统计常见问题

1. Misconception A

Standard scores, such asz-scores andT-scores,are normally distributed.Whatever non-normality exists in the raw scores will be transferred directly toz-scores andT-scores.Thus, if a set of raw scores has a mild, negative skew, the set ofz-scores(orT-scores) based on the raw scores will have a mild, negative skew.If a set of raw scores is “overly peaked”, converting the raw scores toz-scores(orT-scores) will not bring about normality; the standard scores will be just as overly peaked as the original raw scores.

2. Misconception B

In testing a null hypothesis, thep-valuebased on the sample data indicates the probability thatHois true.One important area of statistics is probability, and one important kind of probability is aconditionalprobability. In our everyday lives, we frequently compute conditional probabilities in ourminds. For instance, if we must decide whether to take an umbrella with us whenwe leave home, we typically ask ourselves the question, “Given the weather forecast for today, what’s the likelihood that it will rain?” In making everyday decisions, or in using data to evaluate a null hypothesis, it’s important to realize that the probability of “XgivenY” is often different from the probability of “YgivenX.” It can matter greatly whether it isXorYthat’s assumed to be true. Because thep-value plays such an important role whenever a null hypothesis is evaluated, it’s important for you to understand thatpis a conditional probability and to be aware of what’s given.

3.卡方检验使用的误区

4.统计显著性检验的一个认识误区

1你证明了零假设（即总体平均数之间没有差异）为误。

2你证明了备择假设（即总体平均数之间有差异）。

3你发现了零假设为真的概率是1.5%。

4你可以推断出备择假设为真的概率为98.5%（1－1.5%）。

1和2的错误在于显著性检验只提供统计量检验的概率信息，不能证明某个假设为真或为误。零假设要么为真，要么不为真。它不是随机的，不存在概率。对于研究假设也一样。从本质上讲，显著性检验只是表示在零假设为真的情况下，通过无数次取样某研究证据出现的可能性，不能给任何假设提供成立的概率，所以3和4都是错误的。针对本例，可以认为，鉴于在零假设为真的情况下，出现本研究结果或更极端结果的概率很低（p= .015 a= 0.05）的概率错误地拒绝零假设。

5.混淆显著性和重要性

6.误区：处理组在协变量上没有显著差异时无需使用协方差分析

7.误区：处理组在协变量上有显著差异时才需使用协方差分析

8.误区：裂区设计中错误使用协方差分析

• 发表于:
• 原文链接https://kuaibao.qq.com/s/20180603G00C9F00?refer=cp_1026
• 腾讯「腾讯云开发者社区」是腾讯内容开放平台帐号（企鹅号）传播渠道之一，根据《腾讯内容开放平台服务协议》转载发布内容。
• 如有侵权，请联系 cloudcommunity@tencent.com 删除。

2018-06-22

2018-06-13

2022-08-11

2022-08-11

2022-08-11

2022-08-11