Statistical Analysis Plan
For Ahlam’s thesis
Motivation
The thesis has been rejected due to a poor results chapter and incorrect statistical analyses.
This document outlines the proposed statistical analysis plan.
- This is a basic, graduate-level analysis that should comply with academic practices.
- More intricate analyses can be conducted, but they are likely to be under-powered given the small sample size.
In the following section the proposed changes to each section of the thesis are stated.
- Small, ad-hoc adjustments may be required during the analysis;
- In any case, the proposal includes a complete write-up of the results section.
- Guidance for restructuring the thesis in accordance with Faculty guidelines will be provided.
Proposed Changes by Chapter
Introduction
No major changes are required.
References for key statetaments (e.g “Muslim Arabs are at risk for sleep disturbance”) will be sought using RefSeek.
The research hypotheses remain unchanged.
Methods
- The questionnaire scores
The explanatory variables
The Total PSQI score will be used (without the subscales) for Hypothesis 1.
The PSQI sleep-quality index will be used for Hypothesis 2.
The IPAQ1 scores will be aggregated in accordance with the WHO guidelines(“Global Physical Activity Questionnaire (GPAQ),” n.d.).
- The data will be re-cleaned and re-processed.
- In particular, the continuous score from the short-form IPAQ will be used.2
Control variables:
- Age - grouped.
- Sex.
- Marital status.
- Education - dichotomous: academic/non-academic.
- Health and Medical history.
Results
- Descriptive statistics:
- Table 1: Subject characteristics
- Table 2: Sleep quality indexes.
- Table 3: Physical activity indexes.
- Table 4: Correlations between study variables.
- Hypothesis 1 - A quadratic regression model will be fitted to the IPAQ continuous score regressed on the PSQI total score as follows:
\[\text{IPAQ}_i = \beta_0 + \beta_1 \cdot \text{PSQI} + \beta_2 \cdot \text{PSQI}^2 + \text{covariates}.\] - Research hypothesis 1 operationalized: \(H_0: \beta_2 \geq 0\) vs \(H_1: \beta_2 < 0\). - The quadratic term is expected to be negative.
- Hypothesis 2 - A linear regression model will be fitted to the IPAQ continuous score as follows:
\[\text{IPAQ}_i = \beta_0 + \beta_1 \cdot \text{PSQI}_\text{Quality} + \text{covariates}\] - The PSQI sleep quality will be handled as an interval scale thus assuming that the effect of PSQI sleep quality is linear.
- Additional analyses - interesting subgroup analyses will be conducted and selected results will be reported as needed.
Discussion
- This section needs amendment, its extent depends on the results.