---
# For reference on dataset card metadata, see the spec: https://github.com/huggingface/hub-docs/blob/main/datasetcard.md?plain=1
# Doc / guide: https://huggingface.co/docs/hub/datasets-cards
license: apache-2.0
task_categories:
- tabular-classification
- tabular-regression
- text-classification
language:
- en
- ms
tags:
- socioeconomic-disparities
- technology-sector
- wage-gap
- malaysia
pretty_name: Malaysian Tech Disparity Analysis 2025
size_categories:
- 100K<n<1M
---

# Dataset Card for Malaysian Tech Disparity Analysis

## Dataset Details

### Dataset Description
A comprehensive longitudinal dataset (2015-2025) analyzing ethnic disparities in Malaysia's technology sector, covering:
- Intergenerational mobility metrics
- Wage gaps across 12 tech subsectors
- Digital literacy scores by state
- Automation risk assessments

**Key Features:**
- 428,000+ anonymized records
- District-level geocoding
- EPF-verified salary data
- Bumiputera subgroup granularity (Malay, Orang Asli, Sabah/Sarawak natives)

**Example Research Questions:**
1. How do Chinese-Malay wage gaps vary between AI and traditional IT roles?
2. What's the correlation between digital literacy and intergenerational mobility?
3. Which states show the fastest narrowing of tech access disparities?

### Dataset Sources
- **Curated by:** Malaysian Digital Economy Corporation (MDEC) Analytics Team
- **Funded by:** Ministry of Science, Technology and Innovation (MOSTI)
- **Primary Sources:**
  - Department of Statistics Malaysia (DOSM) microdata
  - EPF wage records (2015-2025)
  - MyDigital Literacy Census
- **Repository:** [Hugging Face Dataset Link]
- **License:** Apache 2.0 (commercial use permitted with attribution)

## Uses

### Direct Use
- Policy analysis for digital inclusion programs
- Corporate DEI benchmarking
- Academic research on labor economics
- Predictive modeling of automation impacts

### Out-of-Scope Use
- Individual-level discrimination claims
- Automated hiring decisions
- Political redistricting (geographic precision limited to districts)

## Dataset Structure

**Core Tables:**
1. `demographics` - Ethnicity and population shares
2. `mobility` - Intergenerational education/income mobility
3. `wages` - Tech sector compensation by role
4. `skills` - Digital competency assessments

**Field Examples:**
```python
{
  "ethnicity_id": 2,  # Chinese
  "state": "Selangor",
  "job_role": "Data Scientist",
  "median_salary": 12500.00,  # MYR
  "automation_risk": 18.7,  # Percentage
  "digital_literacy": {
    "basic_skills": 82.1,
    "ai_competency": 34.5 
  }
}
```

**Splits:**
- `train` (2015-2023): 380K records
- `test` (2024-2025): 48K records
- `validation` (random 5% sample): 21K records

## Dataset Creation

### Curation Rationale
Created to support Malaysia's 12th Plan (2021-2025) goal of reducing ethnic disparities in tech sector participation by 30%.

### Source Data
**Collection Methods:**
1. **Wage Data:** Aggregated from EPF contributions with industry coding
2. **Skills Data:** Proctored assessments at 126 testing centers nationwide
3. **Mobility Data:** Longitudinal household surveys (N=15,000 families)

**Processing Pipeline:**
```mermaid
graph TD
    A[Raw EPF Data] --> B[Anonymization]
    C[Survey Responses] --> D[Geocoding]
    B --> E[Industry Classification]
    D --> E
    E --> F[Quality Control]
    F --> G[Parquet Export]
```

### Annotations
**Salary Gap Flags:**
- Annotated by MDEC economists using threshold of >15% below sector median
- Inter-rater reliability: κ = 0.82

**Ethnicity Coding:**
- Validated against National Registration Department categories
- 99.3% consistency with self-reported identity

## Bias, Risks and Limitations

**Known Biases:**
1. **Geographic Coverage:** East Malaysia samples 23% smaller than population share
2. **Freelance Workers:** Underrepresented in formal wage records
3. **Age Bias:** Gen Z (18-24) coverage begins only in 2020

**Mitigation Strategies:**
- Clear documentation of coverage gaps
- Sample weighting variables included
- Partnered with Grab/MYWork to supplement gig economy data

### Recommendations
1. **For Researchers:** Use provided sample weights for national estimates
2. **For Policymakers:** Cross-validate with DOSM's granular household surveys
3. **For Companies:** Combine with internal HR data for DEI analysis

## Citation

**BibTeX:**
```bibtex
@dataset{mdec_tech_disparity_2025,
  title = {Malaysian Technology Sector Disparity Analysis 2025 Edition},
  author = {MDEC Analytics Team},
  year = {2025},
  publisher = {Malaysian Digital Economy Corporation},
  version = {3.1.0},
  url = {https://huggingface.co/datasets/MDEC/Tech-Disparity-MY}
}
```

**APA:**
MDEC. (2025). Malaysian Technology Sector Disparity Analysis [Data set]. https://huggingface.co/datasets/MDEC/Tech-Disparity-MY

## Glossary
- **B40/T20:** Income percentile groups (Bottom 40%/Top 20%)
- **MyDigital:** National digital literacy certification program
- **GIGIS:** Government Integrated Geographic Information System (used for geocoding)

## Dataset Card Authors
- Kurnia Kadir

## Dataset Card Contact
kurnia.kadir@chemmara.com