electricsheepafrica/africa-social-connections-survey

---
annotations_creators:
- no-annotation
language_creators:
- found
language:
- en
license: cc-by-4.0
multilinguality:
- monolingual
size_categories:
- 1K<n<10K
source_datasets:
- original
task_categories:
- tabular-classification
- tabular-regression
- other
task_ids: []
tags:
- africa
- humanitarian
- hdx
- electric-sheep-africa
- demographics
- health
- bra
- egy
- fra
- ind
- idn
pretty_name: "Social Connections Survey"
dataset_info:
splits:
- name: train
num_examples: 2107
- name: test
num_examples: 526
---

# Social Connections Survey

**Publisher:** AI for Good at Meta · **Source:** [HDX](https://data.humdata.org/dataset/social-connections-survey) · **License:** `cc-by` · **Updated:** 2026-03-26

---

## Abstract

Gallup, Meta and a group of academic advisors collaborated to design and conduct the Social Connections Survey, which offers a first look at how social connections vary across diverse geographic regions. The survey was administered through face-to-face or phone interviews with people aged 15 and older in seven countries. The data provides an in-depth look at the extent to which people feel connected, socially supported and lonely in different parts of the world. The data also sheds new light on the characteristics of people’s social connections, the ways people interact with others, the groups with whom they have frequent contact, and how they connect with others to get support when they need it.

Please see the [State of Social Connections Study](https://dataforgood.facebook.com/dfg/docs/2022-state-of-social-connections-study) on the Data for Good website for more information and a report with key findings. Aggregate country-level data for each item in the survey is publicly available below. Controlled access microdata is available for request through the [AI for Good Program at Meta](https://ai.meta.com/ai-for-good/datasets/social-connections-survey/).

Each row in this dataset represents country-level aggregates. Data was last updated on HDX on 2026-03-26. Geographic scope: **BRA, EGY, FRA, IND, IDN, MEX, USA**.

*Curated into ML-ready Parquet format by [Electric Sheep Africa](https://huggingface.co/electricsheepafrica).*

---

## Dataset Characteristics

| | |
|---|---|
| **Domain** | Public health |
| **Unit of observation** | Country-level aggregates |
| **Rows (total)** | 2,634 |
| **Columns** | 12 (5 numeric, 7 categorical, 0 datetime) |
| **Train split** | 2,107 rows |
| **Test split** | 526 rows |
| **Geographic scope** | BRA, EGY, FRA, IND, IDN, MEX, USA |
| **Publisher** | AI for Good at Meta |
| **HDX last updated** | 2026-03-26 |

---

## Variables

**Geographic** — `country` (US, ID, IN), `question_text` (Which of the following best describes your current employment status?, Now, think about ALL the friends you may have, including your close friends and ANYONE ELSE you would consider a friend. Overall, about how many friends would you say you have?, For this next question, please think of a CLOSE friend as someone you can talk with about things that are most important to you, including sensitive issues. Overall, about how many CLOSE friends would you say you have?).

**Outcome / Measurement** — `value` (DK/NR, Never, Yes).

**Identifier / Metadata** — `esa_source` (HDX), `esa_processed` (2026-04-06).

**Other** — `variable` (employment_status, num_friends_bucket, num_close_friends_bucket), `estimate_weighted` (range 0.0001–3.3583), `estimate_weighted_se` (range 0.0001–0.0375), `estimate_weighted_95ci_low` (range 0.0–3.3242), `estimate_weighted_95ci_upp` (range 0.0008–3.3923) and 2 others.

---

## Quick Start

```python
from datasets import load_dataset

ds = load_dataset("electricsheepafrica/africa-social-connections-survey")
train = ds["train"].to_pandas()
test = ds["test"].to_pandas()

print(train.shape)
train.head()
```

---

## Schema

| Column | Type | Null % | Range / Sample Values |
|---|---|---|---|
| `country` | object | 0.0% | US, ID, IN |
| `variable` | object | 0.0% | employment_status, num_friends_bucket, num_close_friends_bucket |
| `value` | object | 0.5% | DK/NR, Never, Yes |
| `estimate_weighted` | float64 | 0.0% | 0.0001 – 3.3583 (mean 0.2346) |
| `estimate_weighted_se` | float64 | 0.0% | 0.0001 – 0.0375 (mean 0.0092) |
| `estimate_weighted_95ci_low` | float64 | 0.0% | 0.0 – 3.3242 (mean 0.2174) |
| `estimate_weighted_95ci_upp` | float64 | 0.0% | 0.0008 – 3.3923 (mean 0.2535) |
| `n_unweighted` | int64 | 0.0% | 1.0 – 2076.0 (mean 454.88) |
| `question_text` | object | 1.1% | Which of the following best describes your current employment status?, Now, think about ALL the friends you may have, including your close friends and ANYONE ELSE you would consider a friend. Overall, about how many friends would you say you have?, For this next question, please think of a CLOSE friend as someone you can talk with about things that are most important to you, including sensitive issues. Overall, about how many CLOSE friends would you say you have? |
| `notes` | object | 73.6% | Question only asked if support_need_freq_30d is Rarely, Sometimes, or Often. All estimates for this variable are reported as proportions of response over the entire country-level population., For random half of participants, question only asked if num_friends_bucket is not 0 (otherwise question asked before num_friends_bucket)., Question only asked if 1) support_need_freq_30d is Rarely, Sometimes, or Often AND 2) Yes to any of support_interaction_mode_is_phone_30d, support_interaction_mode_is_video_30d, support_interaction_mode_is_messaging_30d, or support_interaction_mode_is_social_media_30d. All estimates for this variable are reported as proportions of response over the entire country-level population. |
| `esa_source` | object | 0.0% | HDX |
| `esa_processed` | object | 0.0% | 2026-04-06 |

---

## Numeric Summary

| Column | Min | Max | Mean | Median |
|---|---|---|---|---|
| `estimate_weighted` | 0.0001 | 3.3583 | 0.2346 | 0.1704 |
| `estimate_weighted_se` | 0.0001 | 0.0375 | 0.0092 | 0.0102 |
| `estimate_weighted_95ci_low` | 0.0 | 3.3242 | 0.2174 | 0.1497 |
| `estimate_weighted_95ci_upp` | 0.0008 | 3.3923 | 0.2535 | 0.1931 |
| `n_unweighted` | 1.0 | 2076.0 | 454.88 | 333.5 |

---

## Curation

Raw data was downloaded from HDX via the CKAN API and converted to Parquet. Column names were lowercased and standardised to snake_case. Common missing-value markers (`N/A`, `null`, `none`, `-`, `unknown`, `no data`, `#N/A`) were unified to `NaN`. The dataset was split 80/20 into train and test partitions using a fixed random seed (42) and saved as Snappy-compressed Parquet.

---

## Limitations

- Data originates from AI for Good at Meta and has not been independently validated by ESA.
- Automated cleaning cannot correct for misreported values, definitional inconsistencies, or sampling bias in the original collection.
- The following columns have >20% missing values and should be treated with caution in modelling: `notes`.
- This dataset spans 7 countries; geographic and methodological inconsistencies across national boundaries may affect cross-country comparability.
- Refer to the [original HDX dataset page](https://data.humdata.org/dataset/social-connections-survey) for the publisher's own methodology notes and caveats.

---

## Citation

```bibtex
@dataset{hdx_africa_social_connections_survey,
title = {Social Connections Survey},
author = {AI for Good at Meta},
year = {2026},
url = {https://data.humdata.org/dataset/social-connections-survey},
note = {Repackaged for machine learning by Electric Sheep Africa (https://huggingface.co/electricsheepafrica)}
}
```

---

*[Electric Sheep Africa](https://huggingface.co/electricsheepafrica) — Africa's ML dataset infrastructure. Lagos, Nigeria.*