Effects of Interview Mode on Assessments of Erectile and Ejaculatory Dysfunction among Men with Benign Prostatic Hyperplasia (BPH)

Abstract

In a randomized experiment (N = 249; age 50 + years), this study examined if self-reports of erectile dysfunction (ED) and ejaculatory dysfunction (EjD) symptomatology were influenced by the mode of interview administration (computer-assisted self-interview [CASI], audio computer-assisted self-interview [ACASI], or computer-assisted personal interview [CAPI; involving an interviewer]). This study also examined if mode moderated person variables hypothesized to impact self-reports (social desirability, age, or depressive mood). No main or moderating effects of mode were found for self-reports of EjD symptoms. However, mode effects on reports of ED symptoms were observed, and these moderated age and social desirability effects on self-reports. Significantly more older (relative to younger) men reported high levels of ED symptoms when interview administration was by a live interviewer (CAPI) than with self-administration. Alternatively, significantly more younger men reported high levels of ED symptoms when administration was by an interviewer (CAPI) or by ACASI (vs. CASI). The Mode × Social Desirability effects were complex (see the Discussion section), showing hypothesized effects under ACASI and CAPI conditions, but an opposite effect under the CASI condition. The stability of self-reported ED symptoms did not vary by mode (based on test–retest comparisons); test–retest was significantly higher for EjD symptoms within the ACASI condition. The impact of mode of administration on self-reports of ED/EjD symptoms is less predictable and dramatic than one might conclude from prior research with other types of self-report outcomes. The findings are consistent with a small, but growing, body of studies that illustrate highly situational effects of interviewing, which may depend on a complex interplay between modes, person variables, and the interview topic/target items. Self-administered methods, in particular, may not be a universal solution to response bias.

Acknowledgments

This work was supported by funding from Sanofi-Aventis, Bridgewater, NJ. Data collection was conducted by The Henne Group of San Francisco. Data analysis was conducted by Jesse Canchola, Jason Chang, and Dr. Lance Pollack (Health Survey Research Unit at the University of California, San Francisco). We give special thanks to Sharon Jacobs, Josephine Sallano, and Arkady Shpilsky (all formerly of Sanofi-Aventis in the United States) for contributions in supporting research on benign prostatic hyperplasia. Consultation was provided in the early stages of this research by Dr. Richard Brand, Professor Emeritus, University of California, Berkeley and University of California, San Francisco.

Notes

Note. Percentages may not equal 100 due to rounding. Differences between modes on the variables are all nonsignificant (all ps > .30), indicating that the randomization was successful. CASI = computer-assisted self-interview; ACASI = audio computer-assisted self-interview; CAPI = computer-assisted personal interview.

^a n = 82.

^b n = 87.

^c n = 80.

Note. 0 = less than the median, 1 = greater than or equal to the median. Hosmer–Lemeshow goodness-of-fit test, p = .12. The benign prostatic hyperplasia (BPH) severity, age, social desirability, and depressive mood two-way interactions with interview mode were nonsignificant. OR = odds ratio; CI = confidence interval; CAPI = computer-assisted personal interview; ACASI = audio computer-assisted self-interview; CASI = computer-assisted self-interview.

^a Erectile function scale scores were reversed so that higher scores indicated greater dysfunction.

^b BPH severity: severe versus moderate (OR = 2.04, CI = 1.01–4.12; p = .050).

^c Social desirability was rescaled so that a one-unit increase represented a four-point increase on the social desirability scale.

^d Depressive mood was rescaled so that a one-unit increase represented a four-point increase on the short form of the Center for Epidemiologic Studies–Depression scale.

Note. 0 = less than the median, 1 = greater than or equal to the median. Hosmer–Lemeshow goodness-of-fit test, p = .75. The main effects for age, mode, and social desirability are not directly interpretable because of interactions in the model, and they are not presented. OR = odds ratio; CI = confidence interval.

^a Benign prostatic hyperplasia (BPH) condition: severe versus moderate (OR = 1.81, CI = 0.81–4.07; p > .10).

^b Depressive mood was rescaled so that a one-unit increase would represent a four-point scale increase on the short form of the Center for Epidemiologic Studies–Depression scale.

^c Social desirability was rescaled so that a one-unit increase would represent a four-point scale increase in social desirability.

^d See Table 4. ^eSee Table 5.

Note. ED = erectile dysfunction; OR = odds ratio; CI = confidence interval; ACASI = audio computer-assisted self-interview; CAPI = computer-assisted personal interview; CASI = computer-assisted self-interview.

Note. ED = erectile dysfunction; SD = social desirability; OR = odds ratio; CI = confidence interval; ACASI = audio computer-assisted self-interview; CAPI = computer-assisted personal interview; CASI = computer-assisted self-interview.