Calculation Method

1. Calculation of dissimilarity(categories)
   Dissimilarity in informational demographics
   The HR managers provided the demographic information. Education level was measured using five categories. The education major was categorized as fine arts, law, liberal arts, science, engineering, agriculture, medicine, or business. The functional specialty was categorized as operations, R&D, finance or accounting, general management, marketing, personnel, legal administration, or general counsel.
   We then calculated dissimilarity scores for each of the three demographics—level of education, education major, and functional specialty. Following previous studies (e.g., Tsui et al., 1992), dissimilarity was assessed by aggregating the dissimilarity scores of all of the CEO–TMT member dyads. Specifically, a variant of Blau’s (1977) index of heterogeneity was used. This is defined as (1 − Pi)2, where Pi is the proportion of CEO–TMT member dyads that share the ith category (Murray, 1989). Taking education level as an example, the index gives the square of the proportion of CEO–TMT member dyads in which the individuals had different levels of de- gree. The indices for the three demographics were averaged to create an indicator of CEO–TMT dissimilarity in informational demographics.
2. TMT tenure overlap
   Information about TMT tenure was provided by the HR managers. For each firm, we first calculated the tenure overlaps of all the CEO–TMT member dyads by identifying the number of years the CEO and the particular TMT member had worked together in the team. The sum of these scores was then divided by the number of non-CEO TMT members, to represent TMT tenure overlap between the CEO and the other TMT members.

1/4 2/4 3/4 4/4 AND？THEN？
——Yan，2015 Benefiting from CEO's empowerment of TMTs: Does CEO–TMT dissimilarity matter?

HOWEVER 对公式表示质疑

3. Age
   To minimize any variance caused by demographic factors extraneous to the research question, we included differences in gender and age between the CEO and the other TMT members in the analysis as controls. For the categorical variable of gender, the same formula as above, (1 − P)2, was used, with P representing the proportion of CEO–TMT member dyads sharing the same gender. For the
   continuous variable of age, dissimilarity was measured using an analog of the Euclidean distance measure (i.e., the coefficient of variation):
   1/n{[sigma(Sceo-Sj)]1/2}
   where SCEO is the CEO's age, Sj is the age of TMT member j, and n represents the total number of non-CEO TMT members. We also used the age of the CEO and the average age of the other TMT members as a proxy to control for the potential influence of the career stage of the CEO and other TMT members on leadership effectiveness (Super, 1980).