Quote:
I respectfully disagree, I have been watching this for a long time and it is not like that at all. Let's say for arguments sake we are 5.05
A new very dissatisfied will make it 5.01.
A new very satisfied will make it 5.07.
Additionally, if that same Very dissatisfied changes to a very satisfied the score again will only go to 5.07.
|
Someone should have paid more attention in math class regarding the calculation of arithmetic means. The phenomenon you describe is not an error in our system nor does it reflect "weighting". Examples follow:
Lets say you have the following survey response points:
3,3,3,3,3,3,3,3
Each response is 3 out of a total 4 possible points, therefore your rating would be 7.50 for the above eight survey responses.
Now lets take the same set of eight responses and add a ninth response, 4 or "Very Satisfied":
3,3,3,3,3,3,3,3,4
That changes the rating from a 7.50 to a 7.78.
Now lets take the same set of eight responses and add a ninth response in the other direction, a 0, or "Very Dissatisfied":
3,3,3,3,3,3,3,3,0
That changes the rating from a 7.50 to a 6.67.
So as you can see, your initial rating of 7.50 changed by -0.83 points from a "Very Dissatisfied" rating but only changed by +0.28 when you received a "Very Satisfied" rating.
Lets reverse the problem and start out with all "1's", which averages to a rating of 0.25:
1,1,1,1,1,1,1,1
Adding a response of "4" increases the rating from 0.25 to 0.33.
Adding a response of "0" decreases the rating from 0.25 to 0.22.
So as you can see in this second instance, the "Very Satisfied" reponse actually increased the rating by a larger degree (+0.08) than the "Very Dissatisfied" response (-0.03).
Ultimately the impact that any new ratings will have on your overall rating is dictated by what responses you have received previously and what your overall rating is. The rules of arithmetic means are:
1) The more rating responses you have received, the less affect any new responses will have on your average (whether positive or negative).
2) The greater the difference between your average and any new rating, the more that rating will affect your average. For instance, if your average is 10.00, and you receive a "Very Dissatisfied" reponse, that new response will affect your average much more than if you received a "Very Satisfied" response (because your 10.00 rating was already "Very Satisfied" - a new rating isn't going to alter that). By contrast, if your overall rating was 0.00, a new "Very Satisfied" response would affect your overall rating much more than a "Very Dissatisfied" rating.
This is how arithmetic means of number sets work and it has nothing to do with our site. Your average is what your average is and it is based solely on survey points received out of points possible.
Surely you experienced this effect in school. You could be going along getting test averages of 95, 95, 95, 95, then bam, you get a 0 on a test and your 95 average suddenly blows out to 76. Then, you earn a 100 on a test but your average only goes up from a 76 to an 80. The 0 that you earned had a much greater impact on your average then the "100" that you earned.
Linear Mode
