Statistical Learning

TY - CHAP

T1 - Statistical Learning

AU - Leppink, Jimmie

PY - 2020/4/7

Y1 - 2020/4/7

N2 - Where Chaps. 1 and 2 provide the foundation for types of questions and study designs, respectively, this third chapter provides a pragmatic approach to statistical testing and estimation (PASTE). This approach is an extension of PASTE introduced in my recent Springer Texts in Education Series book entitled “Statistical Methods for Experimental Research in Education and Psychology” in two ways: moving beyond experimental research and including small samples as well. Both extensions require specific modifications in the pragmatic approach for larger-sample experimental research as introduced in the aforementioned book, including in criteria used for statistical testing and in terms of causal inference. However, like in the approach presented in my book on experimental research, the approach discussed in this chapter is about uniting traditional and emerging approaches to statistical testing and estimation: Frequentist CIs and equivalence testing, Bayesian posterior intervals and the region of practical equivalence, Likelihood ratio testing, and information criteria that are—through the use of Likelihoods—related to the other methods but provide testing outcomes under slightly different assumptions. Concepts of Big Data, Artificial Intelligence, Machine Learning, Educational Data Mining, and Learning Analytics are introduced in this chapter, with examples that are revisited in later chapters in this book. Finally, PASTE as introduced in this chapter incorporates a decision-making framework for dealing with different types of missing data as well.

AB - Where Chaps. 1 and 2 provide the foundation for types of questions and study designs, respectively, this third chapter provides a pragmatic approach to statistical testing and estimation (PASTE). This approach is an extension of PASTE introduced in my recent Springer Texts in Education Series book entitled “Statistical Methods for Experimental Research in Education and Psychology” in two ways: moving beyond experimental research and including small samples as well. Both extensions require specific modifications in the pragmatic approach for larger-sample experimental research as introduced in the aforementioned book, including in criteria used for statistical testing and in terms of causal inference. However, like in the approach presented in my book on experimental research, the approach discussed in this chapter is about uniting traditional and emerging approaches to statistical testing and estimation: Frequentist CIs and equivalence testing, Bayesian posterior intervals and the region of practical equivalence, Likelihood ratio testing, and information criteria that are—through the use of Likelihoods—related to the other methods but provide testing outcomes under slightly different assumptions. Concepts of Big Data, Artificial Intelligence, Machine Learning, Educational Data Mining, and Learning Analytics are introduced in this chapter, with examples that are revisited in later chapters in this book. Finally, PASTE as introduced in this chapter incorporates a decision-making framework for dealing with different types of missing data as well.

M3 - Chapter

SN - 978-3-030-43082-5

SP - 35

EP - 65

BT - The Art of Modelling the Learning Process

ER -