Panel Data Analysis: A Survey On Model-Based Clustering Of Time Series
January 31, 2020
Explain And Execute Statistical Design And Analysis Of Two Variable Hypothesis
February 7, 2020Research Design Decisions And Be Competent In The Process Of Reliable Data Collection And Analysis
Research Design Decisions And Be Competent In The Process Of Reliable Data Collection And Analysis
Research Design may be described as the researcher’s scheme of outlining the flow of his project. It is based on research design, that the researcher goes about gathering data to answer his research question. If the idea is to complete a building, then it has to be decided whether it is going to be an apartment, stand-alone house or a shopping complex, who are its occupants? and what are the materials needed? The plan of the project, namely the planning for the materials and the logistics involved follows this. Similarly, in research as well, the researcher chooses his data collection process based on his Research design decision. It enables the researcher to prioritize his work, create better questionnaires and arrive at conclusions with greater clarity.
It would be worthwhile to take a look at an example:
Table 1
Evaluation matrix:
Matching data collection to key evaluation questions
| Examples of key evaluation questions (KEQs) | Programmed participant survey | Key informant interviews | Project records | Observation of programme implementation |
| KEQ 1 What was the quality of implementation? |  |
|
|
|
| KEQ 2 To what extent were the programme objectives met? | |
|
|
|
| KEQ 3 What other impacts did the programme have? | |
|
||
| KEQ 4 How could the programme be improved? | |
|
Source: Peersman,(2014)
In the above diagram, table1 shows the type of questions and the data collection methods that were used for the same. For instance, Key informant interviews and Project records were used for collecting information on the quality of the implementation. Quantitative research design may be sub-divided into experimental, Quasi-experimental, Survey and Correlational, while, Qualitative research may be divided into Ethnography, Case study, Historical and Narrative.
Broadly, RD can be classified into Exploratory and Conclusive. Exploratory research is a research conducted for a problem that has not been studied more clearly, intended to establish priorities, develop operational definitions and improve the final research design.(Shields & Rangarajan, 2013) It does not seek to arrive at a conclusion. Conclusive Research can be classified into descriptive and causal. Descriptive research tries to answer questions such as what and How? While, Causal research tries to establish the cause-effect relationships among the variables of the research.
Table 2
- Major differences
between exploratory and conclusive research design
| Research project components | Exploratory research | Conclusive research |
| Research purpose | General: to generate insights about a situation | Specific: to verify insights and aid in selecting a course of action |
| Data needs | Vague | Clear |
| Data sources | Ill defined | Well defined |
| Data collection form | Open-ended, rough | Usually structured |
| Sample | Relatively small; subjectively selected to maximize generalization of insights | Relatively large; objectively selected to permit generalization of findings |
| Data collection | Flexible; no set procedure | Rigid; well-laid-out procedure |
| Data analysis | Informal; typically non-quantitative | Formal; typically quantitative |
| Inferences/ Recommendations | More tentative than final | More final than tentative |
Source: Pride-Ferrell,(2006)
Data Collection Techniques and How to choose one
Using a mix of both Qualitative and Quantitative methods can be most beneficial. The most widely used data collection techniques are Interviews and Questionnaires. Interviews may be one to one or in groups. The Questionnaire is developed with the research question in mind. But it is very difficult to determine if the participant is lying or not. Hence reliability is a problem here.
Here are a few tips on developing effective survey Questionnaires:
- Ensure that that the length of the survey questionnaire does not run to more than five minutes.
- Avoid complicating the Questionnaire by using questions which may refer to answers of previous questions. For instance, ‘If your answer was yes to Q. No 3 then…’.
- Take care to see that the Questions don’t look biased. ‘You would not refer XYZ Baby oil to your friend. Would you?’
- Ensure that you keep the Demographics in mind and use uncomplicated words.
- Make sure that the questions do not carry conflicting ideas, such as ‘Which is the best and cheapest restaurant in town?’ The best restaurant need not be the cheapest.
Using Data Collection tools such as ‘Device Magic’ which helps you to pre fill form data. ‘Fulcrum’ allows for custom maps with geo location while ‘Fast Field’ enables exporting to word and pdf. ‘Magpi’ has features for interactive data collection. ‘Zapier’ helps automate the Data Collection process.
Data Analysis
Probability and non-probability methods are used in Data Analysis. Probability sampling uses random or semi-random methods to select a sample from among the given population and it uses Statistical generalization with a margin for error as no sample will exactly reflect the population exactly.
Random Sampling uses a simple process where there is equal likelihood of every member from the sample being chosen. Stratified Random Sampling uses a method of segregating the sample into mutually exclusive groups and then selecting simple random samples from a stratum. Example:
Strata1: Gender Strata2: Income Strata3: Occupation
Male 1 lakh Self-employed
Female 1 to 2 lakhs Clerical
2-5lakhs Professional
In the above sample we can choose females with income range of 1 to 2 lakhs using simple random sampling. We are now able to make inferences across these 3 strata. After stratifying the population, simple random sampling is used to generate the complete sample.
Among non-probability sampling methods Purposive sampling is used where particular cases which are information-rich are selected with a view to drawing inferences about the population. Convenience sampling is used only in cases of insufficient data.
Mixing methods can improve credibility of the research findings as each data source possesses its own limitations and advantages and triangulating data from different sources or integrating different collection methods will help answer the research question more accurately.
Some methods of Numerical analysis are given below:
Source: Peersman, (2014)
| Numeric analysis : Analysing numeric data such as cost, frequency or physical characteristics. Options include: |
| Correlation : A statistical technique to determine how strongly two or more variables are related. Cross tabulations: obtaining an indication of the frequency of two variables (e.g., gender and frequency of school attendance) occurring at the same time. Data and text mining: computer-driven automated techniques that run through large amounts of text or data to find new patterns and information. Exploratory techniques: taking a ‘first look’ at a data set by summarizing its main characteristics, often through the use of visual methods. Frequency tables: arranging collected data values in ascending order of magnitude, along with their corresponding frequencies, to ensure a clearer picture of a data set. Measures of central tendency: a summary measure that attempts to describe a whole set of data with a single value that represents the middle or centre of its distribution. Measures of dispersion: a summary measure that describes how values are distributed around the centre. Multivariate descriptive: providing simple summaries of (large amounts of) information (or data) with two or more related variables. Non-parametric inferential: data that are flexible and do not follow a normal distribution. Parametric inferential: carried out on data that follow certain parameters. The data will be normal (i.e., the distribution parallels the bell curve); numbers can be added, subtracted, multiplied and divided; variances are equal when comparing two or more groups; and the sample should be large and randomly selected. |
| Summary statistics: Providing a quick summary of data, which is particularly useful for comparing one project to another, before and afterwards. Time series analysis: observing well defined data items obtained through repeated measurements over time. |
| Textual analysis : Analysing words, either spoken or written, including questionnaire responses, interviews and documents. Options include: |
| Content analysis: Reducing large amounts of unstructured textual content into manageable data relevant to the (evaluation) research questions. Thematic coding: recording or identifying passages of text or images linked by a common theme or idea, allowing the indexation of text into categories. Narratives: construction of coherent narratives of the changes occurring for an individual, a community, a site or a programme or policy. Timelines: a list of key events, ordered chronologically. |
Source: Peersman,(2014)
References
- Peersman, G. (2014). Overview: Data Collection and Analysis Methods in Impact Evaluation: Methodological Briefs-Impact Evaluation No. 10. Retrieved from https://ideas.repec.org/p/ucf/metbri/
- Peersman, G. (2014). Overview: Data Collection and Analysis Methods in Impact Evaluation: Methodological Briefs-Impact Evaluation No. 10. Retrieved from https://ideas.repec.org/p/ucf/metbri/
