Experimental research and design

Experimental Research and Designs

INTRODUCTION

The term experiment means an observation under controlled conditions as defined by Chapin.

Experimental research- It may be defined as a research that leads researchers to arrive at some reliable and valid conclusions by establishing possible cause-effect relationships on the basis of a well planned and systematic observations under controlled conditions.

Experimental research is mainly aimed at establishing the possible cause and effect relationship

CERTAIN PROCESSES INVOLVED IN AN EXPERIMENTAL RESEARCH

1.     The fundamental goal of an experimental research is to establish a causal (cause and effect)relationship between two variables to demonstrate that changes in one variable are directly responsible for changes in other variable (complex experiments may involve several variables). In abstract term, it is nothing but to establish the fact that a certain action X is alone responsible for creating the effect Y.

2.      To explore how much amount of cause will produce how much of effect .

3.     In the experimental research, a researcher is supposed to manipulate

           or introduce variations in a independent variable to observe the changes.

4.      Research process involves scientific observation of the subsequent changes in the dependent variable   by varying the conditions in the independent variable while holding the confounding   variables influencing the results of the study nullified or constant.

5.     The researcher is also supposed to ensure that the effect Y has been caused by the factor  X alone and not by any other factor that can affect the results of the study

6.     In addition to making an attempt to eliminate or nullify the effect of the factors or conditions other than X on the variable Y, he  tries to control the conditions prevailing in his experiment by

(i)Creating two groups of participants matching in almost all characteristics other

than that inherent in the causative factor and by naming them as experimental and control groups.

 (ii) Subjecting experimental group to experimentation and Keeping away the participants  of the control group from the exposure/effect of the factor X and then comparing the two groups for arriving at the conclusion.

BASIC CONCEPTS USED IN EXPERIMENTAL RESEARCH

1.Independent, Dependent and Extraneous or Intervening Variables

A variable is defined by any characteristic or condition (of man and his environment)

that can be varied and thus can be attributed with different values.

The condition itself is not a variable, but its characteristics are.

 While conducting experiment behavioural sciences, a researcher encounters a number of variables, i.e. These variables are known as independent, dependent and extraneous or intervening variable

Independent variables: These in an experimental study are those variables (characteristics and conditions of the participants and their environment) that are very much in control of the experimenter and are subjected to deliberate manipulation and variation by the experimenter for observing their effects on the dependent variables.

Dependent variables: These are those variables that can to get affected or changed with the introduction and manipulation of the independent variables. These are dependent on the independent variables

Extraneous or intervening variables: These are those variables that are of

no interest to researchers but which might influence and interfere with the results of the study. These variables needs to be controlled and held constant by the researcher to clearly establish the cause and effect relationship between the independent and dependent variable.

In case these variables are not controlled they can turn Confounding and the Confounding variables exert a great effect on DV by getting systematically related to the IV

Control Over the Variables

The experimenter has to take control over all the variables involved in his study of independent, dependent and extraneous or confounding.

STEPS CONTROLLING THE VARIABLES:

1.  Identification task: The first step is to identify and name the variables.

The primary goal of an experimental research is to establish a cause and effect relationship between two set of variables .Thus, the researcher, first of all had to identify which one of the variable involved in his study is the cause and which variable is the effect and then naming them as independent variable and dependent variable.

In addition to these two specific variables of interest, there may be a many other variables in the study, out of these many variables the researcher had to be careful mainly about those confounding variables that might affect the result of the study.

For ex. The colour of wall may be an extraneous variable but not affect the study but temperature of the room may become a confounding variable as it might affect the performance of the participant and thus might affect the result of the study

IDENTIFYING CRITERIA OF CONFOUNDING VARIABLES

i) An extraneous variable becomes confounding variable only if it influence the dependent variable

(ii) A confounding variable must vary systematically with the independent variable not randomly

2. Controlling the independent variable

In experimental research a researcher is supposed to take control of the independent variable by getting it manipulated or varied for creating difference in the dependent variable

3. Controlling the dependent variable In Experimental research, a researcher can take control of the dependent variable by subjecting it to the process of observation and measurement.

4. Controlling the extraneous or confounding variables

In order to say with certainty that the differences observed in the dependent variable is because of the independent variable and not due to any extraneous variable, the researcher employ the following methods and techniques:

Methods of exercising control over extraneous or confounding variables:

1.Eliminating or Holding constant - Eliminating or holding participants and environmental variables constant and restricting the range of variability.

2.Randomization -The most convenient and powerful tool of controlling. In this the researcher use the techniques of random Sampling for selection of participants in research study

May not be useful when the sample is small

3.Making use of additional independent variable- In this the confounding variables play the role of an additional independent variable. Since it varies systematically with the IV, it works to help in the study. Instead of holding it constant or eliminating it, it's included in the study. This helps in exercising control over the confounding variables as well as studying the effects of an additional IV along with its nature of interaction with the original IV

4.Statistical control-Use of statistical methods may help in controlling the effects of confounding variables

5.Matching-Getting the participants of the control group and experimental group matched with respect to suspecting confounding variables may help in controlling the effects of these variables.

CONTROL AND EXPERIMENTAL GROUPS

To establish the fact that the changes occurred in the DV are not accidental or a matter of chance, it is required to form two identical groups of participants known as experimental and control group to use them for comparison needed in the study.

Experimental and control groups are similar on many grounds, the only difference is that the experimental group receives the treatment while control group does not receive the treatment.

                         EXPERIMENTAL DESIGNS

Purpose served by experimental design:
1.Formulating the experiment and and control group
2.Identifying and defining the IV and DV
3.How to manipulate IV how to measure DV
4.Ways to make observation and record it

Types of Research Design in behavioral research
1 experiment or within subject design
2 true experiment or between subject design
3 quasi experiment design
4 ex Post facto design

(quasi and ex post facto are not discussed here)

                          (1) Pre- experiment or within subject design

These are characterized by:

1.lack of control group

2. Failure to provide for equivalence of control group

3. Two sets of scores are obtained from same sample of participants hence the name within   subjects

4. Two sets of scores are obtained by subjecting the same group of participants to two treatment conditions and then comparing the scores for establishing causal relationship existing between IV and DV.

5.It is termed as pre-experimental as it does not fulfill the criteria of a true experiment 

6.Only compares the intra-individual differences existing in same group of participants resulting from different levels of treatment


Key features:

1.Using a single group of participants

2.Exposing each participant to each level of the IV (treatment) at least once

TYPES:

A.One Shot case study
B.Pretest -posttest design
C.Repeated measure design
D.Static group comparison design

Symbols used

1.IV- Independent variable
2.DV-Dependent variable

A. One Shot case study – Conducted in a single shot at one time to compare it with the general expectation of what would have happened if treatment have not been applied

In this,

1.There is only one group

2.No randomisation in selection of participants

3.No pre-test is conducted to know the characteristics of the participants before subjecting them to some treatment

4.Only post test is carried out
  

Limitations
1.provide weak basis for generalisation

2.unable to determine if the result obtained is due to the treatment introduced or it was already present in the participants since no pre-test is conducted

B. Pre-test posttest pre-experimental or within subject design

In this pre-testing is done before introducing the treatment to the experimental group
it is used mainly to study the effect of an treatment introduced.

 

Steps:

1. Only one group of participants is used 

2. Randomisation is not done before selecting the subject 

3. First pre testing is done to know the pre-existing effect of the DV, then experimental  treatment is introduced 

4. Following the treatment post testing is conducted

5.Finally the researcher take note of the difference between the pre-test and post-test scores by accounting it due to the effect of treatment

C. Repeated measure design

Repeated measures design allows multiple measures per design

In such research design the researcher may test the outcomes of the different levels of treatment or a different treatment altogether by taking repeated measurement of the same group of participants

D.Static group comparison design

In this ,the researcher use two groups of experimental and control group ,similarly like true experiment design but does not provide the equivalence between the two groups.

The control group remain static in the sense that it is used only for the comparison not for testing the influence of the treatment aimed in the study

Advantages of within subject design:

1.Can be conducted with less number of participants so useful in situation where it is difficult to obtain the participants or they are rare (such as astronauts players)

2.Advantage of not getting affected by individual differences because use of same participant in all different level of treatment

3.Low level of error variance so, less likelihood of making type 2 error

Disadvantages:

1.No proper control over extraneous variable 

2.No guarantee of availability of participant for repeated measures required for completion of the study

3.History or time related problem may arise 

4.Problem of carry over effect or practice fatigue effect
  Carry over effect- it means after completing the first part of the experiment, experience gained or altered environment may influence the result .

Practice effect -can cause improvement or decline in performance due to fatigue

Solution for carryover or practice effect:

The researcher can use the technique of counterbalancing that is using more than one sequence or order in his study

Counterbalancing has two forms:

1.counterbalancing in complete within subject design-Testing the participant to each condition several times using different orders till all possible orders are done.

 In this any effect associated with presenting condition is counterbalanced by presenting the condition in opposite order to same participant.

2.Counterbalancing in incomplete within subject design-In this carry over or practice effect are balanced by administering each condition to each participant only once

 Each participant then receives a unique order of condition at least once, but not all possible sequence of that condition

The experimental condition are ordered randomly.

                  (2).True experiment for within subject design

A researcher here obtain two sets of scores from two equivalent experimental and control group.
Two groups( experiment and control) are obtained from same population ,they are subjected to two different treatment conditions and comparing the scores of these groups to arrive at some conclusions

Characteristics of true experiment design:

1.participants are divided into two equivalent group (experimental and control) by using randomisation or matching technique

2.Two sets of scores are obtained from the groups to compare the different treatment condition

3.Each participant is tested once only

Advantages:

1.Since only one score is obtained from each participants so each individual score is independent of other score

2.Each participant is tested once only show the score is generally not affected by practice, experience ,fatigue, boredom or contrast effect

3.Random selection of participant may be helpful in controlling confounding variables

Disadvantages of between subject design:

1.Not beneficial in using when number of participants is small

2.Require a great deal of time and energy to gather large number of participants

3.Error variance is relatively high since the two groups not only differ on the IV but also on  many different factors

4.The difference between the condition may be due to the difference between the two groups and not only due to the IV

Types of true experimental design:

1.Post-test only equivalent group design

2.Pre-test- posttest equivalent group design

2.Solomon three or four equivalent group design

1.Posttest only equivalent group design

Steps:

1.Researcher randomly select participants from the population

2.Then randomly assigns participant to experimental and control group to form two equivalent groups

3.Equivalent and control groups are then subjected to two different treatments

4.No pre-test is carried out hence the name posttest only

5.The posttest scores of these groups are then compared

Major limitations :

The researcher has no idea about the initial behaviour of the participants prior to the introduction of the treatment conditions so, it becomes difficult to know whether the difference are due to the treatment given or due to the initial differences existing in the group

2.Pretest posttest equivalent group design

This design is similar to the posttest only equivalent design except the fact that participants here are pretested

It is useful in ruling out the possibility of difference in the result due to the initial differences existing between the groups 

Disadvantages:

1.Can be affected by interaction effect of testing that is; the difference between the pretest and posttest scores may have been influenced by carry over effect ,fatigue or boredom

but this can be controlled by adding two more groups that do not experience the pre testing also known as Solomon three or four equivalent design

3.Solomon three or four equivalent group design

Steps:

1.The researcher forms four groups through randomisation, two as experimental group and two as control group

2.Out of four groups, two groups (experimental and control) follow the same process of pretest -posttest equivalent group design

3.No pre testing is done on the other two groups (experimental and control) so they follow the same process of posttest only equivalent group design

4.It is nothing but combination of pretest only and pretest posttest equivalent group design

Advantages:

We can compare the first two groups to know the difference due to the treatment given and the other two groups to know whether the treatment is more effective in the absence of protesting