2.20 Design of Experi, 4 D, (19) D3 (22) Dy (20) Dy (20) Ds (29) D2(24)_ Ds (30) D3 (24) D2 (26) Dy (25) Dy (16) Dy (22) Ds (28) D3 (25) Ds (31) Da (28) Dy (27) Dy (16) D2 (27) D3 (20) Analyse, the given data. 2.3 TWO-WAY CLASSIFICATION In two-way classification of analysis of variance, we consider o, classification along column-wise and the other along row-wise. RBD [Randomised Block Design] Let us consider an agricultural experiment using which we vig to test the effect of ‘K fertilising treatments on the yield of crops. Wi assume that we know some information about the soil fertility of t plots. Then, we divide the plots into ‘A’ blocks, according to the si fertility each block containing ‘k’ blocks. Thus, the plots in each blod will be of homogeneous fertility as far as possible within each bloc the 4 treatments are given to the ‘%’ plots in a perfectly randoc manner, such that each treatment occurs only once in any block. Be the same k treatments are repeated from block to block. This desi: is called Randomised Block Design. ‘The scheme is most readily understood by visualising a field pl for an agricultural experiment, say for four treatments (A, B, C D in six blocks of four plots. PLOTSDesign of Experiments nat Merits and Demerits of Random Design m 1 2 ye YD in Tt hos a simple layout. The design controls the variability in the experimental units and gives the treatments equivalence to show their effects. The analysis of the design is simple and straight forward as in the case of two-way classification of analysis of variance. The analysis is possible, even in the case of missing observations. The design is not suitable for large number of treatments, since in this case the block size is large and hence homogeneity of units may not be possible. Unequal number of replications for equal treatment is not possi ctangular. ible. ‘The shape of the experimental material should be re It controls the variability in one direction only. ‘The analysis of this decision is not as simple as a completely randomized design. The analysis of variance table for a randomized block destsn will, general, have the following form. Variance Source of | Sum of | Degrees of | oan square variation | squares | _ freedom ratio Column | SSC €-1) MSC treatments Row SSR (r=) treatments (Block) Remainder | SSE (or Error) (c= 1) or [N-c-r+]] TSS x-1 | Pre)2.22 Design of Experian By comparing the treatment mean square with the remainder jy). square, we can decide by an F-test, whether the treatments haye effect regardless of whether there is a significant variation from py)! to block. i Working Rule 1, Hp : There is no significant difference. 2. Hy : There is a significant difference. Take the hypothesis that variation between Varieties and betwee, blocks do not differ significantly from the variance due to random errors. Arrange calculation of sum of squares. s 5c Te Step 1: Find N. Step 2: Find T. Step 3: Find 37 Step 4: Find TSS Step 5: Find SSC. _ Step 6 : Find SsR Step 7: SSE = TSS — SSC — SSR Prepare the ANOVA Table Step 8 : Find Table F, and Fy Step 9 : Conclusion ce Example 2.3.1 ‘An experiment was designed to study the performance of 4 different detergents for cleaning fuel injectors. The following ‘cleanliness’ readings were obtained with specially designed equipment for 12 tanks of gas distributed over 3 different models of engines : Engine 1 | Engine 2 | Engine 3 Total Detergent A 45 43 51 139 Detergent B] 47 6 2 145 Detergent C} 48 50 55 153 Detergent D} 42 37 49 apesign_of Experiments 2.23 Perform the ANOVA and test at 0.01 level of significance, : a whether there are differences in the detergents or in the engines, [A.U. Model] [A.U. N/D. 2004][A.U CBT A/M 2011] Solution : The above data are i (i) Detergent (ii) Engine. lassified according to criteria In order to simplify calculations, we code the data by subtracting 50 from each figure. [|]. Engine Detergent xy Pomp | Tom! 2a | 9 AY) | -7 1 | -u | 25 | 49 1 B (¥2) 3 4 2 us |' 9 16 4 C (¥3) 2 0 5 3 4 0 25 D(a) | 8 | | 1 | 2 | ot ion | 1 Total aig [224] 7-35) [ime | 23a |. on 1L.Hy: There is no significant difference between column means as well as row means. 2. Hy: There is significant difference between column means or the row means. Step 1: N = 12 [Total number of entries] Step 2:T = -35 Pr (-35 Step 3: —— = = 102.08 P3257 oO 1 Bale Step 4: T9g = EXP+EXZ+EXS— QV = (102) + (234) + (31) — 102.08 = 367 — 102.08 = 264.92 2 xy eX 2 Step 5 ; _ &Xy Xz 3) 7? = +N TON, UN [Ny > Number of elements in cach column]\ 3 2. 2. = Cast , Can? , OY — 0008 = 81+ 144 + 12.25 — 102.08 = 135.17 2 y, 2 eyy , oe _ oxy , GX" 2 Step 6 SSR = ee % ‘ a » Number of elements in each roy 6 cup, cH, o. (2° —s0208 3 Ag : = 4033 + ro +3 ; ey - 102.08 = 110.91 \S SSE = TSS - SSC - SSR = 264.92 - 135.17 — 410.91 = 18.84 Step 7 : Table of analysis of variance Sources of of at. teen, | Variance ——, variation | squares Between [SSC [C1 usc ~ SSC |p, - MSC |Fe 28) columns | = 135.17) =3-1=2 =1 MSE | =10.92 _ 13517 | _ 87.585 2 3.14 = 67.585 21.52 Between [SSR |r-t SSA MSR 3, Rows =togtf=a1—3 [MSR = 1 |FR ~ MSE a : _ 2697 1314 = 1177 Residual [SSE =18.84 Total Step 8 : Conclusion : Cal. Fe > Table Fe. So, we reject Hy Cal. Fx > Table Fy. So, we reject Hysign of Experiment: pe P : 2.26 Example 2| Perform two-way ANOVA for the given below : IA.U. N/D 2003] Treatment Plots of land A B c 38 40 41 45 a2 49 40 38 42 Use coding method, subtracting 40 from the given numbers. Solution : Subtract 40 from all the numbers. By doing so, F ratio is unaffected and reduces the numbers to smaller numbers. = —j 3 = 1. Hy: There is no significant difference between column means as well as row means. 2.1: There is significant difference between column means or the row means. , Step1:N=3+3+343'= 22 Step2:T=3+0+ 12-3= 12 2 2 Seo3:2 2-2 N 12 22 Step 4-95 = EXP+UXP+EXGHEXG- = 29 + 8 + 86 + 21 — 12 = 1322.26 it of Exper thus | Step §: sg¢ = EXP, EX)? EX? Xo" ve + Ni Ny Ny N,N IN, = Number of clements in cach coly 940, M49 Sis tg 3404+ 443-2 = 42 ey ev e% 2 Ny Ny N,N [Np = Number of elements in each; Step 6: SSR = 1444 Go aaae SSE = TSS —SSC - SSR = 132 - 42-26 = 64 12 = 260 Step 7 : ANOVA Table Source of Variation | S° DE Between [SS0=42 [O-1 columns =4-1 =3 Between ISSR= 26 _ SSR _MSR Fa(2.6) rows MsR="Ty [FR MSE | =5.14 26 -3-8 Error SSE=64 [N-o-r#1 | gp, SSE =6 MSE o—rHt TSS= 122, 10.67 Step 8 : Conclusion = i Cal. Fe < Table Fe. So, we accept Ho Cal. Fp < Table Fp. So, we accept Hy : A : ea Hence there is no significant differ nce between column ™ | as well as row means && Design of Experiments 2.27 Ss Teeny 7 Jl Analyse the following RBD and find your conclusion. [A.U. N/D 2013) Solution : 1, Ho: There is no significant difference between blocks and treatments. 2 Hy: There is significant difference between blocks and treatments, Subtract 15 from cach number.2.28 -Design_of Experimen, N 20 Step2: T = 49 (40 “ Step 3; w=‘ = 8 ae bk Step 4: TSS = ExX}4+2Ex342X34+5x9— Nv ; = 38 + 147 + 119 + 68 — 80 = 292 Cx) CF C4? ex r GS ayaa i ™ ON [Nj + number of elements in each colums] 2 2, 2 = gar +e +0-80 = 572 2 2 = 2 2 step 6: sr = SW, EHP yy? ery? ey? 7 Ny Ny Ny Nz Note [Nz number of elements in each row] 82 | 192 (-3 8 @ Sac s aotgt 4 7780 = 50 SSE = TSS ~Ssc~ssp = 292-572~59 = 1848= pesign of Experiments sep 7 ANOVA Table Rows (olock) Table Source of | Sum of | Degrees | Mean sum (value variation | squares |of freedom] of squares at 5% level Coun [SSC = 572] c-1 MSC MSC [Fe (3, 12) treatment =4-1=3] _ ssc |" ~ MSE|= 874 e-1 = 1.24 = 72 3 = 19.1 Between | SSR = 50 _ MSE |Fa(12,4) 26 Ramainder | SSE= 184.8 or Error = 154 Step 8 : Conclusion : Cal Fe < Table Fo, so accept Hp Cal Fp < Table Fr, so accept Hy Example 4) The following table gives monthly sales (in thousand rupees) of a certain firm in the three states by its four salesmen. Salesmen