The topic of misleading statistics have already reached the tables of academic scientists. And they already starting to academically examining it. One of this academic sources were made by Maiz Mohammed Masi in 2019, and the name of it is "Analysis of Misleading Statistics" from The International Journal.
The paper presents a broad and analytical view on the topic of misleading statistics. It incorporates sections as Introduction, Literature Review, Some relevant terms (Which includes the important terms as Statistics and Data Analysis), Methods of Persuasion, Constraints in analysing the data, The analysis in scientific domain, and Overcoming faulty statistics (Masi, 2019).
Literature Review emphasises the importance of statistics based on the American Statistical Association ASA) and presents sources that were used to explain the material presented in other parts of the paper. Additionally, the quote from A. Wolf's paper were discusses, and it stated that it is better to have a mislead number than no number because "it can stimulate the correction", but this number might become embedded with no further scrutiny. And the less the number is, the less it will be debatable from misinterpretation and misuse. (Wolf, 2007, from Masi, 2019).
Section of Methods of Persuasion includes some of the most applicable methods for making a misleading statistics. They include subparagraphs as false correlation (making the outcomes of event 1 to be applicable to event 2), correlation does not necessary mean causation (to establish a causation, a statistical evidence should not be only one), confirmation bias (is a tendency to search for or interpret information in a way that confirms someone's findings), manipulation of graphs (there are different ways to graphically influence the ones interpretation of statistics), testing many hypotheses (reporting ones that fit in the study or analysis), cherry-picking data (choosing the data samples that gives desired outcomes and results ) (Masi, 2019). All of these methods are used in the misleading statistics, and it is important to understand how they are made.
Section of Constraints in analysing the data presents the so-called Simpson's paradox. It states that the aggregated datasets show a reverse trend than the original one when they were separately presented. For example, of one of our test results indicated that one of the groups is more biased compared with other, however, on a more separate levels it was indicated as reversed. One of the famous examples of it is the case with UC Berkeley was sued for potential gender bias in admissions, which by further investigation was rejected. I recommend reading about this case.
The analysis of scientific domain, however, indicated that 1 in 4 Statisticians admit they were asked to commit scientific fraud by manipulating statistics, however, the number could be higher because some of respondents could be hiding that fact. The main methods were to cherry-picking, dropping of some unsuitable records, and manipulating with survey questionnaires which then lead to misleading or falsified reports (Masi, 2019)
Ad for the overcoming faulty statistics, the authors suggested a various methods to find the misleading statistics. The important factor there would be to search for inconsistency in graphs, and understand the context of it. Additionally, the one important factor is the amount of sample, because it is hard to create a misleading statistics, and bigger samples provide more accurate results (Masi, 2019). Lastly, the statistical significance would be an important factor to consider, because it can dramatically change the outcomes of the test. For example, if you are completing a test based on a statistical significance variable as 5%, it gives you the results that support your claims. However, if you decide to increase that to 10%, the outcomes would not support your claims. That is why you should involve that factor in analysis too.
To conclude all of that, I must say that this article is a broad and interesting for those who want to learn more about the topic of misleading statistics. The number of methods that are used are well-explained and presented, as well as the outlook on the scientific domain and paradoxes. Additionally, the solutions to avoid that is presented. That is why I recommend you to read that article, the link to it is the references. And also, do not forget that ones who are able to process with the possible misleading statistics have a full and clear picture on events happening in the World.
References:
Mazi, M. M. (2019). Analysis of misleading statistics. The International Journal, 2(1). https://maktabahjafariyah.org/paper/volume3_1.pdf
Comments
Post a Comment