Randomized Complete Block Design and LSD Test

Lesson 44 of STA301 explores the Randomized Complete Block Design (RCBD) and post-hoc multiple comparison procedures via Fisher's Least Significant Difference (LSD) Test. In agricultural, industrial, and clinical trials, experimental units are frequently heterogeneous (such as varying soil fertility, machine operator experience, or patient baseline health). In RCBD, units are grouped into homogeneous subsets called Blocks, with each treatment allocated randomly once within every block. By isolating block-to-block variability, experimental error is substantially reduced, improving test precision. The linear mathematical model for RCBD is Y_ij = mu + alpha_i + beta_j + e_ij, where alpha_i is the treatment effect, beta_j is the block effect, and e_ij is random error. Total variation is partitioned into three additive components: Total SS = Treatment SS + Block SS + Error SS, with degrees of freedom partitioned as (k*b - 1) = (k - 1) + (b - 1) + (k - 1)*(b - 1). A Two-Way ANOVA Table is constructed, computing Mean Squares for treatments (MST = SST / (k - 1)), blocks (MSB = SSB / (b - 1)), and error (MSE = SSE / [(k - 1)*(b - 1)]). Two separate F-ratios are evaluated: F_treat = MST / MSE testing treatment equality, and F_block = MSB / MSE testing blocking efficiency. When ANOVA rejects the omnibus null hypothesis indicating significant treatment differences, researchers must determine which specific treatment pairs differ. The lecture presents Fisher's Least Significant Difference (LSD) Test: LSD = t_(alpha/2, df_error) * sqrt[2 * MSE / r], where r is the number of replications (blocks). Any absolute difference between two sample treatment means exceeding the LSD threshold (|X-bar_i - X-bar_j| > LSD) is declared statistically significant at level alpha. The lecture also introduces the Chi-Square Goodness of Fit test comparing observed and expected distributions. In Virtual University final term examinations, two-way ANOVA table calculations, computing error sum of squares, and executing LSD comparisons represent major numerical questions. Practicing sta301 handouts and solved past papers by Moaaz guarantees success.