Penentuan Besarnya Sampel Uji Coba berdasarkan Presisi Estimasi Alpha Cronbach

The current article shows the importance of involving an adequate size of try-out samples based on two considerations: the power of the analysis and the accuracy of parameter estimates. Two simulation studies were conducted to achieve two purposes. The first was to show the disadvantages of calculating reliability estimates with inadequate sample sizes due to lower power or high inaccuracy. The second objective was to show the performance of two formulas for calculating the required sample size: Feldt and Bonett. Based on their performance, the recommendation on which formula should be used was made. The results of the first simulation show that inadequate sample size resulted in lower power to test estimated reliability from the data and large fluctuation of reliability estimates across samples. Such large fluctuations made higher possibility of having sample values of reliability estimates far from the population value. The second simulation results in better estimation of sample size required to have large power and accuracy using Bonett’s formula.