TeacherGroupie you are right, but if you put it on a 100 points scale it will still give you the same results or close to it
"Or close to it" is key, and I wasn't seeing that in the explanation. One of the reasons for scaling is to compensate for inevitable differences in difficulty among questions and across test versions.
Just took my test today in Arkansas (5169) Algebra 20, Geometry 14, total raw score 34 According to your information this should be good enough to pass. Is this accurate. I am so nervous. Any comments would be appreciated
No, the computer only gave me my raw scores for each category. I guess the scaled score will be in 2 weeks or so.
HI Everyone, To Alymark: Thanks for posting your scores, and congratulations! To dmccoy: Well done, the regression predicts your scaled score will be 174.3 give or take up to two points. Now the raw scores to scaled scores that are shared here are as follows: Raw total vs Scaled Score from ets 28 157 29 161 29 163 31 165 32 171 33 172 34 173 34 175 35 178 36 180 36 182 37 183 39 184 41 190 41 190 41 195 44 200 To TeacherGroupie: That’s a good point about the regression having an intercept of 88 while the scaled scores have a minimum of 100. I have no explanation, ETS would have to share their criteria or formula they use for us to understand why. I do know that with the new data set the current regression comes out to y = 2.5394x + 87.963 R² = 0.9787 Congrats to those who passed. To those who didn’t please don’t give up, you will try again or you will use your talents in some other way, this is just one test, you are a whole person.
I do not know how to message on here but you can either post it on here or email me at yahoo. My email is goodlife4545@yahoo.com
The reason a y-intercept of ~88 makes little sense is that the regression line was computed in the domain of 28-44; that is, we are assuming that raw scores lie ONLY between those 2 ranges. But the y-intercept indicates that when the Raw score = 0 [which is NOT in our domain of 28-44], the scaled score is ~88. Bottomline: IF we had access to a candidate with a really low Raw score (close to 0!) and discovered his / her Scaled score (!), with that additional datum, we could create a New regression equation whose y-intercept WOULD make eminent sense (it'd be close to 100), haha! This is all assuming of course that the association between the 2 is strictly linear. Jay. http://praxismathguru.weebly.com/
