In estimating the parameters of a population if one can sacrifice the property of unbiasedness, better estimators in view of minimum mean squared error(MMSE) can be obtained.The present investigation brings out such an estimator σ2 of a population, employing a prior information namely the co-efficient of Kurtosis. Relative efficiency REF(T*,T) where T is the conventional unbiased estimator of σ2 is calculated.
It is found that T* issuperior to T in case of Leptokurtic populations.Larger gains are observed in efficiencies for small sample sizes. Specific populations like Poisson, Gamma, Exponential, Rectangular and Normal are examined and results are illustrated.
