Executive Summary
The dataset does not establish staff members’ personal political views. Most records are classified as UNKNOWN or neutral, and the available material predominantly concerns official NADI/PEDi communications rather than personal accounts.
Of 79 staff records, 67 have unknown government stance and 10 are labelled neutral: 97.5% combined. Two records carry pro-government labels, but their supporting summaries describe institutional promotion of government programmes—not personal political endorsement.
Confidence: High in the reported counts; low in their ability to measure staff political sentiment.
Key Findings
- Public-service information, digital literacy, health advisories and community programmes dominate the supplied examples.
- References to PADU, MADANI initiatives or elected representatives do not, by themselves, demonstrate political allegiance.
- No anti-government classifications are reported. This is an observation about the dataset, not proof that opposition views are absent.
- UNKNOWN means insufficient evidence; it should not be interpreted as neutrality, support or opposition.
- Institutional pages appear to have been attributed to individual staff, creating a significant risk of misclassification.
Party Inclination
| Supplied classification | Staff records | Share | |---|---:|---:| | UNKNOWN | 70 | 88.6% | | BEBAS / Independent | 9 | 11.4% |
No records are assigned to a named political party.
The BEBAS label appears to reflect non-partisan content or lack of clear affiliation. It should not be presented as confirmed independent political identification.
Government Stance
Relative to Kerajaan MADANI (Federal unity government):
| Supplied classification | Staff records | Share | |---|---:|---:| | Unknown | 67 | 84.8% | | Neutral | 10 | 12.7% | | Pro-government | 2 | 2.5% | | Anti-government | 0 | 0.0% |
The two pro-government labels have supplied confidence scores of 0.74 and 0.56. These scores are not validated probabilities of personal support. Both underlying summaries concern official programme communications, making personal-stance attribution unsupported.
The taxonomy also requires care: BEBAS is marked outside government, but non-affiliation must not automatically be treated as opposition.
Sites of Interest
For account-attribution and classification review—not political targeting:
- Kampung Padang, Pokok Mangga: One record is labelled pro-government based on official-site promotion of a MADANI programme.
- Bukit Beruang: One pro-government and one neutral record. Programme promotion and event acknowledgements should be assessed consistently as institutional content.
- Air Panas: Both listed records are neutral; examples cover registration services, health information and digital safety.
Each listed site has only two staff records. These small counts cannot support reliable conclusions about site-wide sentiment.
Notable Individuals
No individual should be designated politically notable from this evidence. The named examples largely concern workplace or community pages, and associating those communications with employees’ private political beliefs would be unwarranted.
Individual names are therefore omitted from this brief; review should focus on account ownership and published content rather than inferred personal allegiance.
Data Coverage & Limitations
- 79 staff; 77 crawled (97.5%); 64 accounts; 53 posts.
- Crawl completion does not establish successful personal-account identification or adequate evidence per employee.
- The site table covers 20 staff across 10 sites, not the entire workforce.
- Only 12 sample posts are reproduced; the complete underlying corpus is unavailable here.
- No collection window, deduplication method, account-verification procedure or confidence-calibration methodology is provided.
- Supplied notable-record confidence scores range from 0.32 to 0.92, but confidence in content classification is distinct from confidence in personal political attribution.
Recommended Actions
- Do not use these labels for employment, disciplinary or loyalty decisions.
- Separate institutional communications from personal-account material; remove unsupported personal political attributions.
- Reassess programme promotion using consistent, non-partisan content criteria.
- Document collection dates, account ownership, missing coverage and classification methodology.
- Report aggregate communications themes rather than employee political profiles. If staff feedback is needed, use voluntary, anonymous workplace surveys without collecting party preferences.