Level of significance refers to the percentage at which the researcher is willing to accept the probability of committing a Type I error or false positive (Hurlburt, 2003). When the researcher identifies the level of significance to 0. 05, it means that the researcher will accept a 5% probability that the results of the data analysis indicate a relationship when in truth it does not, as such a statistically significant result at 0. 05 level of significance will be assumed as true and not due to chance.
Researchers usually try to lessen the probability of committing errors, but when the chance of committing a Type I error is decreased, it also causes the chance of Type II error to occur or false negative. Type II error occurs when the researcher assumes that there is no relationship between the variables of the study when in fact it does (Hurlburt, 2003). So that, when significance level is increased to 0. 01 then it also increases the chance of committing Type II error.
This means that a 1% probability that there is a relationship when there is none, due to the very stringent significance value, it would become very difficult to have statistically significant results, and hence the assumption would be that there is no relationship. In the social sciences, significance levels are usually set at 0. 05 but for healthcare and public health researches, a 0. 01 level of significance is desired (Hurlburt, 2003). Identifying the significance level prior to data collection enables the researcher to identify what kind of data he/she should collect and which statistical test will be appropriate for the study.
A nurse researcher can set the level of significance at some other level depending on the nature of the study, for example if the study is to determine the relationship of tender loving care to patient satisfaction of quality of nursing care, then a 0. 05 level can be used, but if the study is to determine whether an herbal supplement can control blood sugar level better than just exercise alone, then the level of significance should be increased to 0.
01, thus if the result would say no relationship, we would accept the null hypothesis that herbal supplements do not control blood sugar better, then the patient could still benefit from exercising to control blood sugar levels. If instead the significance level is 0. 10 which would probably lead to a significant value, then we would reject the null hypothesis and say that the herbal supplement can control blood sugar better when it really does not. Reference Hurlburt, R. (2003). Comprehending behavioral statistics 3rd ed. California: Brooks/Cole.