Experimental Research

DIBRUGARH UNIVERSITY

TUTORIAL ON

“EXPERIMENTAL RESEARCH”

CENTRE FOR MANAGEMENT STUDIES

Presented by: (GROUP: 3)

06 : Arup Bordoloi17 : Joyshree Borah36 : Nikhil Modi

48 : Rabby Ullah55 : Sandeep Das

MASTER OF BUSINESS ADMINISTRATION – 2nd SEMESTER

An experiment is generally used to infer a causality where a researcher actively manipulates one or more causal variables and measures their effects on the dependent variables of interest.

What is an Experiment?

It is a collection of research designs which use manipulation and controlled testing to understand causal processes. Generally, one or more variables are manipulated to determine their effect on a dependent variable.

What is an Experimental Research?

In simple terms Causality means the cause and effect relationship. The cause and effect relationship often tends to be probabilistic in nature as the changes in the dependent variable may be caused by a number of other variables.

Concept of causality

Concomitant variation Time order of occurrence of variables Absence of other possible causal factors

Necessary Conditions for making Causal Inferences

CONCEPTS USED IN EXPERIMENTS

• INDEPENDENT VARIABLES

• DEPENDENT VARIABLES

• EXTRANEOUS VARIABLES

VALIDITY IN EXPERIMENTATION

1. INTERNAL VALIDITY

2. EXTERNAL VALIDITY

FACTORS AFFECTING INTERNAL VALIDITY

• MATURATION

• TESTING

• INSTRUMENTATION

• SELECTION BIAS

• TEST UNIT MORTALITY

FACTORS AFFECTING EXTERNAL VALIDITY

METHODS TO CONTROL EXTRANEOUS VARIABLE

• RANDOMIZATION

• MATCHING

• STATISTICAL CONTROL

AQUARIUM EXPERIMENT

DILEMMA: FISHES ARE HAVING WHITE SPOT DISEASEHYPOTHESIS: USE OF RID-ALL GENERAL AID CAN CURE THE FISHES

EXPERIMENTALBOWL

CONTROLLEDBOWL

X : FEW DROPS OF THE MEDICINE ARE ADDED IN THE EXPERIMENTAL BOWLNO MEDICINE IS ADDED IN CONTROLLED BOWLTHEN TWO FISHES FROM THE AQUARIUM ARE PUT INTO THE TWO BOWLS

AFTER APPROX. 24 HOURS

O1 : FISH FROM EXPERIMENTAL BOWL BECAME FREE OF WHITE SPOTSO2: FISH FROM CONTROLLED BOWL IS STILL HAVING WHITE SPOTS

EXPERIMENTAL DESIGN

PRE-EXPERIMENTAL STATISTICALTRUE - EXPERIMENTALQUASSI -EXPERIMENTAL

STATIC GROUP

ONE GROUP PRE TEST- POST TEST

ONE-SHOT CASE STUDY TIME SERIES

MULTIPLE TIME SERIES

SOLOMON FOUR GROUP

POST TEST ONLY CONTROL GROUP

PRE- TEST- POST TEST CONTROL

GROUP

COMPLETELY RANDOMIZED

RANDOMIZED BLOCKS

LATIN SQUARE

FACTORIAL

PRE- EXPERIMENTAL DESIGN

Do not use any randomization procedureCannot control extraneous variables Internal validity is highly questionable

ONE SHOT CASE STUDY

Symbolic representation:

X O

One Group No random assignment

ONE GROUP PRE TEST- POST TEST

Symbolic representation:

O1 X O2

One Group Not random selection Test units are subjected to both pre and post treatment Treatment effect : O2 – O1

STATIC GROUP

Symbolic representation:

Group 1: X O1

Group 2: O2

Two Groups Not random selection Group1 is experimental group, Group 2 is control group Treatment effect : O1 – O2

QUASSI- EXPERIMENTAL DESIGN

Can Control when measurement are taken and whom they are taken

Do not use randomization Cannot control extraneous variables

TIME SERIES DESIGN

Symbolic representation:

O1 O2 O3 O4 X O5 O6 O7 O8

Not randomization

TIME SERIES DESIGN

TRUE- EXPERIMENTAL DESIGN

Uses randomization procedureCan be able to eliminate extraneous variables both

from experimental and control group

PRE- TEST- POST TEST CONTROL GROUP

Symbolic representation:

Experimental group: R O1 X O2Control group : R O3 O4

Two Groups Random selection Findings: O2 – O1 = Treatment +extraneous Variable O4 – O3 = Extraneous Variable Cannot find interactive testing effect

POST TEST ONLY CONTROL GROUP

Symbolic representation:

Experimental group: R X O1Control group : R O2

Two Groups Random selection Findings: O1 – O2 = Treatment effect

SOLOMON FOUR GROUPSymbolic representation:

Experimental group1: R O1 X O2Control group1 : R O3 O4Experimental group2: R X O5Control group2 : R O6

Can find both effect of treatment and interactive testing effect

STATISTICAL DESIGN

Allow for statistical control and analysis of external variables

Effect of more than one level of independent variable on the dependent variable can be manipulated

The effect of more than one independent variable can be examined

Effect of specific extraneous variable can be controlled

COMPLETELY RANDOMIZEDUsed when effect of only one independent variable is

considered. Independent variable should be measured in nominal

scale i.e. it should have number of categories. Each category is considered as treatment All test groups are selected at random Statistical technique: Analysis of Variance (ANOVA) Limitation: Does not consider extraneous variables

Sample shops

Price of goods

Low Medium High

1

2

3

4

5

6

7

8

9

10

Average

RANDOMIZED BLOCK DESIGN

Overcomes the limitation of Completely Randomized block design

Considers the extraneous variables

Sample shops

Price of goods

Low Medium High

Large

1 2 3 4 5

Medium

1 2 3 4 5

Small

1 2 3 4 5