Nqoubiler/GHS-Income-Analysis

Analyzing South African household salaries using GHS data with SRS, stratified, and cluster sampling methods in SAS. ## General Household Survey (GHS) using SAS --- This project involved a statistical analysis of South African household income data using the General Household Survey (GHS) to estimate national salary indicators and evaluate the performance of different probability sampling techniques. Using SAS, I applied survey sampling methods to produce reliable estimates that can support government planning, poverty reduction strategies, and socio-economic decision-making. Completed as part of the STAT395 final project at the University of KwaZulu-Natal. --- ## Project Objective The purpose of this analysis was to answer key economic questions: - What is the estimated average household monthly salary in South Africa? - What proportion of households earn below the poverty threshold of **R4,500** per month? - What is the estimated total monthly household salary generated nationally? - Which sampling method provides the most precise and reliable estimates? --- ## Dataset Summary ### Source: South African General Household Survey (GHS) ### Main Variable: - `LAB_SALARY_HH` – Household Monthly Salary ### Additional Variables Used: - `FIN_INC_MAIN` – Main source of income filter - `PROP` – Indicator for households earning below R4,500 - `FIN_EXP` – Household expenditure used for stratification - `PROV` – Province used as cluster unit --- ## Data Preparation Performed Before analysis, data was cleaned and prepared by: - Removing zero salary values where contextually invalid - Filtering households whose main income source was salary - Creating binary poverty indicator (< R4,500) - Structuring variables for sampling design This ensured cleaner estimates and stronger statistical validity. --- ## Sampling Techniques Applied ### 1. Simple Random Sampling (SRS) - With replacement - Without replacement ### 2. Stratified Sampling Used household monthly expenditure groups to improve estimate precision. ### 3. Cluster Sampling Used provinces as natural cluster units. …