Range, standard deviation, correlation between the different parameters and determination of p value is the important statistical analysis parameters. Except this any other parameters are there which can make a research paper much more scientific and easily understandable ? Please share your views.
Using statistical analyses does not make a paper "more scientific". Using statistics inappropriately rather makes a paper less scientific. And applying methods, including but not limited to, statistical analyses, that are not sufficiently well understood to recognise their limitations and to give a clear, correct and useful interpretation of their results, is also more a feature of bad science.
You can start with the p-value. It is embarrassing how many authors are not able to correctly interpret them, despite of the fact that they can be found in almost every paper. Many papers present plain wrong interpretations. That's not science. I think the correct English word for this is bullshit (https://en.wikipedia.org/wiki/Bullshit#In_the_philosophy_of_truth_and_rhetoric).
@Jochen: To some extent previously my thought was rather same as you. But after going through the big journals and observing their high impact carrying papers I am really puzzled. In fact at present it is looking like that if I don't add statistical analysis part it may be possible that my hard work will not get so much impact to the scientific arena. So it is creating really a hidden pain to me as well. Thank you so much for sharing your views.
Statistical tests should be selected based on the type of data as well as on the distribution of the data especially for quantitative data. More simple stat analysis get more better and readable paper.. Combining both experience (by practice or via an expert mentor) and good knowledge about statistics and research methods make your research work more scientific and easily understandable..
Publishing in high-impact journals is not neccesarily indicative for good science. These are different goals: "making good science" and "publish papers in high-impact journals". Rarely, they may coincide, but often they don't. Many (most?) really good papers are not published in the top journals. The "impact" of these journals is not high because the papers are scientifically good, but because the papers in these journals are often cited. And this has mainly other reasons (e.g. the topic is "sexy", there is interest in the topic of a larger community [like for medical topics], and famous influential names in the author list).
If your work includes quantitative methods, you will usually have to present the data in some form, graphically or as summary - and this is already applying statistics. You will further likely use the data to fit some model of interest, and there you will have to provide some measure of uncertainty associated with estimates and possible conclusions that can be made from your data w.r.t. the model. Doing this is a neccesary but not a sufficient condition for a work to be scientific, I'd say. Adding analyses that do not adress a scientific question do not make a work "more scientific", and adding analyses that are not understood and/or wrongly interpreted can easily destroy an otherwise good work.
@Jochen, thank you so much for your excellent answer. Really it is a view that is related with hard reality and ground fact. Really I am thankful to you for sharing your thoughts.
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers