Experimental design
-
Upload
h9460730008 -
Category
Documents
-
view
500 -
download
0
Transcript of Experimental design
Experimental Research designs
Concepts
• Experiment: Lab or Field
• Treatment
• Treatment effects
• Factor-independent variable
• Blocking factor
Experimental Design
Examine possible cause and effect relationship among variables
To establish variable X causes variable Y, all three conditions should be met:
• Both X and Y should covary
• Time sequence X should precede Y
• No other factor should possibly change the dependent variable Y
Principles of Research Design
• Principle of Replication
• Principle of Randomization
• Principle of Local control
Informal Experimental Design
1. Quasi Experimental Design
2. Pretest and posttest Experimental Group Design ( caution: testing and instrumentation)
3. Post test only control design
4. Pretest posttest experimental and control group design
Validity
• External – Generalizability to other setting
• Internal- History, maturation effects, testing, instrumentation, selection bias etc
Quasi E.D.
• Experimental group treatment and measure effects
Pre test & Post test ED
Experimental group
Pretest Treatment Post test
O1 X O2
Treatment Effect=(O2-O1)
Post test only control design
Group Treatment Outcome
Experimental group X O1
Control group O2
• Treatment Effect=(O1-O2)
Pretest posttest experimental and control group design
Group Pretest Treatment Posttest
Experimental group O1 X O2
Control group O3 O4
Treatment Effect=[(O2-O1)-(O4-O3)]
Solomon Four-Group Design
Group Pretest Treatment PosttestExperimental O1 X O2Control O3 O4Experimental X O5Control O6Treatment Effect E=O2-O1 =O2-O4 =O5-O6 =O5-O3 =[(O2-O1)-(O4-O3)] * all Es are similar if cause and effect is highly valid
Double blind studies
• Researcher-subjects are unaware
• Drugs
Design a study to examine the following situation.
An organization would like to introduce one of two types of new manufacturing processes to increase to the productivity of the workers, and both involve heavy investment in expansive technology. The company wants to test the efficacy of each process in one of its small plants.
Formal E.D.
Completely Randomized
Randomized Block
Latin Square
Factorial
They are required to judge simultaneous effect of two or more variables on dependent variable.
Concept
• Factor denotes independent variable• Level denotes various gradations of factor
(high, medium and low price)• Treatment refers to various levels of
factors• Blocking factor is a preexisting variable
that has an effect on dependent variable in addition to the treatment, the impact of which is important to assess
Completely randomized design
A transportation compnay manager wants to know the effect of fare reduction by 5, 7, and 10 rupees, on the average increase in number of passengers using bus as a means of transportation.
He chooses 27 routes and randomly assign nine routes to each of treatments for a two week period.
The design would look like
Routes Number of Treatment Number of
passenger before passenger after
Group 1 O1 X1 O2
Group 2 O3 X2 O4
Group 3 O5 X3 O6
* OS SIGNFY NUMBER OF PASSENGERS
Randomized Block Design
Now company manager was interested in targeting price reduction of right routes or sectors. Reduction would be more welcomed by the senior citizens or people living in crowded areas were driving is a problem than the suburbs.
First the manager would identify the routes fally into three categories i.e. retirement areas, crowded areas and suburbs. Thus now 27 routes would get assigned to one or the other of three blocks and then randomly assigned, within the blocks to three treatments.
Randomized Block Design
Blocking Factor: Residential AreasFare Reduction Suburbs Crowded Retirement
Urban Areas
5 X1 X1 X1
7 X2 X2 X2
10 X3 X3 X3
* OS are not shown but these measures will be taken
Latin Square Design
Two blocking factor (nuisance) across rows and columns.
• Day of the week 1. Midweek (Tue to Thrus)2. Weekend3. Mon and Friday
• Residential localities
Latin Square Design
Day of the Week
Residential Mid Weekend Mon/ Fri
Area
Suburbs X1 X2 X3
Urban X2 X3 X1
Retirement X3 X1 X2
Factorial Design
It enables us to check manipulations of two or more manipulation at the same time on dependent variable
The manager now is interested in knowing passenger increases if he used three different types of buses( Luxury, standard and regular). Using fare reduction and type of vehicle simultaneously
Fare reduction and Vehicle used
Bus Fare Reduction Rates
Type of Bus 5 7 10
Luxury X1Y1 X2Y1 X3Y1
Standard X2 Y2 X1Y2 X3Y2
Regular X3Y3 X2Y3 X1Y3
Any doubts?Thank you