u5.2-randomizedblockdesigns.ppt

Upload: jacob-huston

Post on 04-Jun-2018

216 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    1/21

    ExpDes-1

    Randomized Block Designs:

    RBD and RCBD ( 15.2, 15.5)

    Randomized block designs: Randomized Complete Block Design Randomized Block Design

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    2/21

    ExpDes-2

    Randomization in Blocked Designs

    For all one blocking classification designs : Randomization of treatments to experimental units takes place

    within each block.

    A separate randomization is required for each block. The design is said to have one restriction on randomization .

    A completely randomized design requires only one randomization.

    Note: The randomized block design generalizes the paired t-test tothe AOV setting.

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    3/21

    ExpDes-3

    Analysis of a RBD

    Traditional analysis approach is via the linear (regression on indicatorvariables) model and AOV.

    A RBD can occur in a number of situations:1. A randomized block design with each treatment replicated once

    in each block (balanced and complete). This is a randomizedcomplete block design (RCBD).

    2. A randomized block design with each treatment replicated oncein a block but with one block/treatment combination missing.

    (incomplete).3. A randomized block design with each treatment replicated two or

    more times in each block (balanced and complete, withreplication in each block).

    We will concentrate on 1 and discuss the others.

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    4/21

    ExpDes-4

    Single Replicate RCBD

    Design: Complete (every treatment occurs in every block) blocklayout with each treatment replicated once in each block(balanced).

    Data:

    BlockTreatment 1 2 3 ... b1 y 11 y12 y13 ... y1b2 y 21 y22 y23 ... y2b... ... ... ... ... ...t y t1 yt2 yt3 ... ytb

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    5/21

    ExpDes-5

    RCBD Soils Example

    Design: Complete block layout with each treatment (Solvent)replicated once in each block (Soil type).

    Data:

    BlockTreatment Troop Lakeland Leon Chipley NorfolkCaCl2 5.07 3.31 2.54 2.34 4.71NH4OAc 4.43 2.74 2.09 2.07 5.29 Ca(H2PO4)2 7.09 2.32 1.09 4.38 5.70Water 4.48 2.35 2.70 3.85 4.98

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    6/21

    ExpDes-6

    Minitab

    Note: Data must be stacked .From here on out, all statisticspackages will require the data tobe in a stacked structure. Thereis no common unstacked formatfor experimental designs beyondthe CRD.

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    7/21

    ExpDes-7

    Linear Model: A Two-Factor (Two-Way) AOV

    ij jiij y b jt i

    1

    1

    ij jiij y E )( BlockTreatment 1 2 3 ... b mean1 11 12 13 ... 1b 12 21 22 23 ... 2b 2... ... ... ... ... ...t t1 t2 t3 ... tb t mean 1 2 3 b

    ii

    i

    i

    0

    0

    constraintstreatment i effectw.r.t. grand mean

    block j effect w.r.t.grand mean

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    8/21

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    9/21

    ExpDes-9

    RCBD AOV

    Source SS df MS FTreatments SST t-1 MST=SST/(t-1) MST/MSEBlocks SSB b-1 MSB=SSB/(b-1) MSB/MSEError SSE (b-1)(t-1) MSE=SSE/(b-1)(t-1)Totals TSS bt-1

    Partitioning of the total sums of squares (TSS)

    TSS = SST + SSB + SSE

    df Total = df Treatment + df Block + df Error

    Regression Sums of Squares

    Usually not of interest! Assessed only todetermine if blocking wassuccessful in reducingthe variability in the

    experimental units. Thisis how/why blockingreduces MSE!

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    10/21

    ExpDes-10

    Sums of Squares - RCBD

    TSS y y

    SST b y y

    SSB t y y

    SSE y y y y

    ij j

    b

    i

    t

    i i

    t

    j j

    b

    ij i j j

    b

    i

    t

    ( )

    ( )

    ( )

    ( )

    2

    11

    2

    1

    2

    1

    2

    11

    SSBSST TSS SSE bt

    yt

    ySSB

    bt y

    b y

    SST

    bt y

    yTSS

    b

    j

    j

    t

    i

    i

    t

    i

    b

    jij

    1

    2

    1

    21 1

    2

    2

    2

    2

    )(

    )()(

    MSE E

    t MSB E b MST E

    B

    T Expectation under H a T Expectation under Ha B

    1

    2

    t i i

    T

    1

    2

    b j

    j

    B

    Expectation of MST andMSB under respective

    null hypotheses is sameas E(MSE)

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    11/21

    ExpDes-11

    Soils Example in MTB

    Must check Fitadditive model(no interaction).

    Stat -> ANOVA

    -> Two-Way

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    12/21

    ExpDes-12

    Soils in MTB: OutputTwo-way Analysis of Variance

    Analysis of Variance for SulfurSource DF SS MS F PSoil 4 33.965 8.491 10.57 0.001Solution 3 1.621 0.540 0.67 0.585Error 12 9.642 0.803Total 19 45.228

    Individual 95% CISoil Mean ---+---------+---------+---------+--------Chipley 3.16 (-----*------)Lakeland 2.68 (------*-----)Leon 2.10 (-----*------)

    Norfolk 5.17 (-----*------)Troop 5.27 (-----*------)

    ---+---------+---------+---------+--------1.50 3.00 4.50 6.00

    Individual 95% CISolution Mean -----+---------+---------+---------+------Ca(H2PO4 4.12 (------------*-----------)CaCl 3.59 (-----------*------------)

    NH4OAc 3.32 (-----------*------------) Water 3.67 (-----------*------------)

    -----+---------+---------+---------+------2.80 3.50 4.20 4.90

    Note:

    You must know whichfactor is the block, thecomputer doesnt knowor care. It simply doessums of squarescomputations.

    Conclusion:Block effect is

    significant.Treatment effect is

    not statisticallysignificant at

    =0.05.

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    13/21

    ExpDes-13

    Soils in SAS

    data soils;input Soil $ Solution $ Sulfur;datalines;Troop CaCl 5.07Troop NH4OAc 4.43Troop Ca(H2PO4)2 7.09Troop Water 4.48Lakeland CaCl 3.31Lakeland NH4OAc 2.74Lakeland Ca(H2PO4)2 2.32Lakeland Water 2.35Leon CaCl 2.54Leon NH4OAc 2.09

    Leon Ca(H2PO4)2 1.09Leon Water 2.70Chipley CaCl 2.34Chipley NH4OAc 2.07Chipley Ca(H2PO4)2 4.38Chipley Water 3.85Norfolk CaCl 4.71Norfolk NH4OAc 5.29

    Norfolk Ca(H2PO4)2 5.70Norfolk Water 4.98;

    proc glm data=soils;class soil solution;model sulfur = soil solution ;title 'RCBD for Sulfur extraction acrossdifferent Florida Soils';

    run ;

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    14/21

    ExpDes-14

    RCBD for Sulfur extraction across different Florida Soils

    The GLM ProcedureDependent Variable: Sulfur

    Sum ofSource DF Squares Mean Square F Value Pr > FModel 7 35.58609500 5.08372786 6.33 0.0028Error 12 9.64156000 0.80346333

    Corrected Total 19 45.22765500

    R-Square Coeff Var Root MSE Sulfur Mean0.786822 24.38083 0.896361 3.676500

    Source DF Type I SS Mean Square F Value Pr > FSoil 4 33.96488000 8.49122000 10.57 0.0007Solution 3 1.62121500 0.54040500 0.67 0.5851

    Source DF Type III SS Mean Square F Value Pr > F

    Soil 4 33.96488000 8.49122000 10.57 0.0007Solution 3 1.62121500 0.54040500 0.67 0.5851

    SAS Output: Soils

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    15/21

    ExpDes-15

    SPSS Soil Once the data is input use the following commands: Analyze > General Linear Model > Univariate >

    Sulfur is the response (dependent variable)

    Both Solution and Soil are factors. Solutionwould always be a fixed effect. In somescenarios Soil might be a Random factor(see the Mixed model chapter)

    We do a custom model because we only canestimate the main effects of this model andSPSS by default will attempt to estimate theinteraction terms.

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    16/21

    ExpDes-16

    SPSS Soils Output

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    17/21

    ExpDes-17

    Soils RCBD in R

    > sulf chem soil rcbd.fit = aov(sulf~soil+chem)> # anova table> anova(rcbd.fit)Analysis of Variance Table

    Response: sulf

    Df Sum Sq Mean Sq F value Pr(>F)soil 4 33.965 8.491 10.5683 0.0006629 ***chem 3 1.621 0.540 0.6726 0.5851298

    Residuals 12 9.642 0.803

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    18/21

    ExpDes-18

    Profile plot: Soils > interaction.plot(chem,soil,sulf)

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    19/21

    ExpDes-19

    Nonparametric Analysis of RCBD: Friedmans Test

    The RCBD, as in CRD, requires the usual AOV assumptions for theresiduals: Independence; Homoscedasticity; Normality.

    When the normality assumption fails, and transformations dont seemto help, Friedmans Test is a nonparametric alternative for the RCBD,

    just as Kruskal-Wallis was for the CRD. For example: ratings by apanel of judges (ordinal data).

    The procedure is based on ranks (see 15.5 in book), and leads tocalculation of FR statistic.

    For large samples, we reject H 0 of equal population medians when:2

    1, t FR

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    20/21

    ExpDes-20

    Diagnostics: Soils > par(mfrow=c(2,2))> plot(rcbd.fit)

  • 8/14/2019 U5.2-RandomizedBlockDesigns.ppt

    21/21

    ExpDes-21

    Friedmans Test: Soils

    > friedman.test(sulf, groups=chem, blocks=soil)

    Friedman rank sum test

    data: sulf, chem and soilFriedman chi-squared = 1.08, df = 3, p-value = 0.7819

    Check group and block means:

    > tapply(sulf,chem,mean)ca2 cac h2o nh4

    4.116 3.594 3.672 3.324

    > tapply(sulf,soil,mean)Chip Lake Leon Norf Troop

    3.1600 2.6800 2.1050 5.1700 5.2675