Having two normal distributions with close means (i.e. -0.1, +0.1), points far in the tails over-represent the "closer" bell by a surprising factor. Psychometric tests show that there is 60% chance that a random male is more aggressive than a random female [1], but the top 0.00001% of the population in aggressiveness is all male.

tailseffect.py

#!/usr/bin/env python3
from multiprocessing import Pool
import scipy.stats

N = int(1e10)
STDDEV = 1

norm_f = scipy.stats.norm(-0.1, STDDEV)
norm_m = scipy.stats.norm(0.1, STDDEV)

# 99.993% | 1 in 15787
stddev_4 = 4 * STDDEV
# 99.99994% | 1 in 1744278
stddev_5 = 5 * STDDEV
# 99.9999962% | 1 in 26330254
stddev_55 = 5.5 * STDDEV
# 99.9999998% | 1 in 506797346
stddev_6 = 6 * STDDEV

results_4 = {
    'M': 0,
    'F': 0
}
results_5 = {
    'M': 0,
    'F': 0
}
results_55 = {
    'M': 0,
    'F': 0
}
results_6 = {
    'M': 0,
    'F': 0
}


def generate_scores():
    m_4, m_5, m_55, m_6 = 0, 0, 0, 0
    f_4, f_5, f_55, f_6 = 0, 0, 0, 0

    scores_m = norm_m.rvs(int(1e6))
    for score in scores_m:
        if score > stddev_4:
            m_4 += 1
        if score > stddev_5:
            m_5 += 1
        if score > stddev_55:
            m_55 += 1
        if score > stddev_6:
            m_6 += 1

    scores_f = norm_f.rvs(int(1e6))
    for score in scores_f:
        if score > stddev_4:
            f_4 += 1
        if score > stddev_5:
            f_5 += 1
        if score > stddev_55:
            f_55 += 1
        if score > stddev_6:
            f_6 += 1

    return (m_4, m_5, m_55, m_6, f_4, f_5, f_55, f_6)


if __name__ == '__main__':
    with Pool() as pool:
        for r in [pool.apply_async(generate_scores) for _ in range(N // int(1e6))]:
            m_4, m_5, m_55, m_6, f_4, f_5, f_55, f_6 = r.get()
            results_4['M'] += m_4
            results_5['M'] += m_5
            results_55['M'] += m_55
            results_6['M'] += m_6
            results_4['F'] += f_4
            results_5['F'] += f_5
            results_55['F'] += f_55
            results_6['F'] += f_6

    print("> µ + 4σ = {}".format(results_4))
    print("> µ + 5σ = {}".format(results_5))
    print("> µ + 5.5σ = {}".format(results_55))
    print("> µ + 6σ = {}".format(results_6))

[1] Eagly, A. H., & Steffen, V. J. (1986). Gender Stereotypes, Occupational Roles, and Beliefs About Part-Time Employees. Psychology of Women Quarterly, 10(3), 252–262. https://doi.org/10.1111/j.1471-6402.1986.tb00751.x