Background: Ecological momentary assessment (EMA) is a methodology involving repeated assessments/surveys to collect data describing respondents’ current or very recent experiences and related contexts in their natural environments. The use of EMA in audiology research is growing.
Purpose: This study examined the construct validity (i.e., the degree to which a measurement reflects what it is intended to measure) of EMA in terms of measuring speech understanding and related listening context. Experiment 1 investigated the extent to which individuals can accurately report their speech recognition performance and characterize the listening context in controlled environments. Experiment 2 investigated whether the data aggregated across multiple EMA surveys conducted in uncontrolled, real-world environments would reveal a valid pattern that was consistent with the established relationships between speech understanding, hearing aid use, listening context, and lifestyle.
Research Design: This is an observational study.
Study Sample: Twelve and twenty-seven adults with hearing impairment participated in Experiments 1 and 2, respectively.
Data Collection and Analysis: In the laboratory testing of Experiment 1, participants estimated their speech recognition performance in settings wherein the signal-to-noise ratio was fixed or constantly varied across sentences. In the field testing the participants reported the listening context (e.g., noisiness level) of several semicontrolled real-world conversations. Their reports were compared to (1) the context described by normal-hearing observers and (2) the background noise level measured using a sound level meter. In Experiment 2, participants repeatedly reported the degree of speech understanding, hearing aid use, and listening context using paper-and-pencil journals in their natural environments for 1 week. They also carried noise dosimeters to measure the sound level. The associations between (1) speech understanding, hearing aid use, and listening context, (2) dosimeter sound level and self-reported noisiness level, and (3) dosimeter data and lifestyle quantified using the journals were examined.
Results: For Experiment 1, the reported and measured speech recognition scores were highly correlated across all test conditions (r = 0.94 to 0.97). The field testing results revealed that most listening context properties reported by the participants were highly consistent with those described by the observers (74–95% consistency), except for noisiness rating (58%). Nevertheless, higher noisiness rating was associated with higher background noise level. For Experiment 2, the EMA results revealed several associations: better speech understanding was associated with the use of hearing aids, front-located speech, and lower dosimeter sound level; higher noisiness rating was associated with higher dosimeter sound level; listeners with more diverse lifestyles tended to have higher dosimeter sound levels.
Conclusions: Adults with hearing impairment were able to report their listening experiences, such as speech understanding, and characterize listening context in controlled environments with reasonable accuracy. The pattern of the data aggregated across multiple EMA surveys conducted in a wide range of uncontrolled real-world environment was consistent with the established knowledge in audiology. The two experiments suggested that, regarding speech understanding and related listening contexts, EMA reflects what it is intended to measure, supporting its construct validity in audiology research.
The use of construct validity to establish predictive relationships between hearing aid performance and speech understanding