Spread the Word: Enhancing Replicability of Speech Research Through Stimulus Sharing

Area of Science:

• Speech and Hearing Science
• Psychology
• Research Methodology

Background:

• The replication crisis highlights how research flexibility affects outcomes.
• In speech research, "researcher degrees of freedom" include choices in stimuli, talkers, masking, and listener populations.
• Even minor methodological variations can influence study results.

Purpose of the Study:

• To reanalyze existing speech data to assess the impact of methodological variations on spoken word identification accuracy.
• To investigate how differences in talkers, stimulus processing, masking type, and listeners affect identification rates.
• To underscore the importance of addressing "researcher degrees of freedom" in speech research.

Main Methods:

• Reanalysis of six existing data sets on spoken word identification in noise.
• Examination of variations in talkers, stimulus processing, masking type, and listener characteristics.
• Statistical assessment of correlations in word identification rates across different study parameters.

Main Results:

• Relatively low correlations were found among word identification rates across the analyzed studies.
• Methodological details, often unreported, introduce variability that may compromise replicability.
• Specific factors like talker variability and stimulus processing showed potential to influence outcomes.

Conclusions:

• Seemingly minor methodological choices in speech research can significantly affect study outcomes and replicability.
• Publicly sharing research stimuli is crucial for enhancing the reliability and reproducibility of findings in speech research.
• Addressing "researcher degrees of freedom" through transparent data sharing is vital for scientific progress.