Page 430 - DMTH404_STATISTICS
P. 430

Statistics



                      Notes                   This value is compared with the value of t from tables, to be denoted as t (n  + n  -
                                                                                                        a/2  1  2
                                              2), at 100a% level of significance with (n  + n  - 2) d.f.
                                                                              1   2
                                         (b)  To  test  H   :     £     against  H   :     >     (one  tailed  test),  the  test  statistic  is
                                                      0   1   2         a   1   2
                                                   (X -  X 2 )   n n
                                                     1
                                              t cal  =      ´     1 2  . This  value is  compared with  t (n  +  n  -  2)  from
                                                       s        n +  n                           a  1   2
                                                                 1   2
                                              tables.
                                         (c)  To test H  :   ³   against H  :   <   (one tailed test), the test statistic, i.e., t  is same
                                                     0  1  2        a  1  2                              cal
                                              as in (b) above. This value is compared with - t (n  + n  - 2).
                                                                                    a  1  2
                                    2.   Large Sample Tests (when both n  and n  is greater than 30)
                                                                    1     2
                                         In this case s  and s  are estimated by their respective sample standard deviations S  and S .
                                                   1    2                                                 1     2
                                                                                      X -  X 2           X - X 2
                                                                                        1
                                                                                                          1
                                         The test statistics for two and one tailed tests are  z  =   and  z cal  =
                                                                                 cal    2    2             2   2
                                                                                       S    S            S 1  S 2
                                                                                        1  +  2             +
                                                                                       n    n             n 1  n 2
                                                                                        1    2
                                         respectively. The remaining procedure is same as in case I above.
                                    Remarks:

                                                                                                   E
                                    1.   100(1 - a)% confidence limits for m  - m  are given by  X -  X ±  z  /2 S . .  .
                                                                                          2
                                                                                      1
                                                                     1  1                           (X 1 X-  2 )
                                         If X - X  ~ t - distribution, z  is replaced by t (n  + n  - 2).
                                            1
                                                2
                                                                 a/2
                                                                                  1
                                                                                      2
                                                                               a/2
                                    2.   If the  two sample are  drawn  from populations  with  same  standard  deviations,  i.e.,
                                                                         1    1
                                         s  = s = s (say), then  . .S E  =  s  +   for problems covered under case I and
                                          1  2                (X 1 X-  2 )  n  n
                                                                          1   2
                                                        1   1
                                         S . .     =  S   +     for problems covered under case II, large sample tests. S is a
                                           E
                                             (X 1 X-  2 )  n  n
                                                        1    2
                                         pooled estimate of s, is given by
                                                                 2              2
                                                     å  (X -  X 1 ) + å (X -  X 2 )  n S +  n S 2
                                                                                        2
                                                                         2i
                                                          1i
                                                S =                               =   1 1   2 2
                                                                n +  n                 n +  n
                                                                 1   2                  1   2
                                           Example 16: An investigation of the relative merits of two kinds of flashlight batteries
                                    showed that a random sample of 100 batteries of brand X lasted on the average 36.5 hours with
                                    a standard deviation of 1.8 hours, while a random sample of 80 batteries of brand Y lasted on the
                                    average 36.8 hours with a standard deviation of 1.5 hours. Use a level of significance of 1% to test
                                    whether the observed difference between average life times is significant.
                                    Solution.
                                    Let X and Y denote the life time of flashlight batteries of type X and type Y respectively and let
                                      and   be their respective population means.
                                     X     Y
                                    It is given that    X =  36.5 ,  S  = 1.8,  n  = 100,  Y =  36.8 ,  S  = 1.5,  n  = 80.
                                                                                            Y
                                                            X
                                                                   X
                                                                                    Y
            422                              LOVELY PROFESSIONAL UNIVERSITY
   425   426   427   428   429   430   431   432   433   434   435