Please read article online: (may be helpful)

TESTING OF HYPOTHESIS Parametric Test (T, Z, F) Chi-Sq

http://www.scribd.com/doc/242931155/TESTING-OF-HYPOTHESIS-Parametric-Test-T-Z-F-Chi-Sq#scribd

Mushtaq gave you the lead. Gr8.

Primarily you need to view the variation (fluke-ness),  lest that should defy your requirements ( the case of "performance improvement" in your case.

If the new situation (seen in context with the old one) is significantly disparate  (in terms of stipulated sigma (or square of sigma - variance) ), you can rest assure the endeavour has been successful.

The choice of test - t /z/ F depends upon the information available to you for making above decisions, and as you may wish to analyse the sample, and extrapolate results to the population. Hope this serves.

dear Mushtaq ahemad sir i couldnot download paper which was given by you in a link

Dear Lakhan,

1. t-test is used to find significant difference between two means --- up to sample of 30 respondents.

2. z-test is also used to find significant difference between two means --- when  sample is more than 30 respondents. When sample exceeds 30, t-test automatically converted into z-test.

3. F-test is used to find significant difference between more than two means. When factor is one and means are three or more, F-test is called one way ANOVA.

When factors are two and means are four or more, F-test is called two way ANOVA.

When factors are more than two and means are eight or more, F-test is called multiple way ANOVA.

For Further details,

1. Please consult Book on Marketing research by Malhotra and Dash.

2. Consult my papers on my RG account.

Thanks