- Internal Code :
- Subject Code :
- University :
- Subject Name : Statistics
Applications of Statistical Tools - Part 1
Reflective Summary
In recent years, there is an increased scope of the application of statistical tools and methods. This is due to the increase in the quantification of scientific research on one hand, and the use of large data sets on the other hand. This can be understood as the opening of new avenues and progress in science. But, with the increase in the scientific progress, there is also an increase in the concerns related to the conclusions drawn by these research data.
In the paper “Moving to a World Beyond p<0.05”, it specifies the do’s and don’ts while publishing the paper. It highlights that the researchers should not conclude the results based on the threshold of p value. It focuses on the point that things of scientific and practical importance should not be interpreted or concluded based on statistical significance. The p value should not only be used to conclude the result that there exists or not exists an association. This paper also recommends to stop using the word ‘statistical significance’. The term, “statistical significance” has lost its meaning. It was initially used as a statistical used to stipulate that the result warrants further scrutiny. It was never meant to show scientific importance. In this paper, the author also specifies positive and constructive ideas that should be adopted. The paper summarises the recommendation in two sentences, that is, “Accept uncertainty. Be thoughtful, open and modest.” Significant results have made many researchers to escape to their happy place and they avoid dealing with uncertainty. Thus, the researchers and statisticians should accept uncertainty and should try to bring changes. Although, it is not easy and time is required to bring any type of change.
In the other paper, “What Have We (Not) Learnt from Millions of Scientific Papers with P Values?”, the paper shows that how the bias emerges from a multi-layered selection process that lead to reported p values. The paper recommends remedies to overcome these biases. The paper emphasises that the misuse of null hypothesis significance testing (NHST) p value is increasing over time. It has been observed that cherry picking is done while reporting p value in the abstracts by the authors. The p values in the abstracts of the research papers are more significant than the p values in the full text. It is possible that strong selection biases can make everything significant. There are few optimal situations where null hypothesis significance testing and p value is optimal. There are other more suitable inferential methods than can be substituted for NHST and p value to ensure better comprehend the results. Therefore, this paper suggests reproducible research practices and careful layout of the design of study and the hypothesis.
Both paper highlights that good statistical practice should be conducted with various numerical and graphical summaries of data. The interpretation of statistical result should be made with complete quantitative and logical understanding. Scientific reasoning should not be made using a single index like p value.
Applications of Statistical Tools - Part 2
1) The result after using statistical tool, that is, null hypothesis significance testing (NHST) and p value made the supervisor happy. This was because the p value turned out to be large. The null hypothesis was that there was no difference in the male and female proportion while incubating eggs at different levels of humidity. The p value was more than 0.05. Therefore, the null hypothesis was not rejected. The supervisor interpreted this as that there is no impact of humidity on the sex ratio of crocodile egg. But, this is wrong to consider. I will explain the supervisor that it is wrong to interpret the result on the basis of p value. In the research paper, “Moving to a World Beyond p<0.05”, it is mentioned that p value does not show that there does not exist any association. It is incorrect to conclude that the humidity does not affect crocodile egg sex ratio. Also, the number of eggs collected is very less. This can lead to sample collection bias.
2) The supervisor instruction to conduct one way ANOVA and the calculation of the proportion of males for each incubator to check whether it varies with temperature is correct. However, I will suggest him to consider alternative methods to conclude that temperature affects the sex ratio of crocodile egg. This is because the thermostats in the incubator are not accurate. Also, it is not possible to get each incubator at the same temperature.
3) After conducting an appropriate analysis, the results drawn are in the favour of my supervisor. The table shows that temperature has a positive impact on the crocodile egg sex ratio. This impact is statistically significant as the p value is very small (0.0025). On the other hand, the p value of the coefficient of humidity is statistically insignificant because the p value is more than 0.05. Therefore, it can be concluded that temperature have an impact on the crocodile egg sex ratio whereas, humidity have no impact on the crocodile egg sex ratio. However, there should be other inferential methods to conclude this result as shown in the paper, “What Have We (Not) Learnt from Millions of Scientific Papers with P Values?”
Remember, at the center of any academic work, lies clarity and evidence. Should you need further assistance, do look up to our Statistics Assignment Help