Date of Award

1-1-2022

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

College/School/Department

Department of Educational and Counseling Psychology

Program

Educational Psychology and Methodology

Content Description

1 online resource (x, 197 pages) : illustrations (some color)

Dissertation/Thesis Chair

Mariola Moeyaert

Committee Members

Kimberly Colvin, Wim Van den Noortgate

Keywords

Bootstrapping, Methodology, Monte Carlo Simulation, Non-Overlap, Sampling Distribution, Single-Case Experimental Design, Bootstrap (Statistics), Resampling (Statistics), Monte Carlo method

Subject Categories

Educational Psychology | Statistics and Probability

Abstract

Statistics can be either parametric or non-parametric, depending on whether distributional assumptions about the data and sampling distributions are required. Parametric and non-parametric approaches each include a variety of inferential methods. These methods can be seen in the field of single-case experimental design (SCED). By reviewing one of the most used groups of statistics for SCED research (Jamshidi et al., 2022; Maggin et al., 2011), the non-overlap indices, one major issue arose. It is challenging to make inferences and interpretations about non-overlap indices. The main reason for this issue is that non-overlap indices have inconsistent and unknown sampling distributions (under the alternative hypotheses that there is an intervention effect). Therefore, statistical hypothesis testing cannot be applied to make inferences about the non-overlap indices. Benchmarking, the current method used to make inferences about non-overlap indices, is problematic as it is not context-specific and often misused. This dissertation intends to address these issues by (a) investigating the commonly used resampling technique (i.e., bootstrapping) in patterning the sampling distributions of non-overlap statistics through a Monte Carlo simulation, and (b) applying the bootstrapping method to establish Critical Intervals (i.e., the 95% interval range established by the bootstrapping sampling distribution) of non-overlap statistics. Bootstrapping is a generic method that can be applied to different statistics and data; thus, it is promising in patterning sampling distribution of non-overlap indices. Using the Critical Interval, formed by the bootstrapping method, is demonstrated in this dissertation to provide consistent inferences about intervention effectiveness across different non-overlap indices. Researchers are encouraged to use Critical Intervals over benchmarks in making inferences about non-overlap indices.

Download

Share

COinS