Understanding where biases exist is crucial for promoting gender equality
and creating an environment in which everyone is given equal
opportunity to demonstrate their talents.
Therefore, we first attempted to visualize the biases that exist within our organization.
To measure gender bias, we adopted two methods: the SESRA-S
(Scale of Egalitarian Sex Role Attitudes - Short Version)
and the Gender-IAT (Gender Implicit Association Test).

SESRA-S (Scale of Egalitarian Sex Role Attitudes - Short Version)

The SESRA-S evaluates respondents’ conscious attitudes towards gender equality through 15 questions. Specifically, it measures their attitudes towards gender-based stereotypes (such as "women should do the housework" or "men should be the ones to work"). A high SESRA-S score indicates less stereotypical view on gender and greater support for gender equality, while a low score suggests greater support for traditional gender role divisions.

The Gender-IAT (Gender Implicit Association Test)

The D-score derived from the Gender-IAT indicates the strength of the association between targets (e.g., men, women) and attributes (e.g., career, family). This D-score corresponds to the level of what is often referred to as "unconscious bias."
The test is conducted on a computer and requires participants to quickly categorize words related to gender (such as "men" or "women") in relation to words like "work" or "family" as they appear on the screen. Strong unconscious beliefs lead to faster recognition of related items. For example, someone who strongly associates men with work will respond quickly to questions linking men and work but take longer to respond to word combinations they consider to be less related. The greater the difference in their response time, the stronger their unconscious bias is evaluated to be.

As shown above, these two scores measure different mental phenomena, and we believe that both measurements can be very effective in "visualizing" potential biases within organizations.
Specifically, one's biases when making deliberative judgments are likely to be reflected in the SESRA-S, while tendencies towards bias in one’s instant judgments, such as spontaneous comments, are more likely to be reflected in the Gender-IAT results.

Research Question 1:
Is there a difference in bias between men and women?

We first focused on the differences in bias between men and women.

We found that women scored statistically significantly higher on both the SESRA-S test (conscious gender equality attitudes) and D-score (unconscious bias).

The higher SESRA-S scores (conscious gender equality attitudes) for women can be explained by gender equality being more of a personal issue for them, creating a stronger psychological effect that allows them to be more conscious of equality. On the other hand, gender equality may be less of a personal issue for men, making it more difficult for them to be conscious of gender equality.
The higher D-score (unconscious bias) for women could be interpreted as having stronger unconscious beliefs about "how women should be" or "how men should be" than men do. This unconscious conditioning may not only manifest itself, but also potentially affect the women themselves and constraining their own behavior.

Although it may seem strange, these two indicators can sometimes contradict each other, as seen here. Women demonstrating higher levels of both gender equality consciousness and unconscious bias may indicate that they are more strongly aware of gender equality compared to men, while at the same time being more likely to be bound by internalized notions of femininity and masculinity, which, could be the result coming from internal conflict in themselves.
Our empirical results align with observations from previous studies conducted by other research institutions, suggesting that these conclusions are highly reliable.

Research Question 2:
How does gender equality within an organization relate to bias?

How does the level of women's advancement within an organization relate to bias?
We examined the relationship between each score and the percentage of women in management positions, considering those organizations with higher percentages as progressive organizations advanced in women’s empowerment. (Figure 2)

The SESRA-S scores (conscious gender equality attitudes) were lower in organizations with lower percentages of women in management positions, and scores increased as the percentage of women in management positions rose.
The D-scores (unconscious bias) were high in organizations with both lower (A) and higher (D) percentages of women in management positions.

The trend in SESRA-S scores (conscious gender equality attitudes) suggests a certain relationship between the permeation of ideas regarding gender equality and the presence of women active in the organization. This could be interpreted as either the promotion of women's advancement influencing the formation of gender equality-related values, or as the permeation of gender equality-related values influencing women's advancement.
On the other hand, the trend in D-scores (unconscious bias) showed that organizations with an imbalance in the gender ratio of their active leadership tend to demonstrate higher levels of unconscious bias. This suggests that in order to achieve gender equality within an organization, it's important to resolve organizational homogeneity (bringing the male-female ratio closer to parity) on top of focusing on women’s advancement.

Now, the organizations that lag behind in women's advancement (= highly homogeneous organizations), as introduced in Part 1, correspond to “Organization A” in this study (organizations with low conscious gender equality attitudes and high unconscious bias). Compared to other organizations, Organization A has relatively strong biases, suggesting that resolving organizational homogeneity would be key to the formation of ideas regarding gender equality.

Research Question 3:
Is the gender gap in evaluations caused by supervisors' biases?

The previous study showed that in organizations lagging behind in women's advancement (= Organization A), there was a tendency for male employees to receive higher skill evaluations at the Intermediate grade level*. Additionally, this study found that Organization A had lower gender equality values and higher unconscious bias scores compared to other organizations. (*Shiseido's grade levels include the Junior, Intermediate, and Senior levels.)
So, what about the biases of supervisors in organizations that lag behind in women's advancement? We examined each indicator for Organization A, separating "managers," those who have subordinates and conduct evaluations, from "non-managers," those who are evaluated.

There was no statistically significant difference between the scores of managers with subordinates and the scores of non-managers on either the SESRA-S (conscious gender equality) or D-score (unconscious bias) tests.

Given the gender gap in skill evaluations in Organization A, there were concerns that supervisors in this organization might lack gender equality values or have strong unconscious biases. However, this analysis did not support such hypotheses. Instead, non-managers tended to show relatively lower gender equality values, although not to a statistically significant degree. This suggests that the gender bias reflected in skill evaluations might be caused by the values and preconceptions of "how women/men should be" held by everyone in the organization, as well as the organizational culture and values fostered by these biases, rather than being solely caused by the biases of supervisors. Therefore, it seems necessary to address this as an organization-wide issue rather than assuming that the problem lies with supervisors specifically.

What became clear through this research process was the relationship that exists between organizational homogeneity and the presence of bias. Organizations with both extremely low and extremely high percentages of women in management positions tended to demonstrate stronger unconscious biases. This suggests that gender imbalances may strengthen organizational bias. Increasing the diversity of organizational leaders appears to be an important step towards reducing bias.
Moreover, the fact that our initial hypothesis that “supervisors possess stronger gender biases" was not supported indicates that bias is not merely a problem that exists within specific layers or among individuals, but represents an organization-wide issue. Bias must be confronted by those at all levels of the organization.

Based on these insights, the following approaches might be effective:
1. Ensuring Gender Balance among Organizational Leaders:Ensure gender balance among organizational leaders to resolve homogeneity and incorporate diverse perspectives. 2. Organization-Wide Approaches:
Rather than targeting specific layers of the organization, provide opportunities for all members to confront gender bias, fostering a culture that questions biased assumptions and what has previously passed as "common sense” in order to recognize and accept different ideas. 3. Continuous Verification:Regularly measure levels of organizational bias and determine response measures based on the results.

These approaches have deep connections to the concept of DE&I (Diversity, Equity, and Inclusion). By respecting diversity, providing fair opportunities (equity), and accepting different perspectives (inclusion), organizations can make progress towards achieving true gender equality.

This cannot be just a one-time effort; what’s important is embedding these practices into the organizational culture. Confronting bias requires continuous effort; constant verification and improvement are necessary. While not easy, this is an essential step towards creating a fair and productive work environment. True gender equality helps to build an organization where all employees believe they can maximize their own potential. The Shiseido DE&I Lab will continue to conduct research that leads to organizational transformation.

Statistical Analysis: Shintaro Yamaguchi (University of Tokyo), Yoko Okuyama (Uppsala University), Rentaro Utamaru (University of Tokyo), Keisuke Tsugumi (Hitotsubashi University), Yuma Oshima (University of Tokyo).