Suggestions for new organizational-level item pools for the national Stress Check Program from management philosophy and mission statement: A qualitative study using unsupervised learning.

Objective: This study aimed to obtain suggestions for new organizational-level item pools that companies could utilize to accomplish management philosophy and mission statements in the context of survey and work environment improvements for the national Stress Check Program.

Methods: A qualitative study was conducted using unsupervised learning. A large amount of text data related to management philosophy and mission statements were collected, that is, management messages described on the websites of all companies listed on the Tokyo Stock Exchange. For the main analysis, topic modeling was performed on the nouns from the management messages using Latent Dirichlet Allocation (LDA) to build a model consisting of 10 latent topics, each represented by a group of the 10 most frequently reoccurring nouns. Each group of nouns was qualitatively summarized based on the topic model.

Results: In total, 22 524 nouns were extracted from the management messages of 3575 companies. A topic model consisting of 10 latent topics was constructed using the LDA. The suggestion for new item pools included new technologies, business plans/strategies, company shareholders, health/happiness/wealth, profits/sales, development of society, a sustainable society, safety and security, customer/consumer satisfaction, corporate social responsibility, fairness, transparency, and human rights.

Conclusion: The suggestions for potential item pools were derived from management philosophy and mission statement that are not covered in the existing survey. The suggestions could be useful for motivating employers to implement work environment improvement. Future studies need to make definite items and investigate whether they correlate to job stressors and mental health among workers.