In scientific research, data analysis comprises the formulation and testing of hypothesis. The analysis of the information obtained from samples provides presumptions on a specific population. The outcome of the analysis may be either a null or an alternative hypothesis (Spiegel, 2000). The null hypothesis represents the opposite of the experiment’s expectations. It is the default result. The acceptance of the alternative hypothesis depicts the rejection of the null hypothesis and vice versa.
The acceptance of the alternative hypothesis does not depict the experiment’s failure. For instance, in medical research, when an individual wants to investigate the effect of a certain treatment, he or she first has to assume that initially there were no treatment effects at all. In such a context, there is a possibility of the null hypothesis’ truthfulness hence there is a need to avoid its rejection (Spiegel, 2000).
The absence of evidence is the main reason why experimenters should not reject some of their null hypotheses. One should not come up with an immediate conclusion in case of a default outcome. This is because the lack of evidence is not a proof of the failure. In the medical field, when a particular medication fails to exhibit effects on an individual, it does not mean that the medication does not affect all people (Murphy, 1998). There is usually some degree of effects of the treatment, but the current methods of observations fail to detect the minute changes.
-
0
Preparing Orders
-
0
Active Writers
-
0%
Positive Feedback
-
0
Support Agents
If the null hypothesis has a strong scientific and practical importance, avoiding its rejection is essential as it may contradict with some logical considerations. Logical treatment should state that, as the number of observations approaches infinity, the likelihood of getting the appropriate conclusion is equal to 1. However, the null hypothesis testing (NHST) formulation contradicts this conclusion by stating that, for a true null hypothesis the probability of rejection should be 0.5 or 0.1, regardless of the number of observations (Murphy, 1998). Therefore, the experimenter should know that proving the null hypothesis with certainly is impossible.