Objective:

The data for the study was obtained via paper-based surveys, including scale-based, Likert-scale, binary-response, and multiple-response items for a total of 462 participants. The primary objective was to identify post-diagnosis dietary changes among breast cancer patients and evaluate how these trends vary by ethnicity. A secondary objective was to explore the way patients evaluate dietary and medical information as trust-worthy.

Data Preparation:

  • Cleaned and validated data collected via paper-based surveys, removing any duplicate cases and correcting any erroneous imputations by the researchers.
  • Matching variables for Hebrew and Arabic surveys, and removing any irrelevant features.
  • Transforming features to the correct data types (e.g., numeric, binary, etc…)
  • Removing items that did not meet a pre-defined response rate

Statistical Analysis

1. Sample Size and Power

  • During the data collection phase, calculated the needed sample size to obtain a given effect size and power level, taking into account multiple comparisons and p-value corrections.

2. Descriptive and Exploratory Analysis

  • Compared demographic and clinical characteristics by ethnicity, using parametric and non-parametric tests.

3. Comparing Dietary Preferences

  • Statistical analyses were selected based on the nature of the data: Wilcoxon and Kruskal-Wallis tests for Likert-scale and non-normally distributed continuous data, and chi-squared tests for binary and categorical responses.

The main results of the study are visualized below:

Figure 1:

Figure 2:

Figure 3:

Figure 4:

Reporting Results

  • Cleanly summarized statistical methods and results in a journal oriented fashion.
  • Summarized results cleanly in tables.
  • Created publication-quality visualizations (histograms and boxplots).
  • Communicated results to non-technical medical researchers and assisted in drawing appropriate conclusions.