavior

    J Behavior Assess

    (1983)
  • RJ Larsen et al.

    Affect intensity and reactions to daily life events

    J Personality Social Psychol

    (1986)
  • JC Witschi

    Short-term dietary recall and recording methods

  • BK Armstrong et al.

    Principles of Exposure Measurement in Epidemiology

    (1992)
  • KJ Roghmann et al.

    The diary as a research instrument in the study of health and illness behaviorExperiences with a random sample of young families

    Med Care

    (1972)
  • AA Stone et al.

    Measuring daily events and experiencesDecisions for the researcher

    J Personality

    (1991)
  • H Tennen et al.

    Personality and daily experienceThe promise and the challenge

    J Personality

    (1991)
  • SE Feinberg et al.

    Large-scale social surveysPerspectives, problems, and prospects

    Behavioral Science

    (1983)
  • T Alanko

    An overview of techniques and problems in the measurement of alcohol consumption

  • P Lemmens et al.

    Weekly recall and diary estimates of alcohol consumption in an general population survey

    J Studies Alcohol

    (1988)
  • N Schwarz

    Assessing frequency reports of mundane behaviorsContributions of cognitive psychology to questionnaire construction

  • MB Sobell et al.

    Is self-monitoring of alcohol consumption reactive?

    Behavior Assess

    (1989)
  • FV O'Callaghan et al.

    Young adult drinking behaviorA comparison of diary and quantity-frequency measures

    Br J Addiction

    (1992)
  • JC Duffy

    The Measurement of Alcohol Consumption in Sample Surveys

    (1982)
  • RK Fuller et al.

    Drinking practices recorded by a diary method

    Q J Studies Alcohol

    (1972)
  • EK Gerstel et al.

    The reliability of drinking estimates obtained with two data collection methods

    J Studies Alcohol

    (1980)
  • Cited by (75)

    • Momentary versus retrospective reports of alcohol or cannabis use, sexual activity, and their co-occurrence

      2021, Addictive Behaviors
      Citation Excerpt :

      Specifically, participants tend to underreport drinking alcohol or heavy drinking in retrospective surveys compared with daily surveys (Monk, Heim, Qureshi, Price, & Ryabinin, 2015; Patrick & Lee, 2010), whereas sexual activity is retrospectively overreported (Horvath, Beadnell, & Bowen, 2007; Willis & Jozkowski, 2018). Such discrepant recall biases across these behaviors can increase random error and consequently decrease the statistical power of an analysis—making it more difficult for studies using retrospective surveys to identify otherwise significant correlations between substance use and sexual activity (Leigh, Gillmore, & Morrison, 1998; Leigh & Stall, 1993). In addition to distinct methodologies and samples being used to separately assess the recall biases of substance use and sexual activity, a few researchers have investigated these biases for each behavior—as well as their co-occurrence (i.e., substance-involved sexual activity)—within the same study.

    • Perceived control of anxiety as a moderator in the relationship between anxiety sensitivity and problematic alcohol use among adolescents

      2018, Journal of Adolescence
      Citation Excerpt :

      This single-item measure of binge drinking frequency has been used extensively in the adolescent literature (Behnken, Le, Temple, & Berenson, 2010; Long Foley, Altman, & Wolfson, 2004; Miller et al., 2007). This face-valid, single-item metric of binge drinking from the YRBS has been shown to have adequate test-retest reliability (TCC = 0.89; Rosenbaum, 2009), with similar items showing adequate convergent validity with more thorough assessments of 30-day drinking rates (e.g., 61% concordance with daily diaries; Leigh, Gillmore, & Morrison, 1998). The Childhood Anxiety Sensitivity Index (CASI; Silverman, Fleisig, Rabian, & Peterson, 1991) was used to assess anxiety sensitivity.

    View all citing articles on Scopus
    View full text