Ohh well, that's actually a good question. It depends. Knowing statistics somehow helps you to make, visualize and think on how to investigate questions. we could go through the whole curriculum on descriptive, inferential, modeling, etc, but deep down it really depends on what is your research object. For some people topics like factor analysis can be needless, for others like me, when work with item analysis it would be a key skill. Now if you ask me about GIS statistics, I would be lost. Another point is that at user level of statistical packages you can go through without knowing that much about the "how calculations are done", but again, it helps to know how it is. Still I've seen international researchers that barely know. Some of them just pay for people to do the data analysis. Well, I like to do my own things. That gives you the "feeling" about the data. How participants answered, quality of it and I dare to say even response patterns.
So how much is needed? The basics at a user level at least. After that, as much knowledge as you can just helps. What makes you an international researcher goes more deeper into the questions you pose, if they are relevant, how is the quality of your research design and so on.
And believe me. Knowledge in English language is the key I think. I've been struggling in my papers because it is extremely challenging to wrote in English sometimes.
Oh Thanks you raised an important point that is statistic is necessary for the reader him/her self not only for the researcher to publish.
This is a good point too, I think the role of the statistician begins early even before collecting the data e.g sample size etc if we missed this point we may miss alot
Dear Dr Ishag it is a questions always raised in a discussion when proposal is introduced for acceptance, what is the least study sample taken to make thesis accepted? In dermatology and even in some other specialties study sample could be rare and prevalences data are deficient which make rational of significance and study is valuable i.e to make minimum number in some studies could be difficult.
Dear Adil thanks yes it is a difficult area but the sample size is basic for the power and any study with minimum power will have to contribution to the science
The minimum statistics level required for any researcher is:
1. Generation, tabulation, and computation of data
2. Data reduction technique - like factor analysis
3. Reliability checking technique
4. Other technique for final analysis like ANOVA, regression, discriminant analysis technique, etc.
5. What statistical technique required to understand is depending on how you want to treat your data?
Dear Ishag. What about outsourcing statistics? Now one of the authors in a group takes the responsibility of the overall jargon (sorry..) and informs the others on the summary of findings. Broad knowledge of the statistical methods is mandatory, but the intricate details of advanced statistics is too much. I am not sure how much Einstein or knows of modern statistics. The level of significance is the what is needed by the reader,
Dear Qurashi Ali
Thank you very much for the comments. I think you are in favor to keep the statistician beside you and argue with him in most of the case. Your comment regard the reader him/herself but what about the authors themselves.
Dear Ishag
Thanks for your response. Nowadays the time of the authors is very precious and keeping them busy with the intricates of statistics is not only unfair, but footing upon the others' territory. Many authors in they of pretending that they understand the topic forget the helping statistician and they dilute them in the author list, restrict them to acknowledgement or ignore them altogether. Being fair is to consider statisticians as valuable members (and not redundant) of the research team.
Dear Qurashi Ali
I agree that and I think statistician is to be considered an essential and valuable members and to be in the core of the work from the start and not only during the results flow.
Dear Adam,
I want to answer from engineering point of view. An engineer (industrial, computer or mechanics) should know below topics for a research;
1. The Role of Statistics in Engineering
2. Data Summary and Presentation
3. Random Variables and Probability Distributions
4. Decision Making for a Sample
and also some important points: cental limit theorem, gaussian distributions family, law of large numbers, bayesian approach, measure of central tendency
Now a days, research becomes a collaborate venture. Scientists rarely publish in solo. Most of the time, experienced scientists write editorials or review papers in solo. Normally all experimental papers are multi-authored papers. Sure, you may find a Phd student publishes his work as a single author paper. Then again, many publishes with their mentor. But that is not a requirement.
Coming to the actual question, it is actually good if the researcher knows little bit of stats to understand. I have been well published author, editor, reviewer, etc. I am not really that good in statistics. That said, people bring multiple expertise in the multi-authored papers. Most certainly there will be one or two people, who will be expert in stats. I normally rely on them.
If someone good with stats, it helps them to write very clearly. They can critique a paper nicely in their review. Thus, it has added advantage.
Normally, PhD students are expected to take at least one course in biostats. In addition, many university has statistics department or statistics help available. They could make use of them. However, when you start to publish aggressively, it is a good idea that you collaborate with someone knowledgeable in this area.
Many time, if someone makes mistake with stats, when they submit a papers, the reviewers will pick on their stats, if it found to be wrong.
The bottom line is, someone could still survive with a minimum statistical knowledge. But, it is better they learn as much as they can. It really helps them to be more productive.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics