05-20-2013 11:05 AM
msb - denotes mean score before coaching,
msa - denotes mean score after coaching
sd - denotes sample standard deviation of the difference between individual scores
r - denotes simple correlation between the scores
If μD denotes the population mean of the difference between
population score before and after coaching, then H0: μD =0 against H1: : μD ≠0 can be tested
using statistic td, where td = 'on clip'
that follows student’s t distribution with n-1 degrees of freedom.
Help - writing data step to create a data set with a variable decision_message that must be
assigned either “H0 is accepted” or “H0 is rejected".
The sample size (n) will be sum of all digits of ID ='bg999999'
level of significance will be sum of last 4 digits ID ='bg999999' divided by 400.
Define ID=’bg999999’, and extract digits from variable ID to get both n and alpha.
05-20-2013 01:14 PM
μD is the population paramater for the DIffernce of scores before the coaching, and the difference of scores after testing. So in your example it is 85-70. However that is just a point estimate, and to do a hypothesis test you must find the upper and lower bounds of this estimate, and determine with X % confidence if the old number is within those bounds.
You are doing a comparsion between two means of Dependent Samples (t test), if that helps you in reference to your statistics book. I say it is a dependent sample because from your description it sounds like each person was given a test, and then coached, and then given a test again.
Question for you, why is your significance level related to your sample size / 400? You set the signifiance level (80%, 90%, 95%, 99%) and this can be totally un-related to your sample size.
Anyways, you are doing a t test with N-1 degrees of freedom, with mean=85-70, standard deviation = 225, and your R is irrelavent (in this case).
SO make an upper and lower bounds at your Alpha level of confidence, and if 70 is in that bounded area, then you cannot reject your null hypothesis. Otherwise you can