What’s Behind Statistics?
It is easier for accountants to agree than scientist, simply because science doesn’t have a one fit answers for a problem. The result depend on what was found to be true at that particular time, based on facts at that particular point in time. But what is it that science communicator has to do to have a better understanding of the research outcomes?
You don’t need to master nitty-gritties of statistics in order to understand and communicate scientific findings, however probing of the findings need to done to separate the truth from trash. Lewis Cope in his article “Understanding and using statistics” is trying to give us the ways which will help in digging for more information behind the statics.
He starts by explaining more on “Probability, Power and large numbers” principle which takes care of results by chance. These results can be avoided simply by having a large sample size. The other thing to be considered here is the likelihood of a certain result to be obtained. He gave an example of cases of ”lung cancer in smokers” , this is an obvious things, we all know that smoking causes lung cancer and it not strange to find a smoker diagnosed with this type of cancer.
Check biasness of the conclusions of the study or explanations of the results. Failure to consider other researchers findings or explanations on similar topic shows unprofessionalism. In doing a research it is either a certain point of view is being verified to be true or not, so it is actually a revolving story from somewhere else, so the new discoveries has to be compared to other findings done in the same study area.
Another principle is that of trusting some studies based on impact or indication the findings will have. As stated by Lewis Cope, epidemiological studies tend to study mites to find the effectiveness or dangers of some medicines, vaccines, or chemicals to human beings, so despite the sample being involved in these kinds of studies is small, the impact is great.
Peer review is another thing that also strengthens statistical findings, in every field of study there are professionals who can either agree or disagree with a research finding, find out if the researcher has considered other explanations from his peers and if there are any disagreements check the reasoning the differences.
Statistical distribution of the results needs to be considered as well, “how many samples are above or below average. This will give a good clue on the general outcome of the research.
Assess if correct sampling technique was used. All the above mentioned “principles” will hold if the sampling technique used was correct. Some researchers can deliberately choose a certain way of sampling just to get the result they want which cannot be a true reflection of realities. As indicated earlier, that the larger the sample the better the result, but if that large sample was not randomly selected, the result will not a true representation .For instance using online survey to assess a general acceptance of something by a nation. Online survey will automatically screen out a population that does not have access to internet. Random sampling is always the best way of achieving a better result.
This article gave an insight of what to look for in research statistics to have a clear idea before communicating to the general public, do you agree?
Cope, L. Understanding and using statistics. In Blum D, Knudson M & Henig RM (Eds), A field guide for science writers (pp 18-25).