Modern statistical software makes it easier than ever to do thorough data screening/cleaning and to test assumptions associated with the analyses researchers perform.
However, few authors (even in top-tier journals) seem to be reporting data cleaning/screening and testing assumptions associated with the statistical analyses being reported. Few popular textbooks seem to focus on these basics.
In the 21st Century, with our complex modern analyses, is data screening and cleaning still relevant? Do outliers or extreme scores matter any more? Does having normally distributed variables improve analyses? Are there new techniques for screening or cleaning data that researchers should be aware of?
Are most analyses robust to violations of most assumptions, to the point that researchers really don't need to pay attention to assumptions any more?
My goal for this Research Topic is examine this issue with fresh eyes and 21st century methods. I believe that we can demonstrate that these things do still matter, even when using "robust" methods or non-parametric techniques, and perhaps identify when they matter MOST or in what way they can most substantially affect the results of an analysis.
I believe we can encourage researchers to change their habits through evidence-based discussions revolving around these issues. It is possible we can even convince editors of important journals to include these aspects in their evaluation /review criteria, as many journals in the social sciences have done with effect size reporting in recent years.
I encourage you to join me in demonstrating WHY paying attention to these mundane aspects of quantitative analysis can be beneficial to researchers.
Modern statistical software makes it easier than ever to do thorough data screening/cleaning and to test assumptions associated with the analyses researchers perform.
However, few authors (even in top-tier journals) seem to be reporting data cleaning/screening and testing assumptions associated with the statistical analyses being reported. Few popular textbooks seem to focus on these basics.
In the 21st Century, with our complex modern analyses, is data screening and cleaning still relevant? Do outliers or extreme scores matter any more? Does having normally distributed variables improve analyses? Are there new techniques for screening or cleaning data that researchers should be aware of?
Are most analyses robust to violations of most assumptions, to the point that researchers really don't need to pay attention to assumptions any more?
My goal for this Research Topic is examine this issue with fresh eyes and 21st century methods. I believe that we can demonstrate that these things do still matter, even when using "robust" methods or non-parametric techniques, and perhaps identify when they matter MOST or in what way they can most substantially affect the results of an analysis.
I believe we can encourage researchers to change their habits through evidence-based discussions revolving around these issues. It is possible we can even convince editors of important journals to include these aspects in their evaluation /review criteria, as many journals in the social sciences have done with effect size reporting in recent years.
I encourage you to join me in demonstrating WHY paying attention to these mundane aspects of quantitative analysis can be beneficial to researchers.