Statistical Test

Statistical test caters that mechanism which is used for devising or making a quantitative decision regarding a procedure. The main intention is determining that whether enough proof is present for rejecting a hypothesis or conjecture regarding that process. We call the conjecture as null-hypothesis. If the hypothesis is not rejected then it means that the result was good and we can continue by believing that the null-hypothesis has been considered true.

Criticisms of Statistical Test

Some of the criticisms which are selected are as follows:

• There present are many number of continuous misconceptions about the test as well as the results of the test.
• The result of the test is sample-size’s function.
• The results of the test lack information.
• Practical-significance is not implied by the statistical-significance.
• Problems are resulted when statistical-significance are used as measure for the publication.
  • It is not easy to rectify the published-Type I-error.
  • Published-effect-sizes tend to lean towards upward side.
  • Errors of Type II are very common.
• Test has been considered as blemished probability-theory’s application.
  • The data given by the null-hypothesis have chances of being unconvincing and same can be the chances of the data provided by alternative null hypothesis.

Each of the above mentioned criticism posses merit or advantage but are subjected to a discussion.

In the literature of statistics, a key role is played by the statistical null hypothesis-testing. The common reasoning line is mentioned below:

• We initiate with that research-hypothesis, regarding which we don’t know any truth.
• The 1^st step done is, stating the relevant alternative and null hypothesis. This step is of great importance as if the hypothesis is not stated properly then the remaining process will become muddy. Most importantly, attaching of an attribute is permitted by the null-hypothesis. Choosing of null hypothesis should be done in such a way that the hypothesis permits us in concluding that whether to accept the alternative-hypothesis or leave it undecided.
• In the 2^nd step we take into consideration the assumptions of statistics which were made regarding sample while practicing the hypothesis test.
• Make a decision in choosing appropriate test & stating T, that is, relevant statistical test.
• From all assumptions, derive statistical test’s distribution under null-hypothesis. This test, in standard situations, is considered as well-known outcome. For e.g. statistical test might follow normal-distribution or T distribution of students.
• T – Statistical test’s observed-value (t_ob) is computed out of the observations.
• Statistical test’s distribution makes partitions of possible T’s values into all those tests whose null-hypothesis was rejected which are called as critical-region.
• Make a decision regarding either to reject hypothesis in alternative’s favor accepting the null-hypothesis. The rule of the decision is rejecting H₀ null-hypothesis if observed-value lies in critical-region or accepting the null hypothesis.