Types of research design
There are several types of research design which can be used depending on the nature of the research. This section gives you details of the following research designs:
1. Action Research Design
2. Case Study Design
3. Causal Design
4. Cohort Design
5. Cross Sectional Design
7. Experimental Design
11. Metal-Analysis Design
12. Observational Design
13. Philosophical Design
14. Sequential Design
Let us study each of these designs in detail.
1. Action Research Design
Action research consists of a family of research methodologies which pursue action and research outcomes at the same time. Action research design tends to be cyclic in nature wherein, initially an understanding of the problem is developed and plans are made for some intervention. This is followed by acting upon the interventions and then the related observations or information is collected. This is followed by critical reflection in which
the researchers reflect upon and critique what has happened. This reflection leads to next stage of planning and the cycle continues.
This research design is thus cyclical in nature and helps in deeper understanding of a given situation starting with conceptualizing the problem followed by several interventions and their evaluations.
Action research designs are used in community situations. They aim at solution driven research outcomes instead of testing theories. Apart from being cyclic, action research is also participative in nature. Its cyclic nature gives more chance to learn from experience provided that there is real reflection on the process and on the outcomes of each cycle. Action research also provides flexibility to allow fuzzy beginnings while progressing towards appropriate endings.
One of the most important features of action research is that the researcher is responsible for both change (action) as well as for research. Hence action research is more time consuming than conventional research. You may also not be able to use the conventional format of documentation in case of action research.
2. Case Study Design
Case study design is a useful tool for in depth analysis of a particular research problem rather than a statistical survey. It helps in narrowing the research problem to a few examples. Unlike surveys, a case study will not provide complete answers to a question but it can help in elaboration and hypothesis formation. This design can also be used to test the applicability of a theory or a model in the real world.
Case study designs help in understanding complex issues by a detailed study of limited number of cases. Examining of contemporary life situations through case studies helps in application of concepts and theories and extension of methodologies. Case studies also help in detailed descriptions of rare cases.
Case studies cannot be used to generalise the findings over wider population. This type of design may lead to researcher’s bias while interpreting the results due to intense exposure of the researcher to the case being studied. A case being studied may not be a representative of the larger problem being studied and hence may not be a good design to
assess the cause and effect relationship.
3. Causal Design
Causal designs are used to study phenomena in which a variation in one phenomenon (the independent variable) leads to a variation in other phenomena (the dependent variable). Causal researches are conducted in order to understand the nature and extent of cause and effect relationships.
One of the examples where causal research design can be used in development studies is when one needs to study the impact of a livelihood programme on the income levels of the people. For a researcher to choose a causal research design there should be an empirical association between the independent variable and the dependent variable and the relation
between the two variables is not spurious. Another important aspect to be kept in mind is that the variation in the independent variable should precede any variation in the dependent variable to establish the causality.
Causal studies help in identifying the reasons behind some processes. They are also useful in impact assessment studies. Replication is possible in causal studies. These studies offer high levels if internal validity due to systematic selection of subjects.
Coincidences in events or spurious relations may be perceived as cause and effect relations. Due to the existence of a wide range of extraneous variables in the social environment, the causality can only be inferred but not proven with high levels of certainty. There exists possibility of existence of causal relationship between two variables but it may be difficult to establish which variable is the cause and which the effect is.
4. Cohort Design
Cohort design is a design in which a group having similar characteristics is closely monitored over time simultaneously with another group. It is generally used in medical research where a population is selected and information is collected on the statistical occurrence of a particular phenomenon (say heart disease) among subjects which have a particular characteristic (diabetics). The entire population is then followed up in time
and the incidence of occurrence of heart disease in diabetic individuals is compared with incidence of heart disease in those who are not diabetic. This design can also be used in applied social science research, specially the researches that measure risk factors.
This research design is specially used in studies where a randomised control study may be unethical and hence researches that measure risk factors usually use cohort design. Cohort studies measure the potential causes before the outcome, thereby avoiding the problem of having to clearly state which the cause is and which is the effect.
While using the cohort design to compare two cohorts, the researcher cannot control for all other factors that might differ between the two groups. Sometimes cohort studies may take long time to complete especially if researcher has to wait for the incidence of interest to occur within the group. Sometimes the time may be long enough for the key variables to change, thereby impacting the validity of the study. Since cohort design lacks
randomization, it has lower validity than those designs in which the sample is chosen randomly.
5. Cross Sectional Design
Cross sectional design lacks the dimension of time. In cross sectional researches, observations are taken for different groups at one point of time. A cross sectional design can only measure differences among the subjects. A cross sectional study can be used to describe the characteristics of a group. However, the process of change cannot be traced in a cross sectional study. It focuses on finding relationship among variables at one point of
Cross sectional studies are based in drawing inferences from the existing differences than measuring change. The selection of groups is purposive and based upon existing differences in the sample and not random. This design uses survey technique to collect data and hence can be done at a relatively lower cost.
Finding subjects for study with similar characteristics can be difficult. The results being static cannot be explained in temporal context. These studies cannot be used to establish cause and affect relationships. There is no follow up to the findings.
6. Descriptive Research Design
Descriptive research designs are the best methods for collecting information that will demonstrate relationships or describe the current status of the phenomenon as it exists. Descriptive studies involve surveys or interviews to collect the necessary information. Descriptive studies typically provide answers to questions related to who, what, when, where and how, however, they cannot provide answers to why.
Descriptive studies observe the subjects in their natural environment as they exist and are generally precursors to qualitative research designs. They are useful tools in more focused studies.
The results from descriptive studies cannot be used to prove or disprove a hypothesis. The results of descriptive studies cannot be replicated as they are based on observational methods.
7. Experimental Design
In the simplest forms of experimental designs, two groups are created which are equivalent to each other. Then one of the groups, also called the treatment group gets a programme and the other group, also called the control group does not get the programme. These two groups are similar in all other aspects. Then the differences are observed in both the groups.
Since both the groups are supposed to be equivalent in all the aspects, the difference arising between the two is considered to be due to the only difference between the two groups which is that one of the groups get the programme.
If people are assigned to the two groups on a random basis from a common pool and if the sample size is large enough to achieve probabilistic equivalence between the two groups, the experimental research designs are capable of achieving strong internal validity. Since the experimental designs allow researchers to control a situation, it also allows researchers to establish a cause and effect relationship.
Since the experimental designs are artificially established, the results may not generalize well to the real world. Also, the artificial setting of the experiments may alter the response of the participants. Experimental designs requiring special equipments may be a costly affair. There may also be ethical considerations to be addressed in experimental research designs.
8. Exploratory Research Design:
Exploratory research designs are used when the researcher has an understanding or observation about something and seeks to learn more about it. It is used in research problems in which there are few or no earlier studies to refer to or rely upon to predict an outcome. Exploratory designs are undertaken with one or more of the following goals:
- To develop familiarity with basic details, settings, and concerns.
- To develop a well grounded picture of the existing situation
- Generation of new ideas and assumptions.
- Development of tentative theories or hypotheses.
- To determine the feasibility of the study in future.
- To lay the groundwork which will lead to future studies.
- Direction for future research and techniques get developed.
It helps in gaining background information on a particular topic. Exploratory research designs are undertaken with one of the two basic purposes to define new topics or a new angle or to clarify the existing concepts. These research designs help in generating a formal hypothesis or in developing a more precise research problem. Exploratory designs are useful in research prioritization or resource allocation.
As the sample size used in exploratory designs are small, they cannot be used to make generalizations about the population. They do not provide a definitive conclusion for a research problem. The outcome of exploratory researchers may not be of much value to the decision makers. As compared to other designs, this design lacks the rigorous standards of data collection and analysis.
9. Historical Research Design
Historical research typically involves studying and interpreting past events to predict the future ones. Historical research typically involves Collection, verification and synthesis of evidences from the past so as to establish or refute a hypothesis. It uses both primary and secondary data sources including documentary evidences such as diaries, reports, archives, official records etc.
Historical researches are unobtrusive and do not affect the result of the study. Historical approach is typically suited for trend analysis. There is no possibility of interaction between the researcher and the subject which could affect the findings of the research.
The fulfilment of the objectives of such researches is based upon the availability and quality of documents related to the research problem. Interpreting historical researches can be very time consuming. Historical researches are weak in terms of the demands of internal validity. Since the availability of entire documents needed to fully address a historical research is rare, such gaps need to be acknowledged in the report.
10. Longitudinal Research Design
In a longitudinal study, the sample is followed over time to make repeated observations. In longitudinal studies, the same group or sample is tracked over time and the changes observed. The researcher then relates them to variables that might have caused the changes. They are useful in studies required to establish causal relationship and help in showing both the magnitude and direction of causal relationships.
Longitudinal studies help in identifying the duration of the particular phenomenon being studied. These studies help in measuring and analysing the pattern of change on different variables over time. They also facilitate predictions to be made on the basis of earlier factors.
Large sample size is needed to be able to accurately explain the causal relationships. It may also be difficult to maintain the integrity of original sample over time. The data collection method may change over time and it may also be difficult to show more than one variable at a time. Longitudinal research is based on the assumption that the present trend will continue which may not hold true all the time.
11. Meta-analysis Design
Meta analysis is a technique designed for combining results from independent studies. It involves evaluation and summarizing of results from a number of individual studies thus increasing the overall sample size and also increasing the ability of researcher to study the effects of interest. This type of study uses synoptic reasoning to develop a new understanding of the research problem. Meta-analysis includes analysing differences in the
results among studies and increasing the precision by giving due weight to the size of different studies included. The validity of meta-analysis depends on the quality of systematic review on which it is based. A well designed meta-analysis depends upon strict adherence to the criteria used for selecting studies and the availability of information in those studies to be able to properly analyse their findings. Larger the heterogeneity in
the results of the individual studies, the more difficult it becomes to establish validity of the synopsis of the results.
Meta-analysis can be used as a strategy to determine gaps in already existing literature. Such studies help in review if researches on a particular topic over a long period of time and published in various sources. They also help in overcoming the problem of small sample in individual studies. They are generally used to generate new hypothesis or suggest problems for future studies.
Large sample may yield reliable but not necessarily valid results. Synthesis of a heterogeneous collection of studies in terms of literature reviewed, methods applied or measurements of findings may be difficult. The whole process can be quite time consuming and small violations in criteria used for analysing the studies may lead to meaningless findings.
12. Observational Design
This type of research design draws a conclusion by comparing subjects against a control group; however, the allocation of treatment is not fully under the control of the researcher. The main difference between experimental and observational research is the lack of manipulation of independent or causal variable in case of observational research design. The researcher simply observes the naturally occurring values of the independent and
dependent variables and then uses techniques to find out if they co-vary.
This design helps ensure that there are no false responses that might be introduced by manipulative research procedures. An observational study allows a useful insight into a phenomenon and avoids the ethical and practical difficulties of setting up a large and cumbersome research project. Observational designs are of two types: direct observations in which subjects know that the researcher is watching them and Unobtrusive observations in which the individuals do not know that they are being observed.
Observational designs are flexible and do not necessarily need to be structured around a hypothesis. The researcher is able to do an in depth investigation about a particular behaviour and hence helps in understanding the interrelation among various dimensions of the study. As the variables being studied are allowed to operate without any intervention in case of observational designs, the external validity is often very high. The results of observational research can be generalised in real life situations.
The internal validity of observational research designs is low. Reliability of the data is low as they are difficult to replicate. The findings may pertain to a unique sample population and cannot be generalised. The study may have an element of the researcher’s bias. Since nothing is manipulated, a cause and effect relationship cannot be established. The presence of a researcher in case of direct observations may lead to skewed data
13. Philosophical design
The philosophical researches challenge the assumptions underlying a particular study. Arguments derived from philosophical theories, models, concepts or traditions form the tools for this type of studies. Such studies can basically be of three types:
- Ontology — the study that describes the nature of reality; for example, what is real and what is not, what is fundamental and what is derivative?
- Epistemology — the study that explores the nature of knowledge; for example, by what means does knowledge and understanding depend upon and how can we be certain of what we know?
- Axiology — the study of values; for example, what values does an individual or group hold and why? How are values related to interest, desire, will, experience, and means-to-end? And, what is the difference between a matter of fact and a matter of value?
This type of research helps the researcher in gaining greater self understanding about the purpose of research. Philosophical studies help to refine concepts and theories that are invoked in relatively unreflective modes of thought and discourse thus offering clarity to the terms, concepts and ideas.
Its application is limited to specific problems and has limited applicability in real life issues. There can be analytical difficulties in moving from philosophy to advocacy and between abstract thought and application to the phenomenal world.
14. Sequential Research Design
A sequential design may be defined as a combination of longitudinal and cross sectional design in which several cohorts are followed at a time. In longitudinal designs, we look at one group over a long period of time and in cross sectional design we look at several groups at a particular period of time. In sequential design, we look at different groups over a period of time.
Sequential design is a useful design for exploratory studies. As the study is repetitive in nature, minor changes or correction warranted in the research methods can be done at the initial stages. The study being conducted serially, the results of one sample are known before next sample is drawn and analysed providing enough opportunities for improvement of sampling methods and analysis.
The sampling method used does not allow choosing a representative sample of the population unless the researcher goes for a very large sample. Also as the sampling technique is not randomised, the results cannot be generalised for the entire population. It is also difficult to interpret variation from one sample to another over time.