擅长:python、mysql、java
<p>从同一分布中随机抽取两个样本,计算t统计量,以检验均值相同的无效假设</p>
<p>因为样本是随机的,所以没有理由将p值分布得更接近1。要理解这一点,请考虑置信区间</p>
<p>置信区间告诉您(1-<em>alpha</em>)*100%的时间,真实参数将位于观察到的区间内。同样,您的p值在0和0.05之间,大约占时间的5%</p>
<p>换言之:</p>
<pre class="lang-py prettyprint-override"><code># Convert `ps` to numpy array
ps = np.array(ps)
# Check how many times you rejected H0
print('We rejected H0', (ps <= 0.05).sum(), 'times out of', len(ps))
print('We did not reject H0', (ps > 0.05).sum(), 'times out of', len(ps))
</code></pre>
<p>返回:</p>
<blockquote>
<p>We rejected H0 246 times out of 5000</p>
<p>We did not reject H0 4754 times out of 5000</p>
</blockquote>