A tool for calculation of the test power for categorical variables based on application of the multinomial distribution and Monte Carlo simulations – user instructions

The Excel spreadsheet file needs to be downloaded and the security unblock box checked in line (sub-menu) “Property”, which appears in the menu after the right click on the file name in the downloads folder. When the security warning red frame in “Property” disappeared, the tool is ready to use. The spreadsheet is accompanied by the dataset related to the study of the intensity of the chlorine odor of the drinking water (*.txt file), discussed in the present paper, as an example. This dataset needs also to be downloaded.

The following parameters can be set in “Run Program” sheet:

  • “Model Type” (ordinal or nominal);
  • “Number of Iterations” (simulations);
  • probability of Type I error 𝛼;
  • effect of the sample size 𝑤.

The accompanying file-example or another file of the same structure with the dataset in *.txt format should be uploaded using the button “Get File”. Then, the chosen model type is initiated using the button “Run Model”. The calculation time with a regular PC for the proposed example of the water chlorine odor intensity is about 3-5 min.
The output includes the total variation and its components, their degrees of freedom, the significance indices, their P-value (probabilities), the critical values of these indices for given level of confidence (1 − 𝛼)100 %, and the test power as described in the present paper. Sheets “Program Chart PDF” and “Program Chart CDF” present plots of the probability density functions and cumulative distribution functions of the significance indices, respectively.
The tool allows also to calculate characteristics of an interaction of two factors (variables) when the experiment was designed correspondingly.

ייעוץ לימודים