Planned Tests Of Statistical Significance
|Related Online Training modules|
|Planned Tests Of Statistical Significance|
|Generally it is best to access online training from within Q by selecting Help > Online Training|
When you select one or more cells in the table and press , an expert system in Q determines the test that you most likely want and conducts this test. This is referred to as a planned test because, unlike the automatic tests shown in tables, the relationship is being tested as though the test was planned before the research was conducted, and so no correction is made for multiple comparisons.
The output below, for example, shows that there is a significant difference between the typical level of coffee consumption and cola consumption.
As shown in the null hypothesis the test performed has actually checked that the median is different (i.e., it has not compared the averages). Further, in this example the test was deemed significant because the p-value was less than or equal to 0.1 (by default it is 0.05, but can be changed by changing the Overall significance level).
Sometimes a result will be shown as being significant when you press , but will not be shown as significant on the table. This is generally because multiple comparison corrections have been applied when highlighting cells as significant, whereas the results obtained by pressing contain no corrections (see Multiple Comparisons (Post Hoc Testing)).
Q works out which test to conduct by taking into account the structure of the data (e.g., categorical, numeric, paired, etc - see Tests Of Statistical Significance and the settings in Statistical Assumptions.
- It is always important to read the precise wording of the Null hypothesis as it will indicate what the result actually means.
- When you press no Multiple Comparisons Correction is applied. Thus, a cell that is not shown as being significant on a table may be shown as being significant when is pressed.
- How to Specify Columns to be Compared
- The SurveyAnalysis.org website contains a section on Formal Hypothesis Testing, which is another name for what is referred to in Q as a planned test.