来源：环球教育
小编：长安 55Reliability的意思是可靠性，很多情况下约等于研究方法可重复度。在心理学的学习中，reliability在评估研究方法时是一个关键因素，环球教育朱平博老师为大家做出如下分析。
判断一个方法是否reliable，一般来说是看它是否容易被重复使用。因而，像实验那样highly structured的研究方法，一般来说就具有比较高的reliability。除了在实验室进行的实验，一些完善制作的closed－question questionnaire（有固定选项让你选）同样也有很高的reliability，因为基本不需要研究人员在答题者答题的时候进行引导，所以同样的问卷即使是又不同的研究者来发放，也会得出相似的效果。
就像我们在不同网站看到的那些心理测试，问题本身是一样的话，测评得出来的结果也是一样的（当然了，野鸡网站上不靠谱的测试除外……）。
刚刚我们讲highly－structured的研究方法会有比较高的reliability，那相对来说，case study等方法的reliability就会比较低。因为它研究个体或少数人，且通常采取open－question的方式，那么不同研究者对同一群人或一个人的研究就可能会有所不同，有时甚至有巨大差异。弗洛伊德曾经对5岁男孩Hans进行个体研究，得出5个性阶段的结论，如果拉康有机会对这个小男孩进行研究，可能会采取不同的方法，那么又会有不同的结论出来。
因此我们有下面这个表：Levels of reliability for the different methods of investigation
| Method | Level of reliability | Explanation |
| Experiment | high | Experiments involve levels of controls so are easy to replicate by following procedures that are clearly set out. 尤其在实验室里进行的experiment具有特别高的reliability |
| Correlation | high | Correlations rely on statistical analysis of scores so variables under investigation need to be quantified. This involves using clear measures of behavior that are easy to scrutinize and use again. 数据分析是关键 |
| Self-report | variable | Questionnaires and structured interviews are easy to replicate because questions are preset and available to be asked again (although the context and interview style may affect reliability). Unstructured interviews are more difficult to repeat because the questions depend on the answers that an individual gives at the time. |
| observation | variable | structured observations are more reliable because they use coding systems, which other researchers can use to conduct an observation of the same or similar participants in the same or similar situation. However, qualitative observations involve simply recording what is seen at any one point in time – which also depends partly on the observer’s own interpretation. 从这两种方法来说，我们可以看到structured的方法有比较高的重复度，就像一个规定每种食材分量以及火候的菜谱，拿给任何一个厨师做都会做出十分相似的味道（一个不是特别到位的比喻） |
| Content analysis | variable | Quantitative content analyses are more reliable because they use numerical coding systems, which other researchers can use to analyse the same or similar material. However, qualitative content analyses involve interpreting material through description and this is much more open to interpretation and therefore difficult to replicate. 比如两个研究者可能在对同一个自闭症患者的图画的描述中有所不同，因而造成比较低的reliability。 |
| Case study | low | Case studies involve high levels of involvement and interpretation on the part of the researcher. This makes it difficult for another researcher to carry out the research in the same way to achieve the same result. |
