Thursday, April 15, 2010

Research part 2

Standard Deviation:
Say for example you have hundred numbers 1,2,3,4,5,6,7,8,9,10...100 (with symmetrical distribution)out of this 69% of the scores fall with in SD+or-1. 69% 100(numbers)is 68 you are left with 32 numbers.
out of this 95% fall with in SD of +or-2 (30.2 numbers out of the 32 are with in that range).
Now 99% of the numbers left fall with SD of +or-3.(what ever remaining)
Hope  i am not confusing you with this example.
Probability Levels:
For PT it is always less than 0.05 P<0.05.
For pharma it is very less(0.01).This means out of handred chances you can go wrong (chance) only 5 times.
Alpha levels:
Higher the controlled Alpha level greater the value of the study.If you find the word 'inflated' before alpha then the value is Low.
Ratio and Intervel Data:
To compare use t-test or ANOVA.
ancova is used when there is an intervening variable
Two unrelated groups and one independent variable-use t-test
Two matched groups with one independent variable-use paired t-test
More than two groups,one independent variable- use ANOVA
More than two groups and more than one variable-use factorial ANOVA
To correlate use Pearson product-moment(r)
For Nominal or Ordinal Data:
To compare use chi square
For correlation use Spearmans rho
(Please read on biserial and rank serial correlation.just know what it is.)

No comments:

Post a Comment