A NEW METHOD FOR DETERMINATION OF ADEQUATE SAMPLE SIZE
Most population variables such as density seldom appear to be normally distributed. Therefore, the equations for normal distributions may overestimate the sample size required to obtain an accurate estimate of the variable. In this research, the total required sample size for obtaining the density with 5 and 10% precision was determined in random, uniform and clumped distribution patterns. A significant power relationship (p< 0.0001) was found between the actual sample size and the square of coefficient of variation for achieving 5 and 10% precision. The sample size calculated for normal distribution based on 95% confidence level and 5 and 10% precision was respectively 1 to 10 and 1 to 4 times the actual sample size obtained in the different distribution patterns. The overestimation increased with an increase in the values of coefficient of variation. A new method is presented based on population and sample coefficient of variation for estimating the required sample size. The new method can be used to obtain a reliable estimate of sample size using a wide range of data types in different study designs.