Can MAD replace significance tests? Comments on 'When "significant" is not significant'
Chuck Chakrapani
Leger Analytics
In their paper, ‘When “significant” is not significant’, Kennedy, Scriven and Nenycz-Theil (IJMR, 56, 5, 2014) present a table that is described as ‘simple averages in the form of mean absolute deviations [MADs] from the average brand user profile for brands in the category’, which can be used as a substitute for statistical significance tests. Since there was no more explanation of the numbers presented, I presume that this table reports the averages of...