Pharmaceutical Statistical Validation

STATISTICAL VALIDATION

INTRODUCTION

After accomplishing the thorough investigation of various aspects of possible ‘determinate errors’ and having applied the relevant corrections, it has been observed that the data thus generated not only show fluctuations but also are found to be random in nature. The powerful and effective technique of statistics may render such results, which scatter in a random manner, into a better form that may be employed intelligently. Besides, the specific statistical treatment of the calibration data, aided by pre-programmable calculators and micro-computers, very often yields a fairly accurate and more presentable determination of the graphs between absorbance and concentration than those produced manually.

STATISTICAL VALIDATION

The statistical validation of analytical results will be discussed with regard to the following six as-pects individually, along with appropriate examples wherever possible, in the sections that follow :

(i) Statistical treatment of finite samples,

(ii) Distribution of random errors,

(iii) Significant errors,

(iv) Comparison of results,

(v) Method of least squares, and

(vi) Criteria for rejection of an observation.