卡方拟合优度检验不会产生遵循卡方分布的图

Question

我正在尝试从《统计入门》 3E的6.3.1节中的陪审团问题进行模拟，但是我的模拟似乎并不正确。请参见下面。

我的仿真代码：

white_proba = 0.72
black_proba = 0.07
hispanic_proba = 0.12
other_proba = 0.09
sample = 275
_sum = 275
n = 4

array_of_proba = np.array([white_proba, black_proba, hispanic_proba, other_proba])

number_of_simulations = 1000
master_list_of_chi_stat = []
master_list_of_chi_p_value = []
for b in range(number_of_simulations):
    observed_values = np.random.multinomial(_sum, np.ones(n)/n, size=1)[0]
    expected_values = array_of_proba * sample
    chi_square_stats = np.sum(np.square((observed_values - expected_values)/np.sqrt(expected_values)))
    master_list_of_chi_stat.append(chi_square_stats)

我的图形代码：

# Add histogram data
x2 = master_list_of_chi_stat

# Group data together
hist_data = [x2]

group_labels = ['Chi_Square_Stat']

# Create distplot with custom bin_size
fig = ff.create_distplot(hist_data, group_labels, bin_size=0.1)

# Plot!
py.iplot(fig, filename='Chi_Square_Stat')

很显然，我的代码出了点问题。预先谢谢你。