Scribd
Upload
194 views11 pages

Basic Statistical Design of Experiments

This document provides an overview of basic statistical design of experiments. It discusses key concepts like treatments, treatment levels, experimental units, experimental error, sampling error, fixed and random effects, replication, randomization, and experimental designs including completely randomized design, randomized complete block design, and Latin square design. The purpose of experimental design is to obtain the maximum information from available sources by properly assigning treatments to experimental units and accounting for sources of variation.

Uploaded by

Kean Cardenas
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
194 views11 pages

Basic Statistical Design of Experiments

This document provides an overview of basic statistical design of experiments. It discusses key concepts like treatments, treatment levels, experimental units, experimental error, sampling error, fixed and random effects, replication, randomization, and experimental designs including completely randomized design, randomized complete block design, and Latin square design. The purpose of experimental design is to obtain the maximum information from available sources by properly assigning treatments to experimental units and accounting for sources of variation.

Uploaded by

Kean Cardenas
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 11

Basic Statistical Design of Experiments

Scientific Research

 A study conducted in contribution to science by a systematic collection, interpretation, and


evaluation of data
 Must be systematically planned before performing it.

What are the roles of statistics in scientific research?

In scientific research, statistics deals with:

T3
Treatment or Factor

 Includes the different procedures or conditions to be compared

Treatment Levels

 Pre-set quantities of a quantitative treatment or categories of a treatment under study

Example:

1. Type of sugar (Icing, Granulated, Dark Muscovado)


2. Type of irrigation system (surface, sprinkler, drip, subsurface)
3. Teaching methods (online, Face-to-Face, Blended)

Try this…
Note: Unit is kg/ha

Experimental Error consists of the pooled variation among the 5 plots in the same sulfur application.

Sampling Error consists of pooled variation among the three hotdogs taken from the same package.

Experimental Error

 variation in the observed values of the response variable from experimental units treated alike

Some Sources of Experimental Error

 inherent variability of the experimental materials used


 errors in experimentation
 errors in observations and measurements
 combined effects of all extraneous factors

Sampling Error

 Measure of variation among sampling units within an experimental unit


Fixed Effects

- All factors under test are ‘fixed’ factors


- A factor is considered ‘fixed’ when its levels are selected on purpose
 Interest is in the effects of the selected levels of a fixed factor
 No inferences shall be directed at any treatment level that is not included in the experiment

Random Effects

- all factors under test are ‘random’ factors


- a factor is considered ‘random’ when the levels of the factors tested are a random sample from
a population of levels
 interest is in the population of treatment levels from which the random sample of treatment
levels was drawn
 population of treatments will then be described in terms of the sampled treatments

Example:

A drug study uses 0 mg, 5 mg, and 10 mg of an experimental drug.

Experiment Randomized Experiment


Characterized by the following:  An experiment where the assignment of
1. Presence of treatments and experimental units to treatment groups is randomized
units to be used,  Goal is to see the effect of the treatment
2. By the way treatments are assigned to while controlling for other factors
units, and  Presents the strongest support for
3. By the responses that are measured. causation
RANDOMIZED EXPERIMENT

Replication

 Is needed for the validity of the experiment


 Is the repetition of the application of treatments on a number of experimental units

Functions of replication:

 To provide an estimate of the experimental error


 To increase the precision of the estimates of the parameter
 To increase the scope of the experiment

Randomization

 Aside from providing ways to estimate experimental error, the estimate should also be valid.
 This is done through randomization or a process that ensures treatments will have an equal
chance of being assigned to an experimental unit.

Local Control (Error Control)

 Is any technique used to minimize the experimental error


 Makes the design efficient by making the tests more sensitive and powerful
 Common Techniques of Local Control:
 Use of the most appropriate experimental design
 Use the proper shape and size of the experimental unit
 Use of concomitant variable
 Grouping, blocking, or balancing
Note:

Experimental Design

 Concerns with planning experiments in order to obtain the maximum amount of information
from the available sources
 Involves the assignment of treatments to the experimental units
Completely Randomized Design

 Treatments are assigned completely at random


 Each experimental unit has the same chance of receiving any treatment
 Any difference among experimental units receiving the same treatment – considered
experimental error
 Appropriate for homogeneous experimental units – lab experiments
o Environmental conditions – easily controlled as compared to field

Example:

- 30 pots (1-30)
- 3 treatments (A, B, C)
- Each pot is going to receive a single treatment
- Treatments are assigned at random

Experimental Layout

Sources of Variation:

 Treatments – the source of variation of interest


 Experimental Error – factors beyond control

Actual Example:

Determine and lay out the experimental design of the effects of fertilizer dose on the fresh weight of
plants.
 Experimental Design – CRD
 Independent variable – fertilizer doses
 Treatments – different doses of fertilizer (A, B, C)
 Dependent (response) variable – the fresh weight of plants

Fresh Weight (g)of Plants under Different Doses of Fertilizer

Note: Raw data table

Randomized Complete Block Design

 Most widely used experimental design in field research


 A primary distinguishing feature of RCBD from CRD is the presence of blocks of equal size
 Each block contains all treatments
 Purpose of blocking is to reduce experimental error
 By eliminating a known source of variation among experimental units other than
treatment
 Experimental units are grouped into blocks according to the known or suspected source of
variation
 Known source of variation
 May exist in the form of a gradient – in one direction
 Can be isolated by the blocks
 Conditions – the source of variation
 As homogeneous as possible within blocks
 May differ greatly within blocks
 Each block has all treatments
 The source of variation other than treatment can be corrected
 Treatments are assigned at random within each block
 Complete randomization within a block
 Each block is with a separate randomization
 Flexible design – any number of treatments and blocks possible
 Long and narrow blocks – designed
 Length of blocks – perpendicular to the direction of the gradient
 If there are two sources of variation that occur in a gradient – use Latin Square Design
Example: Effects of fertilizer dose on the fresh weight of plants

 Experimental design – RCBD


 Independent variable – fertilizer doses
 Treatments – different doses of fertilizer (A-H)
 Dependent (response) variable – the fresh weight of plants

The fresh weight of plants is measured after 6 months of planting to see the effect of fertilizer in the
field trial. There is a river on one side – soil near the river is more moist and decreases as we move
away from it.

 Can be effectively analyzed through ANOVA


 Three sources of variability in RCBD
 Treatment – a variation of interest
 Block – unavoidable variation
 Experimental error
Latin Square Design

 RCBD – minimizes the contribution of one source of variation other than treatment
 LSD can handle two sources of variation that occur in a gradient
 Every treatment occurs only once in each row and column

For example, a field has a river on one side and a road on the other side

 Can have a maximum of only 16 plots


 A, B, C, D are types of treatments
 Each treatment appears only once in a row (river gradient, A) and column (road gradient, B)

Example: Effects of fertilizer dose on the fresh weight of plants

 Experimental design – LSD


 Independent variable – fertilizer doses
 Treatments – different doses of fertilizer (A, B, C, D)
 Dependent (response) variable – fresh weight of plants

Fresh Weight (g)of Plants under Different Doses of Fertilizer

Note: Raw data table

You might also like