Two weeks to go for CAT results: So here is some final analysis based on the spreadsheet I created using crackU data.
1. Though it is very hard to tell the percentiles using a sample data like this, My guesstimates using the data, we are looking at 99%iles at raw scores for 159-161 (Slot 1) & 157-159 (Slot 2). It is important to know, even though last years scaled score of 155 fetched 99%ile, the raw score for slot 1 would have been aroundq 157 to get a 99%ile. So we are barely looking at 2-4 marks increase in cutoffs for 99.
2. Normalisation effects:
a. VARC upscaling for slot 1, but very low in magnitude. My guess, even 99%+ilers will have a barely 1-2 marks increase.
Raw Score: 70
Scaled Score (if Slot 1): 71-72
Scaled Score (if Slot 2): 69-70
b. DILR, upscaling for slot 2, and very high in magniture. My guess, even 99%+ilers are looking at 5-6 marks increase at slot 2.
Raw Score: 70
Scaled Score (if Slot 1): 65-67
Scaled Score (if Slot 2): 76-78
c. Quant, upscaling for slot 1, of average magnitude. My guess 99%ilers are looking at 2-4 marks increase in slot 1.
Raw Score: 70
Scaled Score (if Slot 1): 72-74
Scaled Score (if Slot 2): 68-69
d. Total: 210
Scaled Score (if Slot 1): 208-211
Scaled Score (if Slot 2): 213-217
According to me the lower bound of range is highly likely, than the upper bound of range.
3. The effects of Normalisation will be drastically high for scores around 200, and will start falling in magnitude as the scores fall. So at scores around 150, we might only see a difference of 2-3 marks in normalisation (which will be consistent with last years effects in total marks).
4. There is a certain observation in CrackU data is are very crucial to note:
a. Very high scores from Slot 1, but as we go down, the number of people scoring high in slot 2 increases and overtakes slot 1 very fast. It seems scoring 200+ was easier in Slot 1 but scoring 150+ was easier in Slot 2.
b. This observation will effect the entire normalisation process if it happens to be true for the total data. I am assuming this observation to be true and hence I claim the normalisation will be closer to the lower ranges of the above scores.
P.S: Here is the spreadsheet for reference https://docs.google.com/spreadsheets/d/1ZybjtkYM53OKNhf6mQa7iZZt7bpsTkAlUgXDQy7rR_A/edit#gid=2076101469