RESEARCH DESIGN

Meaning of research design



A research design is the arrangement of conditions for collection and analysis of data in a manner that aims to combine relevance to the research purpose with economy in procedure. A Research design is a road map(blue print) or the detailed plan for how a research study is to be conducted. A Research design is the conceptual structure(Boundary line) within which research is to be conducted.

Meaning of research design



a) b) c) d)

A research Design is a complete scheme or programme of the research. It includes an outline of what the investigator will do from research problem formulation to final analysis of data. Thus, we can say research design includes the following: Clear Statement of research problem. Technique to be used for collecting data. The target population to be studied. Methods to be used in processing and analysis of data.

Characteristics of Good Research Design



OBJECTIVITY

The objectivity of findings is related to the method of collection of data. In a research the researcher should generally use close ended questionnaire instead of open ended questionnaire. With the help of such type of questionnaire the researcher can maintain the objectivity of findings which means every researcher will arrive at the same conclusion using the same data collection method.

CONTINUED.
RELIABILITY
 

IT means Consistency in Response. If a respondent gives out a response to a particular question. He is expected to give the same response whenever he is asked subsequently. If the respondent keeps on changing his response to the same question then we can say there is no reliability.

CONTINUED.
VALADITY It defines the strength of final results and whether they can be regarded as accurately describing the real world.

CONTINUED..
Generalisation
It means application of the research findings and conclusions to a larger group. When a researcher conducts a research by collecting data and analyzing it using appropriate statistical tools and reaches to certain conclusions, if same conclusions can be applied to a larger group then generalisation is said to be prevailing.

Important concepts relating to research design



Dependent and independent variables:

If one variable depends upon or is a consequence of the other variable, it is termed as a dependent variable, and the variable that is antecedent to the dependent variable is termed as an independent variable. For example, if we say height depends upon age, then height is a dependent variable and age is an independent variable.

Extraneous variables
Those variables that are not related to the purpose of the study but may effect the dependent variables are termed as the extraneous variables For e.g.:-Suppose if we study the relationship e.g.:between a crops productivity and the type of fertilizer used. Here, weather will be an extraneous variable since it is not related to the purpose of the study but it will also affect the crops productivity.


Confounded relationship


When the dependent variable is not free from the influence of extraneous variable .the relationship between the dependent and independent v variable is said to be confounded by an extraneous variable.

Research hypothesis
The research hypothesis is a predicative statement that relates an independent variable to dependent variable.  A Hypothesis is a testable statement of a potential relationship between two variables. Note:Note:- Predictive statements which are not to be verified or the relationships that are assumed but not to be tested, are not termed as research hypothesis.


Experimental and nonnonexperimental hypothesis testing When the purpose of research is to test a research hypothesis, it is termed as hypothesis testing research . It can be experimental or nonnonexperimental.

Experimental and NonNonExperimental hypothesis testing research  Research in which independent variable is manipulated is termed as experimental hypothesis testing research.  On the other hand, Research in which independent variable is not manipulated is termed as nonnon-experimental hypothesis testing research.

Continued


For Example, Suppose in order to test the impact of a training program on students performance level, a researcher randomly selects 50 students and divide them into 2 groups of 25 each. Then one group is exposed to usual training programme while the other is exposed to special training programme. Here Training is an independent variable which is manipulated. but if both the groups are exposed to same type of training then independent variable training is not manipulated.

Experimental and control groups

In an Experimental hypothesis testing research When a group is exposed to usual conditions, it is termed as a control group.  But when the group is exposed to some special condition, it is termed as Experimental group  In the above example the group which is exposed to special training programme is an experimental group and the other group is a control group.


Treatments


The different conditions under which Experiment and control groups are put up are usually referred to as treatments. For example, if we want to test the comparative impact of three different varieties of fertilizers on the productivity of a crop then the three fertilizers will be treated as three treatments.

Experiment
The process of examining the truth of a statistical hypothesis, relating to some research problem, is known as an Experiment. E.g.:E.g.:we can conduct an Experiment to examine the usefulness of a certain newly developed drug.


Experimental Unit


A Person or a group of persons or a plot of land etc. on which different treatments are applied are termed as experimental units.

Different research design

In case of Exploratory research study in case of descriptive and diagnostic research In case of hypothesis -testing research studies

Research design in case of exploratory research design



Exploratory research studies are also known as formulative research studies. Since the main purpose of an exploratory research is to formulate a research problem for more detailed investigation later or to develop hypothesis, a flexible research design is appropriate for such research. Generally ,two methods are used for collecting data.

1.The survey of concerning literature or Literature survey



This is most simple and fruitful method of formulating the research problem . This includes reviewing the secondary data available from different sources such as research reports published in trade journals, company records, government reports, books, trade magazines etc.

2. The experience survey



The experience survey means the survey of people who have practical experience with the problem being studied. The object of such survey is to obtain new ideas relating to the research problem. After selecting appropriate persons the researcher can conduct either individual or group interviews. The researcher must prepare an interview schedule for the systematic questioning of the informants. The respondents should be given maximum freedom to express their views.

In case of hypothesis-testing hypothesisresearch studies



In such studies the main purpose is to test the hypothesis of relationship between two or more variables. In order to prove or disprove hypothesis, experiments are conducted i.e. an artificial situation is created in which one variable(independent variable) is manipulated and the effect of it on other variable(dependent variable) is recorded and analyzed. Thus, when we talk of research design in case of such studies, we mean experimental design.

Basic Principles of Experimental Designs

Principle of Replication(Repeatation) Replication(Repeatation)
 

According to this principle, the experiment should be repeated more than once. Thus, each treatment is applied on many experimental units instead of one. By doing so the statistical accuracy of the results obtained from experiments is increased. For example, suppose a pain relieving drug (treatment)is given 4 patients(Exprimental Units) in patients(Exprimental order to test its effectiveness. then we say the treatment(Drug) is replicated 4 times or repeated 4 times.

CONTINUED
Principle of Randomization


According to this principle either the study population or the experimental treatments or both are randomly assigned. Random Assignment in case of study population means that any individual unit has an equal chance of becoming a part of an experimental group or control group. Random Assignment in case of treatments means that a treatment has an equal chance of being assigned to any of the two population groups.

CONTINUED
Principle of Local Control


According to this principle, the experimental units are divided into homogeneous groups or blocks.(the process of dividing population into blocks is known as blocking.) The blocks are so formed that they are within homogeneous and between heterogeneous. Each Block is divided into a no. of parts equal to the number of treatments. Then the treatments are randomly assigned to these parts.

CONTINUED..
For example, we divide the students into 3 groups consisting of 3 students each and we have 3 treatments(3 Coaching Institutes preparing for MAT TIME,Career Launcher,KOTA) Then we will randomly Launcher,KOTA) assign these3 treatments to 3 students included in each group.
BLOCK I BLOCK II BLOCK III A C B B A A C B C

TYPES OF EXPERIMENTAL DESIGNS

Experimental Designs can be divided into two categories:categories:Informal Experimental DesignsDesignsFollowing are the types of informal experimental designs: a) BeforeBefore-and after without control designdesign In such a design a single test group or area is selected and the dependant variable is measured before the introduction of the treatment.

Continued


The treatment is then introduced and the dependant variable is measured again after the treatment has been introduced. The effect of the treatment would be equal to the level of the phenomena after the treatment minus the level of phenomenon before the treatment.

CONTINUED.
AfterAfter-Only with Control Design  In this design two group or areas (test area and control area) are selected and the treatment is introduced in the test area only.  The dependant variable is then measured in both the areas at the same time.  Treatment impact is assessed by subtracting the value of the dependant variable in the control area from the value in the test area

Continued
Before-andBefore-and-after with control design


In this design two areas are selected and the dependent variable is measured in both the areas for identical time period before the treatment . The treatment is then introduced into the test area only, and the dependent variable is measured in both areas for an identical time period after the introduction of the treatment. The treatment effect is determined by subtracting change in the dependent variable in the control area from the change in the dependent variable in the test area.

Formal Experimental Designs

Completely Randomized Design
 It is based on only two principles viz., the principle
of replication and the principle of randomization of experimental designs. The essential characteristic of this design is that subject are randomly assigned to experimental treatment. For instance, if we have 10 subject and we to test 5under treatment A and 5 under treatment B, the randomized process gives the every possible group of 5 subjects selected from the group of 10 an equal opportunity of being assigned to treatment A and treatment B.

Continued
There are two types of Completely Randomized Design: a) Two-Group simple randomized design Two In a two-group simple randomized design, twofirst of all the population is defined and then from the population a sample is selected randomly.  Further requirement of this design is that items, after being selected randomly from the population, be randomly assigned to the experimental and control groups.

Continued.


Then, the two groups (experimental and control groups) of such a design are given different treatment of the independent variable.
Suppose that the researcher wants to compare two groups of student who have been randomly selected and randomly assigned. Two different treatment viz., the usual training and the specialized training are being given to the two groups. The researchers hypothesis greater gain for the group who receives specialized training. To determine this, he tests each group before and after the training, and compares the amount of gain for the two groups to accept or reject his hypothesis.

Random Replication Design



There are two populations in the replication design. The sample is taken randomly from the population available for study and is randomly assigned to, say, 8 groups. Similarly, sample is taken randomly from the population available to conduct experiments (because of eight groups eight such individuals be selected) and the eight individuals so selected should be randomly assigned to the two treatments of independent variable..

Continued..
Randomized Block Design


In the R.B design the principle of local control can be applied along with the other two principles of experimental designs. In the R.B. design, subjects are first divided into groups, known as blocks, such that within each group the subjects are relatively homogeneous and between heterogeneous. The number of subjects in a given block would be equal to the number of treatments and one subject in each block would be randomly assigned to each treatment independent of other blocks.

For Example, Suppose there are 4 treatments T1,T2,T3,T4 and each is replicated three times. Then we will divide experimental units into 3 blocks, each of size 4.
BLOCK I T1 BLOCK II T3 BLOCK III T4 T2 T1 T2 T4 T4 T3 T3 T2 T1

Continued..
Latin Square Design


It is an improvement over the randomized block design. In this design the number of treatments equal the number of replication of the treatments. If k stands for the number of treatments as well as the number of replications of each treatment ,then total number of experimental units required for this design will be k*k=k2.

Continued.


These k2 are arranged in k rows and k columns. Then k treatments are allotted to these k2 units in such a way that each treatment occurs only once in each row and in each column. For ex. Latin square of 4*4 will be:(4 blocks,4 treatments,4 replications)
B A D C C D B A D C A B

A B C D

Continued..
Factorial Designs


Factorial designs are used in experiments where the effects of varying more than one factor are to be determined. Factorial designs can be of two types: (i) (i simple factorial designs and (ii) complex factorial designs.

Continued


Simple Simple factorial designs: In case of designs: simple factorial designs, we consider the effects of varying two factors on the dependent variable, but when an experiment is done with more than two factors, we use complex factorial designs. Simple factorial design is also termed as a two-factortwo-factor-factorial design, whereas complex factorial design is known as multimultifactor-factorial design. Simple factorial factordesign may either be a 2 x 2 simple factorial design, or it may be, say 3 x 4 or 5 x 3 or the like type of simple factorial design.

Continued
For example, suppose a company wants to determine the proper concentration of sugar and flavor in a soft drink. For this purpose we need to test various levels of sugar content combined with levels of flavor. Suppose there are 4 different levels of sugar content and 4 diiff. Levels of flavor then then will be 16 different combinations that can be made from these variations.

Continued.
SUGAR CONTENT FLAVOR INTENSITY 1 2 3 4

1 2 3 4

a e i m

b f j n

c g k o

d h l p

Measurement in Research


In our daily life we measure height, weight, depth etc. of a physical object. we also measure when we judge how well we like a movie or how much we like the personality of a person. Thus, we measure physical objects as well as concepts(qualitative aspects).we can measure the properties like height, weight etc directly by using some measuring device but it is not easy to measure properties like taste, honesty, attitude of a customer towards a product etc.

Continued..
Thus, we can say measurement means process of assigning numbers to attributes of physical objects or concepts(Qualitative attributes of human beings).


Types of Measurement Scales

For the purpose of measuring concepts we devise a scale.  There are four different types of scale which are as follows:follows: Nominal Scale-In this scale ,numbers are Scaleonly used to differentiate one person from another or to identify persons. For ex. Roll nos are assigned to students in order to identify them, Numbers assigned to basketball players of a team.


CONTINUED..


Secondly, a nominal scale enables classification of persons into different categories on the basis of a common characteristic. For ex. A sample of students can be divided as male or female by assigning 0 for male and 1 for female. Population of a city can be divided on the basis of religion as: Hindu, Muslim, Christian, Sikh etc.

CONTINUED.


In case of nominal scale, we cannot compare the numbers assigned to persons for ex. In case of roll nos we cannot say roll no. 2> roll no.1(2>1).Here, with the help of roll nos we can only count the no. of students in the class. Nominal data is thus also known as counted data. Moreover , we cannot write 3-1=43-1=42,1+3=4,4/2=2.

CONTINUED.
ORDINAL SCALE
An ordinal scale has all the properties of a nominal scale (Property of difference)plus one of its own.(Property of magnitude).  In ordinal scale not only individuals are divided into different groups but they are also arranged in an order either ascending or descending. For ex. On the basis of income persons can be classified into three groups as above average .Average, Below Average.


CONTINUED..


Here, it indicates that people in Above Average group have more income than average group. So we can compare the three groups as Above average>Average>Below Average. For ex. On the basis of % marks secured by students in a class we can rank them from highest % to lowest % such that student with highest % will be assigned I st Rank.

CONTINUED.


But here, it is not necessary that difference between rank1&2 is equal to the difference between rank 5&6.This difference may be more or less.(Property of equal intervals) Thus, we can say Rank1>Rank2 but we cannot say 2-1=6-5. Thus, the 2-1=6use of an ordinal scale implies a statement of greater than or less than without being able to state how much greater or less.

CONTINUED.
Interval Scale
This scale possess all the properties of ordinal scale(Property of Difference, Magnitude)and also a property of its own(Property of equal Intervals). This means that this scale has a starting & terminating point that is divided into equally spaced intervals. But these starting and terminating points are arbitrary(temporary) and not absolute.

CONTINUED.


For ex. Celsius scale is an example of interval scale. In this scale, the starting point is 0 and terminating point is 100. The gap between these two points is divided into 100 equally spaced intervals, known as degrees. The higher the degree, the higher the Temperature. so we can say100>70 or 58<63.we can also say 70-40=100-70. 70-40=100-

CONTINUED..


As the starting point of a Fahrenheit or a Celsius scale is arbitrary i.e. the number 0 of neither scale is indicative of the absence of whatever quantity we re trying to measure. Here, we cannot say that 60 is twice as hot as 30 i.e. 60/30=2. To show the reason we only need to change to the Fahrenheit scale, 60 in Fahrenheit will be 60*1.8+32=140 and 30 will be 30*1.8+32=86.Now, the first figure is not as twice as second figure.

CONTINUED
Ratio Scale


Unlike Interval Scale, ratio scale have an absolute or true zero of measurement .we can have an absolute zero of length and similarly we can have an absolute zero of time. for ex .The zero point on a centimeter scale indicates the complete absence of length or height. But an absolute zero of temperature does not mean the complete absence of temperature it simply means lowest possible temperature which is 0 in case of celsius and 32 in case of fahrenheit. fahrenheit. Because of absolute zero point,it is possible in case of ratio scale to express two quantities in the form of a ratio.

CONTINUED


The measurement of Income, Age, height, weight are examples of this scale. A person who is 40 years of age is twice as old as 20 years old person i.e. 40/20=2.Similarly a person earning 40,000 Rs. Earns three times the person who is earning 10,000Rs.

Sources of Error in Measurement

Respondent-Errors may arise due to incorrect Respondentanswers/response given by respondents. It may happen that a person is unaware of what is asked in a question but still he ticks out one of the options in order to hide his ignorance.  Similarly, at times Respondent may be reluctant to express strong negative feelings.  Moreover, out of fatigue, boredom etc. A respondent may randomly tick one of the answers to all questions purely on the basis of Guesswork.

CONTINUED.
Situation
Situational factors may also affect the response given by respondents. Any condition which places a strain on interview can have serious effects. For instance ,if some other person is also present along with the respondent at the time of interview then he can distort responses by joining in. Moreover, if the respondent feels that anonymity(confidentiality) is not assured, he may be reluctant to express certain feelings.

CONTINUED..
Measurer/Investigator or the person who is getting the questionnaires filledfilled The behavior of the interviewer may encourage or discourage a respondent to give replies.  Moreover errors may also arise because of incorrect coding, wrong tabulation or wrong statistical calculations done by the investigator.

CONTINUED.
Instrument(Questionnaire/Samplin g method used)used)

Errors may also arise because of improper Questionnaire. The use of complex words which cannot be easily understood by respondents, poor Printing, inadequate space for replies etc. are some of the reasons that make a questionnaire improper. Moreover, errors may also arise due to use of incorrect or inappropriate method of sampling ,due to improper sample size.