What does the "nature of a study" mean with regard to social sciences research?
2 Answers | Add Yours
I assume that you are asking about the term "nature of the study" with regard to something like the writing of a dissertation. If so, the "nature of the study" section of the dissertation is typically the one in which the research design is laid out and discussed.
In this section of the dissertation, you will describe the basic methodology that you used. You will talk about things like who the subjects were and how you collected data on them. You will describe the procedures that you used to collect that data.
In other words, this will be a relatively brief section of a dissertation in which you give the reader a basic understanding of how you did your research. You will not present your findings here--this is just for setting out how the study was done.
The nature of a study in social sciences research may refer to the statistical design of the study. When writing up a report about a study, the report would usually take the format:
Hypothesis and Aims
Methods - Data Collection and Statistical Methods proposed for Analysis
Results and Analysis
In modern science, all findings are usually required by the research community to be backed up by sound statistical evidence. The target audience and peer researchers need to be sure that the results were obtained in a logical and unbiased way and also ideally quote a number (typically a p-value) expressing how likely the results were to arise by pure chance alone. The experiment should be repeatable, and with the same or similar (reasonable) analysis methods give the same results. For good scientific practice, the study design and intended statistical analysis methodology should be laid out even before data collection so that results aren't 'mined' from the data after the study has been conducted. In fact, the study design (the data collection method and statistical methodology to be used) are typically laid out at the grant proposal stage before money is invested in the experiment/research, especially for large-scale experiments like randomised clinical trials.
The nature of the study in terms of its design would be, for example, whether it is:
- an observational study (where the data are not collected under controlled experimental conditions, but from archived databases or published sources)
- a case-control study (cases and controls are matched and the difference in outcomes from one arm of the study to the other observed)
- a controlled experiment (like a clinical trial), or
- a survey
In short, how exactly were the data collected and what are the strong points and limitations of the method?
The type of statistical methodology used will hinge on key assumptions made and ways around limitations of the data. The method may assume a particular model for the data, classically the Normal distribution, or may be a non-parametric analysis, necessarily if the data a qualitative as opposed to quantitative. There may be an assumption about common variance across strata in the data, or other assumed instances of shared parameters that reduce the dimensionality of the model. The parameters of interest must be clearly identified and ways of maximising the statistical power to estimate them should be addressed carefully. The method might be under the classical/frequentist paradigm, or might, as is seen more and more in modern science, be under the Bayesian paradigm. This would usually be implicit given the outline and description of the models and analysis methods used. Lastly, there may unavoidable biases in the data, such as potential missing data. If the missingness is not at random particular care needs to be taken with this in analysis. Also, the quality of the collected data may strongly affect the usefulness of the findings of the study. Even if there are large amounts of data, if it is of poor quality it is useless for making informed statements about the phenomenon of interest.
Always with the statistical method chosen, there is a trade-off to be made regarding simplicity and interpretability of the model versus accuracy of results, and of power to detect true effects versus controlling type I error (finding seemingly interesting effects by pure chance). Time and resources also govern how much data can be collected and the depth and sophistication of the experiment. To warrant spending a lot of money and time on the research, the outcomes need to be of sufficient interest to the scientific community or private companies (and hopefully the general public too) and the researchers demonstrate enough confidence regarding the findings they expect to record. If the right people cannot be convinced that the research deserves time and money it won't go ahead, so conveying the importance of the hypothesis and engendering confidence in the success of the statistical methodology to be used and its worthiness to demonstrate real findings is key. No matter how convinced the research group is of their ideas, they must convey this effectively to others for the research to be noticed and recognised. There are scientific standards, and these need to be observed for findings to have credence.
Join to answer this question
Join a community of thousands of dedicated teachers and students.Join eNotes