PHPS Baseline — Punjab Health & Population Survey

From 24,540 Household Interviews to One Interactive Dashboard: Exploring the PHPS Baseline Data

The Punjab Health & Population Survey (PHPS) Baseline, fielded by the Bureau of Statistics Punjab, is one of the largest household surveys ever conducted in the province — 24,540 households, 143,951 individuals, 19,556 ever-married women, and 36 districts, collected over three months at the end of 2024 and into early 2025. It’s the kind of dataset that usually lives in a 283-page PDF and a handful of SPSS files that only a statistician can open.

Gallup Pakistan’s Digital Analytics team wanted the underlying data to live somewhere else too: in front of anyone who needs an answer in the next thirty seconds.

So we built an interactive dashboard on top of the raw microdata the Bureau of Statistics Punjab collected — and along the way, turned it into something closer to an analytical tool than a report.

## Built to match the official numbers, not approximate them

The first thing we had to get right was methodology. PHPS didn’t sample households in proportion to population — smaller and rural districts got a fixed household quota regardless of size — so a naive percentage computed straight from the raw rows would be wrong, sometimes by ten points or more. We applied the survey’s own design weights throughout, the same way the Bureau of Statistics Punjab’s official report does.

We checked our work against that report’s own published numbers: household size (5.77 vs. the report’s 5.79), owned-dwelling rate (52.3% — exact match), cooking fuel mix (natural gas 37.8%, wood 31.0% — exact match), antenatal care coverage (90.0% — exact match to two decimal places). When the methodology is right, the numbers converge on their own.

## Every number, four ways

Here’s the part that took the dashboard from “static report” to “exploration tool”: *every one of the 280-plus indicators is clickable*, and clicking one shows the same statistic broken down four different ways —

– *By district* — all 36, ranked highest to lowest
– *On a map* — an actual choropleth of Punjab, shaded by value, with district names labeled
– *By education level* — of the household head or the woman herself, depending on the indicator
– *Urban vs. rural*

Ask “what’s the motorcycle ownership rate?” and you get 62.8%. Click it, and you can see it’s 78% in Lodhran and 43% in Dera Ghazi Khan — or that it climbs steadily from 52% among households with no education to 82% among the most educated. The same drill-down works whether you’re looking at asset ownership, immunization coverage, antenatal care, or household income source.

Every chart — including the map — has a one-click PNG export, so a number you find at 11pm can be in a slide deck by 11:05.

## Beyond the standard tables

A few things in the dashboard go past what the printed report covers:

*Fertility rates, computed from scratch.* The report cites Punjab’s Total Fertility Rate from an external DHS-style calculation. We rebuilt it directly from the birth-history microdata using the standard demographic person-years method — live births in the three years before each interview, divided by the female population actually at risk (not just the ever-married subset, which would badly overstate rates for younger women). Our estimate — 2.77 children per woman — lands close to the official 3.2, with the same real-world pattern showing through: Rajanpur, Muzaffargarh, and Dera Ghazi Khan sit well above the provincial average; Sialkot and Rawalpindi well below.

*A first look at infertility.* The survey doesn’t ask a clinical infertility question, but it does ask every ever-married woman about her future fertility intentions — including the option “I cannot get pregnant.” Cross-referencing that against actual living-children counts gives a defensible, if conservative, estimate of unmet fertility desire, broken down the same four ways as everything else.

*Maternal and infant outcomes by delivery type.* We linked each woman’s most recent delivery record to its outcome in the birth-history file — something that required matching pregnancy sequence numbers across two different files with two different encoding conventions. The result: caesarean deliveries show a higher rate of reported maternal complications but roughly average child survival; normal vaginal deliveries show the opposite. The more informative number sits underneath both: only 62% of normal deliveries had a skilled attendant present, versus 97% of caesareans. Who’s in the room turns out to matter more than how the baby arrives.

## What’s next

This is a baseline survey — the first of what we expect will become a recurring instrument. As future waves come in, the same dashboard architecture is built to take them: same weighting approach, same drill-down structure, same district shapes on the map. A wealth index (built from the asset data already in the dashboard) and a formal panel comparison across waves are next on the list.

For now, the dashboard covers twelve areas — household demographics, housing and WASH, assets and social protection, health facility access, morbidity, tuberculosis, women’s profiles, fertility and infertility, maternal care, child health and immunization, and women’s empowerment — all from the same 24,540 households, all cross-checked against the official report, all one click away from the number underneath.

—

The PHPS Baseline Dashboard was built by Gallup Pakistan’s Digital Analytics team directly from the Bureau of Statistics Punjab’s raw survey microdata (SPSS files, Sections A through M). The survey itself was designed and fielded by the Bureau of Statistics Punjab. Methodology notes, including the weighting approach and its validation against the official report, are documented in-dashboard.