Open Data Is Not Enough

Creating Reusable Census Data Infrastructure

Andrea Gomez Vargas

UNILORIN - R Users Group

July 15, 2026

Who am I?

• Sociologist
• Population Statistics Analyst
• Using R to improve access to public data
• Learned R through the R community
• rOpenSci Champion (2023–2024) & R in Buenos Aires

Before we begin, let me briefly introduce myself.

I am a sociologist and a population statistics analyst. My work focuses on official data, reproducible research, and the use of R to improve access to public information.

I learned R through the R community, so community has always been an important part of my journey. It also shaped the way I think about open-source tools, collaboration, and knowledge sharing.

I was also part of the rOpenSci Champions Program in the 2023–2024 cohort, where I developed arcenso, the project that I will use as a case study in this talk.

Why Open Data Matters

• Transparency
• Research
• Evidence-based policymaking
• Official statistics
• Open science

Before talking about infrastructure, I’d like to start with a simple question: why do open data matter?

Over the last two decades, governments, national statistical offices, and research institutions have made tremendous progress in publishing public data.

Today, open data play a fundamental role in increasing transparency, supporting reproducible research, enabling evidence-based policymaking, and encouraging innovation.

Making official statistics publicly available also helps researchers, journalists, policymakers, and citizens better understand society and make informed decisions.

There is no doubt that open data have transformed the way we work.

So, if open data have transformed research and public policy so much… why do researchers still spend so much time preparing data before they can actually use them?

Is Open Data Enough?

This question became the starting point of my project.

As a researcher working with official statistics, I realized that although many datasets were publicly available, using them was often much harder than accessing them.

Researchers rarely start their work by answering research questions.

Instead, they spend a considerable amount of time searching, downloading, cleaning, harmonizing, and documenting data before any analysis can begin.

So I started wondering:

Is making data available really enough?

Or are we missing something?

Open data ≠ Reusable data

If data are open, why are they still so difficult to reuse?

This is the main idea I would like to share today.

For many years, we have focused on making data open and publicly available. And that has been a tremendous achievement.

But over time, I realized that making data available does not necessarily make them reusable.

In many cases, the data are already online. They are free to access and download.

However, anyone working with public data often faces the same challenge.

Before answering a question or building an analysis, we spend time searching, understanding documentation, cleaning files, harmonizing variables, and combining information from different sources.

So, the question is no longer whether data are open.

The question is whether they are truly reusable.

And this is where data infrastructure becomes essential.

Open data is about access. Reusable data is about use.

The Hidden Cost of Reusing Data

When we talk about reusing public data, the challenge is usually not access.

The challenge is diversity.

The same information may be distributed through books, PDF reports, spreadsheets, databases, or web platforms.

Documentation changes over time.

Variable names change.

Classifications evolve.

Different census years often use different dissemination strategies.

None of these formats are wrong.

But together they create a significant barrier to reuse.

What Does This Mean for Users?

Before answering one question, we first need to make the data usable.

So, what does this actually mean for the people who work with public data?

Before we can answer a research question, build a dashboard, or support a policy decision, we first need to prepare the data.

That means finding the right datasets, understanding the documentation, cleaning the files, harmonizing variables across different years, and integrating information from multiple sources.

None of these tasks are the final goal, but they are essential.

And very often, they take more time than the analysis itself.

Even more importantly, this work is repeated over and over again by different people working on the same data.

Instead of building new knowledge, we spend valuable time rebuilding the same data preparation process.

This is exactly the gap that reusable data infrastructure aims to reduce.

What Makes Data Reusable?

• Accessibility
• Interoperability
• Reusability
• Reproducibility

Making data reusable requires more than publishing them.

So, if open data are not necessarily reusable, what makes data reusable?

In my experience, reusability is not a single feature. It is the result of several characteristics working together.

First, data need to be accessible. People should be able to find and obtain them.

Second, they need to be interoperable. Different datasets should be able to work together, even when they come from different sources or different years.

Third, they need to be reusable. That requires clear documentation, metadata, and a structure that people can easily understand.

Finally, they should support reproducible workflows, so that others can understand not only the data, but also the process used to prepare them.

These principles help transform data from something that is simply available into something that can actually be used.

And This is exactly the idea behind reusable data infrastructure.

Reusable Data Infrastructure

Different Sources

↓

Shared Infrastructure

↓

Reusable Data

So, if we want data to become reusable, we need infrastructure.

Infrastructure is what transforms raw data into reusable data.

Instead of every person repeating the same work, we prepare the data once, document the process, and make that work available for others.

The question is: what does that look like in practice?

Let me show you using a real example.

Why Census Data?

• One of the most important sources of official statistics

• More than 50 years of information

• Essential for research and public policy

• Ideal for studying change over time

Why did I choose census data?

Because censuses are one of the most valuable sources of official statistics.

They provide information that supports research, public policy, planning, and decision-making.

More importantly, they allow us to study change over time.

But working with data collected over more than fifty years also means dealing with changing technologies, formats, and dissemination practices.

That makes census data an ideal case study for thinking about reusable data infrastructure.

The Argentine Census

1970 > 1980 > 1991 > 2001 > 2010 > 2022

Argentina has conducted national censuses for more than fifty years.

The information is available, but every census reflects the technology of its time.

Some results were published as printed books.

Others as PDF reports.

Later, data became available through REDATAM and web platforms.

Each step improved access, but also introduced new formats, structures, and workflows.

As a result, comparing census data across decades became increasingly difficult.

That was the problem I wanted to solve.

How We Built ARcenso

• Download census data from official sources
• Organize and classify historical files
• Transform data into tidy, standardized tables
• Develop reusable R functions
• Package everything for reproducible use

Building arcenso was an iterative process.

First, I collected census data from official sources spanning more than fifty years.

Then, I organized and classified the files, transformed them into tidy and standardized datasets, and documented each step of the process.

Finally, I developed reusable R functions and packaged everything into an open-source R package, making both the data and the workflow easier to reuse.

How We Built ARcenso

From excel tables to ordered tables in R

Original excel download

Tidy table in R

The proposal was to transform those scattered spreadsheets into tidy tables. Where before there was chaos, now there is reproducible structure.

What Can You Do with ARcenso?

• Access historical census data with a consistent interface
• Compare census years using harmonized tables
• Build reproducible analyses
• Focus on research instead of data preparation

The goal of arcenso is not simply to provide another dataset.

The goal is to reduce the time people spend preparing data.

Instead of searching through different formats, cleaning files, and harmonizing variables every time, users can focus on their research questions.

The package provides a consistent interface across census years and promotes reproducible workflows.

Lessons Learned

• Open data is only the beginning

• Infrastructure reduces repeated work

• Documentation is part of the product

• Open source enables collaboration

• Communities make projects sustainable

Building arcenso taught me that publishing data is only the first step.

What really makes a difference is investing in infrastructure that others can reuse.

Documentation is not an extra task; it is part of the product.

Open-source tools make collaboration easier, but communities are what make projects sustainable over time.

These lessons apply far beyond census data.

The Role of Open Source Communities

Mentorship | Collaboration | Feedback | Sustainability

arcenso would not exist without the open-source community.

I learned R through community events, received mentorship through rOpenSci, and continue sharing this work through R-Ladies+, LatinR, and local R User Groups.

Communities don’t just teach us how to code.

They help us build better tools, share knowledge, and create sustainable projects.

Beyond Census Data

• Every dataset has reusable potential
• Small improvements can have a big impact
• Build tools, not only analyses
• Share workflows, not only results

My hope is that this talk is not only about census data.

Every institution, every research group, and every community works with data that could become easier to reuse.

You don’t need to build a large package.

Sometimes documenting a workflow, standardizing a dataset, or sharing a simple function can save hours of work for someone else.

Building reusable infrastructure starts with small, practical steps.

Thank U, Gracias! 😁