12 fallacies that should be avoided when constructing arguments
In this post, 12 fallacies when writing arguments for scientific articles are presented with examples. Some of the presented fallacies seem to be very small and commonly used, but those fallacies completely change the meaning of arguments written in a scientific article.
These fallacies really have negative impacts on the article we write. Many times, these fallacies lead to article rejections in scientific journals.
Because not only do the fallacies create ambiguous arguments, but also the fallacies cause the arguments to be scientifically invalid.
The fallacies are explained as follows.
The wrong use of “can”
In scientific articles, from conference papers to journal articles, very often we present our new method that has advantages over older methods that we compare to. However, we often use “can” word to say that our method can be applied into a wider context than what we present in the articles.
When we use “can” in an argument, it means that the argument has been scientifically proven.
For example: “The presented method to detect circles in images can be applied to other applications, for example, square detections in images”.
In this example, when we use can, it means that our method can really be applied for square detections in images and we have to present evidence, for example, by showing the square detection accuracy tested on several images.
However, in this case, we want to say that there is a possibility that the presented circle detection method can be applied for square detection and we do not mean that the method has been tested for square detection.
To solve this fallacy, we should use “could” instead of “can” word. When we use “could”, it means that there is a possibility, even though we do not have evidences at the moment, that the method may be applied for square detection.
Word ambiguity
Very often, we use a word in a sentence that is not definitive, that is the word can have multiple interpretations by different readers. These multiple interpretations are called ambiguity in scientific articles.
For example, “A large amount of power is required to operate the electric machine”.
In this example, the word “large” is not definitive, it can be 1W, 10W or other values. The sentence should be, for example: “15W of power is required to operate the electric machine”.
Another example is that, we very often incorrectly use the “significant” word in explaining experiment results. This mistake looks simple but, in reality, it is not.
For example: “The effect of temperature on the surface hardness is significant.”
In this example, the word “significant” has multiple interpretations that are it can mean large with certain values or it can mean that there is a causality effect of the temperature to the surface hardness.
Most of the time, what we mean is the causality effect. To solve this issue, we should use “statistically significant” instead of only “significant” in the sentence.
Hence, the sentence should be: “The effect of temperature on the surface hardness is statistically significant”.
Because most of the time this argument is the result from statistical analyses, for example, analysis of variance (ANOVA) method.
Invalid logical argument
Scientific arguments should be logically correct. It means that, whenever the sentence is rephrased, the scientific causality content in the sentences should be still valid.
You may see this post for more details about logical arguments.
For a quick refresher, a handy way to test the logical correctness in an argument is by following this format:
“If X then Y = Y therefore X”
For example, “if the programming code is too long, the program will be complex”.
In this sentence, the sentence is still valid for “if X then Y”.
However, when we flip the sentence to be “the program is complex, therefore the programming code is too long”. In this format, the program complexity is not necessary due to the programming code being too long, there are other possibilities, for example, due to the algorithm itself being complex or due to the program requiring interoperability with other software.
To solve this problem, we can change the sentence to be “if the programming code is too long, the program compilation time will be long”.
In this revised sentence, the sentence is still valid in both “if X then Y” and “Y therefore X” formats. Because, there is a direct causality between compilation time and programming code length.
The idea is that we need to make arguments as specific as possible to minimise multiple interpretations by readers.
Biased argument (with lack of evidences)
There will be a limitation for every experiment result or method that we get from our research. This limitation on the results or method that we present should be clearly stated in our arguments.
In many cases, people try to generalise or make too early conclusions or exaggerated conclusions based on only small data or incomprehensive tests.
For example, when an experiment has, let's say, a temperature set from 100-200 degrees Celsius. The results of the experiment when the temperature is set outside those 100-200 range may not be the same, due to, for example, non-linear (second-degree) effects on the phenomenon we observe.
For this situation, we should clearly state the limitations in our argument. For example, we can say “based on the experiment results with a temperature set between 100-200 degrees Celsius, the reaction time becomes two times faster than the one set at zero degrees Celsius”.
In the above argument, we clearly stated the limit of the experiment from where the conclusion is drawn.
Inappropriate use of authority
In scientific articles, all papers will only be judged by the quality of the content and not by who the authors are (whether the authors are PhD students, professors or noble laureates).
So, do not expect that by putting a “big name” as one of the authors of an article, the “big name” will increase the quality or increase the chance of the article being accepted in a journal or conference.
This “quality content first” is an important mindset when writing an article. We should focus on writing a high-quality article with high-quality content. Only by the quality content, our article will have an impact and will be read by other researchers.
The quality content can only be created by high-quality and well-written arguments.
Based on tradition or common belief (instead of factual evidences)
It is important that in a scientific paper, all arguments should be based on factual data instead of based on, for example, some common practices, methods, cost reduction or time saving.
For example, we cannot say “The experiment parameters are set to have two-level each. Because, with only two-level for each parameter, the number of experiment runs can be minimised”. In this sentence, the argument is not valid because the selection of several parameter levels is not based on a scientific fact.
The argument should be written as “The experiment parameters are set to have two levels each. Because the two levels represent the minimum and maximum range of the operating parameters. From these two-level parameters for the experiment, interaction (non-linear) effects between the parameters can be estimated”.
In this sentence, the reason for the level selection is based on a justified theory.
Inappropriate generalisation of conclusion
All arguments written as conclusion in scientific articles should be less or equal to the number of facts on which the conclusion is based. We should never write an argument “more” than factual data we have.
For example, when we have results only from experiments at a specific condition, let's say, the experiment was performed at a pressure level between 10 MPa and 100 MPa. Hence, we cannot generalise the results to be valid for other conditions outside the pressure range set in the experiment. Unless, we have some scientific proves or theories that can justify the generalisation.
The unscientific generalisation of a conclusion from limited results is one of the common reasons for an article to be rejected in a journal submission.
Lack of Causality
When writing arguments from, let's say, two types of data, it is very important that we should also look at the causality relation between the two data and not only the statistical numbers calculated from the two data sets, such as statistical correlation value.
For example, when we have two data sets: the daily humidity in a city within a month and the daily exchange rate within the same month.
If we calculate the statistical Pearson correlation for the two data sets, there is a possibility we will get the correlation value either close or equal to +1 (proportionally highly correlated) or close or equal to -1 (inversely proportional highly correlated).
However, there is no causality fact between the daily humidity and exchange rate. That is, there seems to be that the daily humidity does not affect the daily exchange rate and vice versa.
From this example, we should consider not only statistical numbers but also causality relations between variables.
Did not present the root cause
This aspect is related to the logical aspect of the arguments explained above. When stating a cause in arguments, the stated cause should be specific so that there is no possible cause other than what we have stated.
A root cause is a common cause that directly triggers an effect or observation. For example, we cannot say that the cause of a material to be brittle is because the material hardness is high.
Although, in many cases, the hardness and brittleness occur together (but not always), both of hardness and brittleness are as effect and not as cause. The atomic structure of the material causes the hardness and brittleness level of the material.
Using “prior event” as a cause without evidence
As has been explained previously, in presenting an argument, especially a conclusion, we must write our arguments based on factual evidence instead of writing the argument based on eventual evidence.
For example, when we write “the surface corrosion on the steel material is caused by the liquid acid stored in the same place”
Just because the steel is stored in the same place (eventual evidence) as the liquid acid, it does not mean that the acid causes the corrosion until there is a proof or fact that support the argument (factual evidence).
Limited conclusion
Very often, when we do experiments to study the effect of one factor with regard to a phenomenon, the experiments' results suggest that the factor does not have a statistically significant effect on the phenomenon of interest.
For example, in this sentence “From the experiment results, the heating temperature does not significantly affect the machine efficiency”.
However, there may be a possibility that the factor will affect the phenomenon in other ways.
For example, maybe the heating temperature will affect the machine's longevity due to, for example, the temperature effect on the surface of the machine components.
So, do not generalise a limited conclusion of a factor about a specific phenomenon at hand for all situations.
Individual attack
This aspect is very obvious. Do not ever attack an individual in writing an article.
For example, in a sentence “The previous method reported by Jack et al. is not valid because their experiment is tailored to the requirements set by a funding body”.
It is not allowed to write the above sentence.
To solve this writing issue, we can state the reason for some drawbacks of the method with scientific reasoning.
For example, we can write “The previous method reported by Jack et al. did not consider the pressure parameter in their experiment procedure. Hence, a new setup of the experiment, that considers pressure and temperature parameters, is performed based on the report by David et al.”
Conclusion
In this post, we discuss and explaine 12 mistakes that are very often found in scientific arguments. Some examples are given to clarify and avoid the mistakes.
By avoiding the mistakes, our scientific arguments become strong and easy to read by readers.
Strong scientific and easy-to-read arguments in a scientific article will increase the opportunity of the article to be accepted by journals or the method and results presented in the article will be followed and implemented by other researchers.
We sell all the source files, EXE file, include and LIB files as well as documentation of ellipse fitting by using C/C++, Qt framework, Eigen and OpenCV libraries in this link.
We sell tutorials (containing PDF files, MATLAB scripts and CAD files) about 3D tolerance stack-up analysis based on statistical method (Monte-Carlo/MC Simulation).