Friday, April 16, 2010

Research part 3

Correlation can be +ve or -ve. 1 is high correlation. Reduce  by .25s you get mod and low correlation.
go throu' the common variance ones-need to know what it is.that's all
Sensitivity and
Specificity:
No need of complex tables and squares!
Sensitivity is true positives.
Specificity is true negatives.
Higher (SS) the better the value of the test.
Example: you have ten boxes of Grapes.3 rotten and 7 good.If you can identify 7 as good ones correctly ,u have true sensitivity.Identifying 3 gone bad correctly is true negativity.
If you falsely identify (wrong id) good as bad=false negative vice versa.
Concentrate on the table "example of statistical analysis according to study purpose and data measurement" in PTRSG (review guide).
Browse thru' Research methods and other stuff.
Likelyhood ratio: just remember the top,middle and lowest values(10,1,-ves)
i'll upload later.

5 comments:

  1. hi wan , I think sensitivity is true positive means sick are identified as sick and specificity - healthy are identified as healthy in following example you said 7 good as true sensitive ... can u please explain ??? i mean in your ex u have high specificity.....not sensitivity.

    ReplyDelete
  2. This comment has been removed by the author.

    ReplyDelete
  3. thanks wan . I got it now ....

    ReplyDelete
  4. Hi Gauri,
    Glad you asked.
    As you understood sensitivity is true possitives.But that does not have to do with "sick" or 'being bad'
    If a given test accurately identifies (measure or test) what it is supposed to,then is has good sensitivity.If the same test accurately tell you that the subjects don't have it,then it is specificity.IT depends on your hypothesis.Here it is to see whether your nose can identify the good ones.
    Another example where you identify the negative aspect is, if allens test can identify 7 out 10 with TOS correctly and 3 as not having TOS then it has both high sensitivity and specificity.
    Anyway,this is the anology i had.Tell me if this works out.if that's confusing i'll change the example.
    Thanks again.
    PS:Pls post some easy examples if you have.i want to make it easy,not to confuse.

    ReplyDelete
  5. could u also explain what is alpha level and common varience..hope am not asking u too much..

    Thanks a lot for the help...

    ReplyDelete