P-values or significance level is commonly used (e.g. clinical trials) to determine whether hypothesis tested (comparison of groups’ characteristics) was significant or not. Often outcomes are labeled with “statistically significant” “non-statistically significant” “unlikely due to chance”, or “due to chance”, and so forth. Researchers make conclusions whether to reject the null hypothesis and accept the alternative hypothesis, vice versa, based on significant level.
Alpha levels are set depending on nature of experiments. Researchers set rate how much error is tolerable. For instance, 0.01 alpha level is more liberal than a conservative 0.001 alpha level. Many clinical trials choose alpha level of 0.05. Those drugs on clinical trial with higher alpha level of 0.05 rarely receive FDA approval. In a Phase II clinical trial, where effectiveness and safety is still being evaluated and there are fewer treatment options (for example for rare type diseases), scientist might set a liberal alpha level (Palesch, 2014). Many higher alpha levels are driven by studies’ cost and lower alpha levels by risks such as possible lifelong disability or death. For instance, in Phase III clinical trial such as testing efficacy of hyperbaric oxygen therapy (HBO) on post stroke patients require a conservative alpha level such as 0.02 (Rusyniak, et al. 2018,) or even more stringent level.
The Greek letter alpha is significant in hypothesis testing. There is not a universal value of alpha that should be used on all statistical tests. Choosing the alpha level is a judgement call of the researcher and it depends on the situation. Raising the alpha level poses more risk to type I and type II errors. Raising the alpha level is
usually related to having a small or limited sample size.
An experiment when the alpha level 0.01l might be used is in a study of a medical screen for a disease. An example, testing for cancer. All possibilities must be considered when the choosing the alpha level, The possibilities that he test may give a false positive report of cancer along with a report of egative when the disease was actually positive but the test did not catch the isease. A false positive will increase the patient’s anxiety leading to other ests and in the end the test was negative. This creates a type 1 error. The patient hat tested negative will not get the treatment that they needed for the isease which can lead to death and this is a type II error. Increasing the alpha level to 0.1 could result n a lower likelihood of a false negative