What Might Happen If The Hypothesis Test Is Performed Before The Researcher Has Decided On The Significance Level?

• The Formal Hypothesis Test.

This is broken down into several steps: First, the idea; second, formulate the testable hypothesis; three, design the experiment; four, establish the decision rule; five, collect the data from the experiment; six, come to a decision; and finally seven, revise the original idea in the light of the data and experimental process.

• The Meaning of Statistical Significance Levels.

A null hypothesis must be established along with an alternative hypothesis. The decision rule should always include a significance level formulation. The correct test statistic that is to be used can then be identified. The p value of the test statistic is compared to the significance level.

This will either verify or reject the null hypothesis. Significance levels are bound up with the null set. Null hypotheses may actually be true even if they are rejected by the data. This gives rise to a Type I error. Failing to reject a false null hypothesis creates a Type II error.

The probability of a Type I error is the level of significance. This, as noted above, must be set before data is collected. The higher the level of significance, the greater the weight of statistical evidence needed to reject the null hypothesis, avoiding the Type I error as above. The significance levels can effectively control the chances of an error occurring.

• Effects of Missing or Mixing Steps.

All steps should be performed in the specific order above. If you collect the data before make the full hypothesis, this will impact unfairly on the null and alternative hypotheses choices. Effectively, the data will damage the theories.

Assigning a significance level after performing the test means that all data objectivity is lost: The results can be made to mean anything you want them to be. This is far from scientific, and will give a skewed, subjective result.