Conducting Experimental Trials Gary Palmer. Scientific Method Formulation of Hypothesis Planning an experiment to objectively test the hypothesis.

Conducting Conducting Experimental TrialsExperimental Trials

Gary PalmerGary Palmer

Scientific MethodScientific Method

Formulation of HypothesisFormulation of Hypothesis Planning an experiment to objectively test Planning an experiment to objectively test

the hypothesisthe hypothesis TreatmentsTreatments Experimental designExperimental design ReplicationReplication

Careful observation and collecting of data Careful observation and collecting of data from the experimentfrom the experiment

Interpretation of the experimental resultsInterpretation of the experimental results

ExperimentExperiment

TreatmentsTreatments• ProwlProwl• OrtheneOrthene• RidomilRidomil

ExperimentExperiment

TrteatmentTrteatment RateRate NozzlesNozzles

MH MH 2 gal2 gal FineFine

MH + Prime+MH + Prime+ 1.0 gal + 0.5 gal1.0 gal + 0.5 gal CoarseCoarse

MH + Prime+MH + Prime+ 1.5 gal + 0.5 gal1.5 gal + 0.5 gal CoarseCoarse

MH + ButralinMH + Butralin 1.0 gal + 0.5 gal1.0 gal + 0.5 gal CoarseCoarse

MH + ButralinMH + Butralin 1.5 gal + 0.5 gal1.5 gal + 0.5 gal CoarseCoarse

ExperimentExperiment

TrteatmentTrteatment RateRate NozzlesNozzles

MH MH 2 gal2 gal FineFine

MHMH 2 gal2 gal CoarseCoarse

MH + Prime+MH + Prime+ 1.0 gal + 0.5 gal1.0 gal + 0.5 gal CoarseCoarse

MH + Prime+MH + Prime+ 1.5 gal + 0.5 gal1.5 gal + 0.5 gal CoarseCoarse

MH + ButralinMH + Butralin 1.0 gal + 0.5 gal1.0 gal + 0.5 gal CoarseCoarse

MH + ButralinMH + Butralin 1.5 gal + 0.5 gal1.5 gal + 0.5 gal CoarseCoarse

Effects of Application of MH by Coarse vs

Fine Nozzles on Yield of Burley Tobacco

0

500

1000

1500

2000

2500

3000

3500

4000

Yie

ld lb

/a

Fine 2398 3105 1713 2352 2074 2697 2194 2911 2355

Coarse 2617 3474 1998 2498 2291 2949 2397 3014 2524

Bath Fleming Jackon Knox LaurelLawrenc

eNelson

Pendleton

Average*

How many reps are needed?How many reps are needed?

A.A. <3<3

B.B. 33

C.C. 44

D.D. 55

E.E. 66

F.F. >6>6

G.G. It dependsIt depends

Why Replicate?Why Replicate?

Treatment is repeated two or more timesTreatment is repeated two or more times In a randomized patternIn a randomized pattern

Provides an estimate of experimental errorProvides an estimate of experimental error Provides a more precise measurement of Provides a more precise measurement of

treatment effecttreatment effect

How many reps are needed?How many reps are needed?

Number requiredNumber required Depends on magnitude of detectable Depends on magnitude of detectable

differences desireddifferences desired• 100 lb/a, 200 lb/a, 600 lb/a, ?100 lb/a, 200 lb/a, 600 lb/a, ?

Inherent variability of subject matterInherent variability of subject matter• CV = Coefficient of VariationCV = Coefficient of Variation• Expressed in %Expressed in %

Inherent variability Inherent variability Coefficient of Variation (CV)Coefficient of Variation (CV)

Small Grain (6-8%)Small Grain (6-8%) Tobacco yields (16-18%)Tobacco yields (16-18%) Black shank (50-70%)Black shank (50-70%) Johnson grass (30-40%)Johnson grass (30-40%)

Inherent Variability Inherent Variability

Actual CV%Actual CV% Tobacco Location Tobacco Location MH Residue MH Residue

• ClaytonClayton 44.444.4• ReidsvilleReidsville 29.929.9

Wheat LocationWheat Location Stand Stand VigorVigor ClaytonClayton 6.3 6.3 55.5 55.5

ReidsvilleReidsville 5.3 5.3 64.7 64.7

Burley Tobacco No-till Variety TrialBurley Tobacco No-till Variety TrialTaylor Co. – Adam Pike FarmTaylor Co. – Adam Pike Farm

394 377 423 451 446 342 384 375 388 400

869 8881146 853 1051

878 933 854 841 874

1156 10771030

11101124

1197 1034 989 1051 1019

570 549532

490606

614625 503 653

97

213

247

548 96

2944278328432900303032282904313028912989 296

0

500

1000

1500

2000

2500

3000

3500

Variety

Yie

ld l

b/a

Flyings Lugs Leaf Tips Total

Burley Tobacco YieldBurley Tobacco Yield Harrison Co. – Ricci Rowland FarmHarrison Co. – Ricci Rowland Farm

203 235 311 269 317 279 281 208

9321399 1153 1248 1095 1071 1257

1064

470

715623 670 701 650

780

692

83

354

183244

336326 315 337

276

322

307 86

1849 2685 2412 2502 2450 2275 2639 2270 625

0

500

1000

1500

2000

2500

3000

Yie

ld lb/a

Flyings Lugs Leaf Tips Total

How can we increase precision?How can we increase precision?

The precision of an experiment can always The precision of an experiment can always be increased by additional replicationsbe increased by additional replications

Degree of improvement falls off rapidly as Degree of improvement falls off rapidly as replication increasereplication increase Example: To double the precision with which Example: To double the precision with which

two means can be separated in a trail with two means can be separated in a trail with four replications, reps would have to increase four replications, reps would have to increase to 16. to 16.

Arrangement of ReplicationsArrangement of ReplicationsConfounded with fields/soil typeConfounded with fields/soil type

Field 1

Field 3

Field 2

Wet Spot

Rep 1

Rep 3

Rep 2

Rep 4

Arrangement of ReplicationsArrangement of ReplicationsConfounded with soil typeConfounded with soil type

Rep 1

Rep 2

Rep 3

Rep 4 Rep 5

Rep 6

Rep 7

Rep 8

Arrangement of TreatmentsArrangement of TreatmentsRandomized complete BlockRandomized complete Block

Rep 2

Rep 3

Rep 4 Rep 5

Rep 6

Rep 7

1 2 3 4 5 6 7

2 45 13 76

Arrangement of TreatmentsArrangement of TreatmentsBalanced Incomplete Block DesignBalanced Incomplete Block Design

1 2 3

4 5 6

1

14

82

5

5

1

Blk 1

Blk 2

7 8 9 Blk 3

4 7 3 6

Blk 6Blk 5Blk 4

9

9

Blk 11Blk 10

Blk 8Blk 7

Blk12

Blk 9

7

7

2

2

3

3

6

6

4

5REP 1

8

REP 2

REP 3

1 8 9

REP 4

Arrangement of TreatmentsArrangement of TreatmentsUnbalanced Incomplete Block Design Unbalanced Incomplete Block Design

3 5 6 7

5 7

2

1 74

4

6

1

1

2

2

7

1 3 6

3

3 4 5 2 4 5 6Blk 6

Blk 2

Blk 3

Blk 4

Blk 5

Blk 1

Blk 7

Replication Confounded with TimeReplication Confounded with Time

Not enough growth chambers available?Not enough growth chambers available?

Simulated DataSimulated DataTreatmentTreatment Lb/aLb/a

AA 33003300

BB 32003200

CC 31003100

DD 29002900

EE 28002800

FF 27002700

Herbicide TreatmentHerbicide Treatment

TreatmentTreatment MeanMean

CheckCheck 18001800

ProwlProwl 23002300

CommandCommand 22002200

SpartanSpartan 22502250

Prowl+CommandProwl+Command 23352335

Prowl+SpartanProwl+Spartan 22952295

Command+SpartanCommand+Spartan 24002400

Herbicide TreatmentHerbicide Treatment

TreatmentTreatment Rep 1Rep 1 Rep 2Rep 2 MeanMean

CheckCheck 11001100 25002500 18001800

ProwlProwl 22002200 24002400 23002300

CommandCommand 23002300 21002100 22002200

SpartanSpartan 23002300 22002200 22502250

Prowl+CommandProwl+Command 20002000 26702670 23352335

Prowl+SpartanProwl+Spartan 16001600 28002800 22002200

Command+SpartanCommand+Spartan 28002800 20002000 24002400

The Completely Randomized Design

The CRD is the simplest of all designs. Replications of treatments are assigned

completely at random to independent experimental subjects.

Adjacent subjects could potentially have the same treatment.

The Completely Randomized Design Sample layout:

Different colors represent different treatments. There are 4 (1-4) treatments with 4 replications (A-D) each

Experimental DesignExperimental Design

Randomized Complete Block DesignRandomized Complete Block Design

RepRep AA BB CC DD

44 11 33 33

11 44 44 22

22 33 22 44

33 22 11 11

Experimental DesignExperimental Design

SamplingSampling

TreatmentTreatment 11 22 33 44

SamplesSamples AA

BB

CC

DD

AA

BB

CC

DD

A A

BB

CC

DD

A A

BB

CC

DD

Experimental DesignsExperimental Designs

FactorialFactorial One or more factors in one experimentOne or more factors in one experiment

Study factors that might make cured Study factors that might make cured tobacco darkertobacco darker Nitrogen (200, 300, 400 lb/a)Nitrogen (200, 300, 400 lb/a) Phosphorus (Low, soil test, high)Phosphorus (Low, soil test, high) Potassium (Low, soil test, high)Potassium (Low, soil test, high) Harvest date (3, 4, 5 weeks)Harvest date (3, 4, 5 weeks)

Experimental DesignsExperimental Designs

FactorialFactorial One or more factors in one experimentOne or more factors in one experiment

Study factors that might make tobacco darkerStudy factors that might make tobacco darker Nitrogen (200, 300, 400 lb/a)Nitrogen (200, 300, 400 lb/a) Phosphorus (Low, soil test, high)Phosphorus (Low, soil test, high) Potassium (Low, soil test, high)Potassium (Low, soil test, high) Harvest date (3, 4, 5 weeks)Harvest date (3, 4, 5 weeks)

3x3x3x3x4=324 plots 3x3x3x3x4=324 plots If 4 rows by 30 ft = 3.12 acresIf 4 rows by 30 ft = 3.12 acres

InterpretationInterpretation

What if plant response to nitrogen What if plant response to nitrogen changes as potassium rates change?changes as potassium rates change? How do you interpret?How do you interpret?

What if plant response to nitrogen What if plant response to nitrogen changes as harvest date increase and changes as harvest date increase and potassium rates decrease?potassium rates decrease? How do you interpret?How do you interpret?

Split-Plot DesignSplit-Plot Design

Main plotMain plot Ex. Nitrogen RateEx. Nitrogen Rate Least precisionLeast precision

Split PlotSplit Plot Ex. Harvest DateEx. Harvest Date Most precisionMost precision

Split-Plot DesignSplit-Plot DesignNitrogenNitrogen

RateRateHarvest days after toppingHarvest days after topping RepsReps

200200 5050 2020 3030 4040 DD

400400 2020 4040 5050 3030

400400 3030 5050 4040 2020

CC200200 2020 4040 5050 3030

400400 3030 5050 2020 4040

BB200200 5050 4040 3030 2020

400400 4040 2020 5050 3030

AA200200 2020 4040 3030 5050

Yield of Burley Tobacco Varieties with Yield of Burley Tobacco Varieties with Different Fertility – UK Woodford Co. FarmDifferent Fertility – UK Woodford Co. Farm

2000

2200

2400

2600

2800

3000

100+100 200+0 300+100 400+0

Nitrogen (units/a broadcast+sidedress)

Yie

ld l

b/a

KT 204 V

NC 6 V

TN 90 V

KT 204 N

NC 6 N

TN 90 N

V = VentedN = Nonvented

Nitrogen was applied in the ammonium nitrate form .

Experimental Unit SizeExperimental Unit Size

In general, variability decreases as plot size In general, variability decreases as plot size increase.increase.

However, once a certain size is reached the However, once a certain size is reached the increase in precision falls off rapidly.increase in precision falls off rapidly.

Little gain in field plots beyond 0.1 acreLittle gain in field plots beyond 0.1 acre Precision is good at 0.01 to 0.02 for most crop (4 Precision is good at 0.01 to 0.02 for most crop (4

rows by 30-35 ft in tobacco)rows by 30-35 ft in tobacco) Individuals may be used as an experimental unitIndividuals may be used as an experimental unit

Trees, cows, goats, etc.Trees, cows, goats, etc.

Experimental Unit SizeExperimental Unit Size

CovarianceCovariance Removing variability of some independent Removing variability of some independent

variablevariable• Difference in beginning weight of cows in testDifference in beginning weight of cows in test• Differences in plant standsDifferences in plant stands