Supplement 2

I recently analyzed the data of an Internet questionnaire on hand (DIMSDRIVE, 4,094 people, 2004). The result by age (10-year increments) and gender are amusing. See the following chart (the higher the score, the less stressful).

As expected, Type A is most stressful while Type B is least. However, comparing A (average) and B (average);

• The people in their 50s are less stressful than ones in their 20s
• Both blood types of men in their 20s are less stressful than women in their 20s
• On the contrary, neither blood type of men in their 50s is less stressful than women in their 50s

The number of subjects analyzed in this case is 1,110 people – much convincing.

The interaction between age, gender and blood type is enigmatic. Thus, analyzing the difference by blood type precisely, conventional psychological or statistical methods will not work unless age and gender are fixed ...

Compare relative figures, but not absolute ones. For example, do not compare a baby boy with an adult woman when you measure the difference of height by gender. Of course, in this case, usually women are taller than boys, because the influence of age is much greater than gender. You must not conclude that women are taller than men. This type of errors or mistakes sometimes occurs with random sampling method.

Situations are almost the same as personality tests such as the “Big Five”. As shown in the following figure, scores are mysteriously affected by age and gender. Two items of “openness” have different effects depending on age and gender (in this case, the five personality factors have two items each; openness has items 5 and 10 as follows).

A total of 4,588 subjects were analyzed, which is sufficient for this kind of analysis. It is possible to clearly recognize specific trends according to gender and age.

Items 5: I love novel things and have unusual ideas.

Items 10: I am a mediocre person who lacks imagination.

<Notes>

1. Those scores are adjusted values because items 10 is a reverse item

2. The correlation efficient (r) between items 5 and 10 is 0.29 (medium)

Source: Tetsuya Kawamoto, Atsushi Oshio, Shingo Abe, Yuki Tsubota, Taro Hirashima, Hiroyuki Ito, Iori Tani. (2015). Age and Gender Differences of Big Five Personality Traits in a Cross-Sectional Japanese Sample. The Japanese Journal of Developmental Psychology.

The result is very confusing. Thus, on average:

• Men are more open than women
• According to the items 5, the people in their 20s are “more” open than ones in their 80s
• However, according to the items 10, the people in their 20s are “less” open than ones in their 80s

This may explain the reason why psychological personality tests cannot detect the difference by blood type properly.