I have observed in various exc. that Zx is computed with an application of a shortcut for sigma = standard deviation which is a square root of Ex * qx instead of the full formula of square root of Ex * qx * (1-qx). I am just wandering if this approximation will always be acceptable on the exam? Just wandering if that could be a source of loss of marks in the situation when some people compute full formula and some an approximation which is quicker.
Hi Viki2010, Excellent question! Really, it depends which test you're doing. You're spot on, if we are doing the ISD test then the s.d. is sqrt of Ex * qx *(1 - qx) I have never seen this be a problem in the exam but, if you were worried that the qx's were not small enough to say 1 - qx = 1 then, you should do it properly as you suggest - you would get full marks for this and the worst case scenario would be the slight loss of time. I am 99.99% certain that you would be OK just saying that 1- qx = 1 since qx's are small and using sqrt of Ex * qx However, careful! If we're doing the Chi-squared test then this is (A - E) ^2 / E and so you have to use sqrt of Ex * qx on the denominator before you square the Zx's Good luck! John