So far we have looked at how we can examine the different scales in the questionnaire by correlation, ANOVA and Tukey's test. Bringing together the results, the table below summarized the Tukey results for binaural case. In general, the binaural reproduction is better, so we use that in preference to the monaural results.
So we can conclude that A is consistently found in the upper set suggesting it is the highest quality machine. C is also found in the upper set for every result. B is consistently found in the lower set suggesting it is the lowest quality machine. C, on the other hand is never found in the lower set. D are also found in the lower set for every result.
While we can't reveal which washing machine is which, it might interest to know that the best sounding (A) is the most expensive, and (B) is one of the cheapest.
Armed with this information, we can then listen to the different sounds for the different washing machines, and try and tell why A is better than B. This is where the adjectives may prove to be useful. For example A, for the spin cycle A and B are in distinctly different groups, so what makes A good and B bad? By adding up the number of times an adjective is used to describe a particular washing machine during the spin cycle, we might get some insight. Below is the chart for A and B for the spin cycle.
So, for example 5 of the subjects thought the spin cycle for washing machine B was "alarming" and none felt that this was true of washing machine A. You can hear the two sound files here:
This chart just shows the adjectives originally on the questionnaire. Subjects were also able to supply their own adjectives if they wished, these are the adjectives that they used:
Armed with this information you can then examine why the particular sounds arise for the two washing machines and thereby improve the sound of washing machine B. This is a complex engineering task, and not the subject of this web page, so we will leave it there.