EXPERIMENTAL RESEARCH IN SOCIAL RESEARCH

BY

WISDOM IYEKEKPOLO
COLLINS ODIGIE
OSAHON USIOBAIFO
JOAN IDUMWONYI
GEORGE OKUNGBOWA

A SEMINAR PAPER PRESENTED IN THE DEPARTMENT OF POLITICAL SCIENCE AND PUBLIC ADMINISTRATION, UNIVERSITY OF BENIN, BENIN CITY

IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR POL 801 (RESEARCH METHODOLOGY)


COURSE LECTURER: DR. O. J. OFFIONG




APRIL, 2011




INTRODUCTION
There are a lot of classifications of research put forward by different scholars over the years. A cursory look at these classifications reveals that a common trend runs through all of them. For example, Cauvery, el al. (2005), categorized research broadly into four types to include: Exploratory study, Descriptive study, Diagnostic study and Experimental study. Experimental research, widely considered to be the best way of determining cause and effect in scientific research are often touted as the most rigorous of all research design.
Against this background, the purpose of this paper is to tackle the following:
i. To examine the concept and characteristics of Experimental research;
ii. More importantly, to examine the rationale, types and limitations of Experimental research.

CONCEPTUAL CLARIFICATION
A lot of definitions abound on Experimental research by different scholars. But, one good way to approach a definition of the concept is to break it down into its component parts i.e. “Experimental” and “Research”.
According to Oxford Advanced Learner’s Dictionary, the word experimental or experiment refers to a scientific test that is carried out in order to study what happens and to gain new knowledge. In this regard, Cauvery, et al (2005:52), conceptualized experiment as “the blue-print of the procedures that enable the researcher to test hypotheses and reaching valid conclusion about relationship between independent and dependent variable”. Agbonifoh and Yomere (1999:213), on the other hand asserts that, “an experiment is a research design in which the researcher manipulates or varies an independent variable in a controlled setting in order to observe or measure the impact of the variation or manipulation on another variable”. The foregoing definition therefore suggest that experiments are indeed the only means for ascertaining whether there is a cause and effect relationship between two variables. In other words, experiments are widely considered to be the best way of determining cause and effect in scientific research (Burnham, et al, 2004:46).
On the other hand, the term research derived from a prefix “Re” meaning to repeat, and a Suffix “Search” meaning Seeking. This is why research often times means to search again, to find out more or to take another closer look. Succinctly put, the Webster’s New Encyclopedic Dictionary views research as “a careful or diligent search, studious inquiry or examination, investigation or experimentation aimed at the discovery and interpretation of facts, revision of theories or laws in the light of new facts”.
Drawing from the above expose, Experimental research can therefore be defined as a kind of study or investigation which is designed or structured in such a way as to enable us observe and measure the impact of one variable on another (Agbonifoh and Yomere, 1999:13). A good example of experimental design or research in history could be found in the series of experiments conducted by Elton Mayo and his associates at the General Electric Plant in the United States between 1927 and 1933. At the end of the Hawthorne experiment, it was discovered that contrary to the popular belief that money is the only and most important incentive that motivate workers to work, there are other factors such as recognition and affiliation that determines workers behaviour to work.
At this juncture, it is therefore pertinent to mention the crucial features or characteristics used to describe the nature of experimental research.



CHARACTERISTICS OF EXPERIMENTAL RESEARCH
In considering the nature of experimental research, there are crucial distinguishing features that are paramount to all researchers when conducting their experiments. These features are as follows:
1. CONTROL- In control, an attempt is made to hold all other variable except the dependent variable constant. All possible threats to internal validity are eliminated, reduced or isolated.
2. MANIPULATION- This is a process whereby the researcher varies the level of the independent variable or varies the administration of a treatment on the experimental subject in order to enable him observe the impact of the variation. (Agbonifoh and Yomere, 1999:216).
3. COMPARISON- Here, the thrust of experimental research involves ascertaining whether a significant change has occurred in the dependent variable of interest as a result of some manipulation. Evaluating a change involves determining the amount of change that has occurred (Agbonifoh and Yomere, 1999:219).
Aside from the aforementioned fundamental features of experimental research, others include:
4. RANDOMIZATION: This involves Random selection and Random assignment. On the one hand, Random Selection means that every member of a population has an equal chance of being selected to be a member of the sample. Random Assignment, on the other hand implies that every individual which participate in the experiment has an equal chance of being assigned to either the experimental or control groups (Fraenkel and Wallen 2006:5).
5. ARTIFICIALITY: Experiments does not deal with the phenomenon as they occur in the ordinary course of nature. In nature, events are complex they take place under too many conditions and one does not know which of them is relevant. A scientist problem is to isolate the irrelevant conditions himself. This involves producing a phenomenon artificially (Cauvery, et. al. 2005:66)
6. PROBABILISTIC EQUIVALENCE: This means that we know the odds that the groups won’t be equal. When we deal with human being it is impossible to ever say that any two individuals or groups are equal or equivalent (Trochim, 2006:22).
7. TESTABILITY- The result of experiments on the assertion made can be tested and the same conclusions will be reached using the same methods under the same condition. Thus, the conclusion reached can be replicated (Osemwota, et al, 1996:33). Simply put, the hypothesis on cause-effect relationship is tested and confirmed.
8. OBSERVATION: By observation we mean, an intentional or explicit examination of a situation or event in order to ascertain facts or particulars about it. Here, the effect of the manipulation of the Independent variable on the Dependent variable is observed (Leedey, 1997: 232).
Given the above, the question that now arises is what is underlying rationale for conducting experimental research?

THE RATIONALE FOR EXPERIMENTAL RESEARCH
The focus of this section is to discuss the purposes of experimental research with a view to re-examining each of them critically within the context of social science research. Undoubtedly, the objectives of researches in the physical and natural sciences are very clear compared to the social sciences where the human beings could place certain limitations on their research. Despite these
limitations, there are some basic needs that experimental research meets in the social science. Among them are:
Firstly, experimental research essentially helps the researcher to test hypothesis about cause-effect relationship under a “pure” or uncontaminated conditions. For example, in a set of assumption or proposition that, the female gender generally does not do well in Mathematics or in the Science subjects compared with the male. The social scientist can handle the task by conducting an experiment to confirm the validity of the above proposition or assertion.
Secondly, experimental research is a conscious systematic process aimed at providing answer to a definite question. This is perhaps, the major reason for any research investigation. We must be reminded that no researcher embarked on a research for its sake, but for the desire to know more about whatever phenomenon he is interested in; or for the purpose of answering a contentious question. For instance, the Hawthorne experiment conducted by Elton Mayo and his associates earlier mentioned in this paper could be used to explain the logic behind the rationale.
Thirdly, experimental research helps in the generation of theories for explaining, describing and predicting other phenomena. In addition, it helps to isolate the extraneous variable(s) that may influence experiment through the control and manipulation of the experimental research process.

STEPS OF THE EXPERIMENTATION METHOD
The steps of the experimental methods are essentially those of the scientific method. For the sake of clarification, they may be outlined as follows:
1. Selecting and Defining the Problem-The problems amenable to experimentation generally should be converted into hypothesis. The hypothesis can be verified or refuted by the experimental data.
2. Review the Related Literature- This serves two purposes: (1) to set the theoretical base for research and (2) to set the current research into perspective to show “the state of art”.
3. Drawing up the Experimental Design- This section should place primary emphasis on the question of control, randomization, and replication and should include a clarification of such basic aspects of the design as the place and duration of the experiment.
4. Defining the Population- A clear definition of the population to which the experiment applies is necessary to ascertain to whom and whom the conclusion applies.
5. Conducting the Study- It is important to adhere to plans as they relate to factors such as control, randomization and replication.
6. Assessing the Outcome- Assessment of outcome must be done with a careful consideration of the criterion.
7. Analyzing and Interpreting the Result- The researcher is concerned with the operation of the factors under investigation. He must be especially sensitive to the possibility that the results of his study arose through the operation of uncontrolled factors.
8. Drawing up the Conclusions- The conclusions of the study must be based on the findings of the study.
9. Reporting the Result- The study must be reported in sufficient detail so that the reader can make an intelligent judgment as to its validity (Osuala, 2005).

TYPE OF EXPERIMENTAL DESIGNS
The researcher can design his experiments to suit the unique situation of his/her study. Fortunately, this is perfectly in order. However, an Experimental researcher can be evaluated on the basis of the extent to which it possesses or lacks the key features of experiments, that is, the extent to which it possesses or lacks the key features of experiment. (Agbonifoh and Yomere, 1999:219). On the basis of these key features, an experiment can be categorized as either being a True experiment, Quasi-experiment or Weak experiment. We can broadly classify into a simple threefold classification by asking some key questions. Firstly, does the design use Random assignment to group? If random assignment is used, we call the design Randomized experiment or True experiment. If random assignment is not used, then we have to ask a second question. Does the design use either multiple groups of measurement? If the answer is yes we would label it a Quasi-experimental design, if no, we would call it a non-experimental or Weak-experimental design. The classification is especially used with respect to internal validity (Trochim, 2006:78)
For the purpose of this paper we shall look at these three broad types of design and the different design under each.

1. WEAK EXPERIMENTAL DESIGNS:
The following designs are considered weak since they do not have built-in control for threat to internal validity. By internal validity, we mean the extent to which observed differences on the dependent variable result directly from manipulation of the independent variable. Among the factors that act as threat to internal validity of experimental study includes: history, maturation, testing,
instrumentation, statistical regression, differential selection of subjects, morality, etc.
a) The One-Short case study: A single group is exposed to a treatment and its effects are assessed. For example, a study of the effect of the introduction of new text book on the interest of students in the course.

(Fig. 1)
X-------------------------------------------------------O
New text book Attitude scale to measure
Interest Dependent variable)

In this example, an Attitude scale questionnaire to measure interest of the students is administered after the introduction of the new text book. There is no random assignment to groups and it does not use multiple groups of measurement. This experiment is weak in internal validity and therefore a weak experiment.
b) The One-Group-Pretest-Posttest Design: Single group is measured both before and after a treatment exposure (Fraenkel and Wallen, 2006:8). Using the same example in (a) above, there is a little modification which is the introduction of pretest, a measurement of the interest of the student before the treatment, then the introduction of the New textbook.




(Fig. 2)
O--------------------------------X----------------------------O
Pretest Treatment Posttest
Attitude Scale to New Textbook Attitude Scale to Measure interest measure interest
(Dependent variable) (Dependent variable)

Though this is an improvement of the One-Short case study, but randomization and multiple group measurement is still absent which make it weak in internal validity and therefore a weak experiment.

2. TRUE EXPERIMENTAL DESIGNS:
The essential ingredient of a true experiment is the random assignment of subject to treatment groups. There is also the multiple group measurement and comparison. These are powerful tools for controlling threats to internal validity (Fraenkel and Wallen, 2006:12). The following designs are considered strong and true experimental designs based on the strength of their internal validity.

a) The Randomized Posttest-Only Control Group Design: There are two groups randomly selected and assigned but the treatment is applied to one while the other group does not receive any treatment. The obvious logic of this design is that the difference between the two observations can be attributed to the impact of the independent variable. In other words, the difference is due to the treatment. This is founded on the probabilistic equivalence assumption. For example, let us experiment to ascertain whether the rumour of impending or imminent scarcity of a Fast Moving Consumer Goods (FMCG) has an impact on its demand by retailers.
We have used Indomie Noodle as the FMCG for our experiment. We used the Indomie Noodle Model Shop on Benin-Auchi road. A population of 20 randomly selected retailers of Indomie Noodles was taken and subsequently assigned randomly into 2 groups of 10 each. We observed the retailers’ weekly demand over a 2 week period. The first week was observed without interference and we took the individual retailers’ total order for the week after which we applied our treatment in the form of a ‘rumour of an impending scarcity’ of the product. This treatment was applied to the first group of 10 retailers, say ‘Group A’ while the other group, say ‘Group B’ did not get the treatment as they did not hear the rumour of the imminent scarcity. The difference in the measurement or in the retailers’ weekly demand is regarded as the effect of the treatment or rumour of an impending scarcity of the Indomie Noodles product.

(Fig. 3)
R---------------------------------X1--------------------------------O
Random assignment Treatment Posttest
of 10 Noodles retailers ‘Rumour of scarcity’ Total goods ordered
Experimental group (Dependent variation)
20 Noodles
Retailers
Randomly
Selected
R--------------------------------X2----------------------------------O
Random assignment No Treatment Posttest
of 10 Noodles retailers ‘No rumour of scarcity’ Total goods ordered Control group (Dependent variation)

This experiment has a multiple groups of measurement and there is Random assignment into groups. This is a true experiment as the internal validity is strong.

(b). The Randomized Pretest-Posttest Control Group Design: Pretest is included in this design. The introduction of pretest is the major difference between this design and the Randomized posttest only control group design. Using the same example as in (2a) above, the difference is the introduction of the questionnaire prior to the treatment and again after the treatment. This experiment is also a true experiment.

(Fig. 4)
R---------------------------O----------------X1---------------------------O
Random assignment Pretest Treatment Posttest
10 Noodles retls Total gds ordered ‘Rumour of scarcity’ Total gds ordered
Experimental group (Dependent variable) (Dependent variable)

20 Noodles
Retailers
Randomly
Selected

R------------------------O-------------------X2---------------------------O
Random Assignment Pretest No treatment Posttest
10 Noodle retls Total gds ordered ‘No rumour of scarcity’ Total gds ordered Control group (Dependent variable) (Dependent variable)

(c). The Randomized Solomon Four-Group Design: The design involves studying four groups made up of two experimental groups and two control groups. One experimental group undergoes a pretest while the second group does not. Similarly, one control group undergoes a pretest while the other does not. There is a post-test for each of the four groups (Agbonifoh and Yomere, 1999:224). It is design to deal with a potential testing threat. This design is meant to check the effect of pretest on posttest which is also a possible testing threat. Within each treatment condition, we have a group that is pre-tested and one that is not. By explicitly including testing as a factor in the design, we are able to assess experimentally whether a testing threat is operating (Trochim, 2006:88)

(Fig. 5)
R---------------------------O----------------X1---------------------------O
Random assignment Pretest Treatment Posttest
10 Noodles retailers Total goods ordered ‘Rumour of scarcity Total goods ordered
Experimental group (Dependent variable) (Dependent variable)

20 Noodles
Retailers
Randomly
Selected

R------------------------O-------------------X2---------------------------O
Random Assignment Pretest No treatment Posttest
10 retailers Total gds ordered ‘No rumour of scarcity’ Total gds ordered Control group (Dependent variable) (Dependent variable)


R---------------------------------X1--------------------------------O
Random assignment Treatment Posttest
of 10 Noodles retailers ‘Rumour of scarcity’ Total goods ordered
Experimental group (Dependent variation)
20 Noodles
Retailers
Randomly
Selected
R--------------------------------X2----------------------------------O
Random assignment No Treatment Posttest
of 10 Noodles retailers ‘No rumour of scarcity’ Total goods ordered
Control group (Dependent variation)


3. QUASI-EXPERIMENTAL DESIGNS:
Quasi-Experimental Designs do not include the use of random assignments but use other techniques to control threats to internal validity (Fraenkel and Wallen 2006:19). An example is the Non-randomized Control Group Pretest-Posttest Design.
Non-randomized Control Group Pretest-Posttest Design: The design is similar to the Randomized Pretest-Posttest Control Group Design only that no random assignment occurs. It is used to investigate a situation in which random selection and assignment are not possible. It is one of the strongest and most widely used quasi-experimental designs (Leedy, 1997:7).

LIMITATIONS OF EXPERIMENTAL RESEARCH
Despite the utility of experimental research, the study is not without some shortcomings or limitations. Tonwe (1996) summarized as follows:

I Problem of Random sampling: There are practical difficulties in random sampling, losses of cases and shrinkage in numbers in each group during the period of observation.
II Problem of Application of Probability Test: There are limitations connected with application of probability tests as means of estimating the significance of the results.
III Artificiality: The result of most laboratory research may be artificial in character and may only help in finding out some explanations to given situations. In contrast, social situation are real and pragmatic and have very limited scope for artificial research.


IV Possibility of Wrong Interpretation Due to Artificiality: Since laboratory situations are artificial, the possibility of wrongly interpreting the experimental findings exists, and poses a threat to the internal validity of such investigation (Osemwota, et al, 1996:32-33).

Experimental research in the social sciences generally is pretty difficult as it has to apply human behaviour which is largely unpredictable. Human beings can change their behaviour from time to time depending on the situation and especially when they are aware that they are being studied.

IMPORTANCE OF EXPERIMENTAL RESEARCH
I. The results of Experiments are subjected to testability.
II. The assertion made can be tested and the same conclusion will be reached using the same methods and under the same condition
III. It has scientific character and it is not based on personal limited experiences and observations (Osemwota, et al, 1996:33).

CONCLUSION
Experimental Research is generally regarded as the gold-standard on the world of research. This is basically true considering it utility in the measurement of cause-effect relationship. This paper has attempted to review experimental design and its broad classification down to specific designs and their utility.
It is worthy of mention that the applicability of the experimental research in the social sciences is pretty difficult as the study of human behaviour can hardly be subjected to a kind of laboratory research.


REFERENCE

1) Agbonifoh B.A. and Yomere G.O. (1999) Research Methodology in the Management and Social Sciences Uniben Press University of Benin, Benin City, Nigeria.

2) Burnham P. et. al. (2004) Research Methods in Politics Palgrave Macmillian.

3) Cauvery R. et. al. (2005) Research Methodology (for students of Social sciences. Chand and company Limited Ram Nagar.

4) Ehiamator Egbe T. and Nnandiani M. (2002) Practical Guide to Research in Education and social sciences NERA Publication Benin City.

5) Fraenkel J. K, and Wallen N.E. (2006) How to Design and Evaluate Research in Education McGraw-Hill Higher Education. Internet WWW page at URL: http://www.highered.mcgraw-hill.com/sites/dl/.../fraenkel4_ppt_ch13ppt.

6) Leady P. D. (1997). Practical Research Planning and Design (6th ed). Upper Saddler River NJ: Prentice-Hall, Inc.

7) Osewmotas O. et al. (eds) (1996) Research and Statistical Methods in Social Sciences, Humanities and Education. Amfitop Book Company, Lagos.

8) Osuala E. C. (2005) Introduction to Research Methodology. Africana-First Publishers PLC (Africana Academic Books), Anambra.

9) Trochim William M. (2006) The Research Methods Knowledge Base 2nd Edition. Internet WWW page at URL: http://www.socialresearchmethods.net/kb/ (Version current as of October 20, 2006.