Negative binomial mixed-effects model of FSW distribution in sub-Saharan Africa — identifying key determinants for targeted HIV prevention strategies
# Analysis of Factors Influencing Female Sex Worker Distribution in Sub-Saharan Africa
## Overview
This project investigates the distribution of Female Sex Workers (FSWs) across sub-Saharan Africa to identify key environmental and sociological factors that influence their prevalence.
The study incorporates statistical modeling, using mixed-effects regression models, to provide insights into how these variables shape FSW distribution, guiding targeted public health interventions and HIV-prevention strategies.
---
## Objectives
1. Analyze the abundance and distribution of FSWs in sub-Saharan Africa using environmental and sociological factors.
2. Identify critical variables influencing FSW counts to prioritize in future data collection efforts.
3. Model spatial and temporal variability through advanced statistical techniques.
---
## Dataset Description
The dataset contains 750 observations, of which 675 were retained for analysis. It includes variables such as:
- **Response Variable**: `FSWCount` (number of Female Sex Workers observed, discrete and non-negative).
- **Environmental Variables**:
- Population Density (`popDensity`)
- Built-Up Index (`built`)
- Growing Season (`growingSeason`)
- Average Annual Rainfall (`rain`)
- Temperature (`temperature`)
- Nighttime Light Activity (`nightLight`)
- Distance to Clean Water (`cleanWater`)
- **Sociological Variables**:
- HIV Prevalence (`hivRate`)
- High Crime Indicator (`highCrime`)
- Use of Insect Nets (`insectNet`)
- Wealth Index (`wealthIndex`)
- Age of First Sexual Experience (`ageFirstSex`)
- **Random Effects**:
- Country
- Region
- Data Year
- Month
---
## Methodology
1. **Exploratory Data Analysis**:
- Summary statistics and visualizations (histograms, scatterplots, correlation matrix).
- Analysis of relationships between FSW counts and predictors.
2. **Model Selection**:
- Initial Poisson mixed-effects models were evaluated for both environmental and sociological variables.
- Diagnostic checks reveale …