Rise of the Machine: Detecting Aberrant Response Patterns in Survey Instruments Using Autoencoder.

Area of Science:

• Psychology
• Educational Research
• Data Science

Background:

• Survey questionnaires are critical in psychological and educational research, influencing conclusions and policy.
• Ensuring data quality is challenging due to aberrant response patterns that can lead to flawed research outcomes.

Purpose of the Study:

• To present a machine learning solution using autoencoder neural networks for detecting aberrant response patterns in survey data.
• To enhance the reliability and trustworthiness of findings in psychological and educational research.

Main Methods:

• Employed autoencoder neural networks as a computational method to identify anomalous response patterns.
• Evaluated the effectiveness of the autoencoder approach using both simulated and real survey data.

Main Results:

• The autoencoder neural network method demonstrated effectiveness in detecting response anomalies.
• The approach successfully identified aberrant patterns in both simulated and real-world datasets.

Conclusions:

• Autoencoder neural networks provide a viable computational method for identifying aberrant survey responses.
• This enhances researchers' analytical options, improving the integrity of psychological and educational research.