Truth and Memory: Linking Instantaneous and Retrospective Self-Reported Cigarette Consumption.

Area of Science:

• Behavioral Science
• Psychology
• Public Health

Background:

• Smoking cessation studies often rely on the time-line follow-back (TLFB) method for retrospective daily cigarette consumption data.
• TLFB is known to be susceptible to recall and digit preference biases, limiting its accuracy in measuring daily consumption.
• Ecological momentary assessment (EMA) using electronic diaries allows for real-time recording of cigarette consumption, potentially overcoming retrospective recall limitations.

Purpose of the Study:

• To analyze and compare cigarette consumption data collected simultaneously using TLFB and EMA methods.
• To develop and evaluate a statistical model explaining the biases in TLFB data as a process of mis-remembering and rounding.
• To identify factors associated with greater discrepancies between recalled and recorded smoking.

Main Methods:

• Collected simultaneous data on daily cigarette consumption from 236 active smokers using TLFB and electronic diary-based EMA.
• Defined a statistical model incorporating fixed and random effects to describe the two-stage process of mis-remembering and rounding in TLFB data.
• Employed Bayesian methods for model estimation and evaluated model adequacy using histograms of imputed latent remembered cigarette counts.

Main Results:

• Both mis-remembering and heaping (rounding) significantly distort self-reported cigarette counts obtained via TLFB.
• Higher nicotine dependence, white ethnicity, and male sex were associated with greater recalled smoking relative to EMA-recorded consumption.
• The developed statistical model effectively captures the biases inherent in retrospective smoking recall.

Conclusions:

• EMA provides a more accurate measure of daily cigarette consumption compared to TLFB due to the absence of recall bias.
• A statistical model can quantify the extent of mis-remembering and rounding biases in retrospective smoking data.
• Understanding these reporting biases is crucial for accurate smoking behavior research and may be applicable to other self-report data.