When data journalism meets science

ALESSIO CIMARELLI
Data scientist at Dataninja
jenkin@dataninja.it | @jenkin27


International School of Science Journalism
The Digital World (Erice, June 10th, 2014)

Who am I?

aka jenkin

PAST
Master Degree in Physics at the University of Rome "La Sapienza"

Master in Science Communication at the International School for Advanced Studies (SISSA-ISAS) in Trieste

Press officer at the European Laboratory for Non-Linear Spectroscopy (LENS) in Florence

PRESENT
Freelance data journalist, web developer, open data activist, citizen scientist, ...

Dataninja

Data journalism & data visualization made in Italy

Science journalism

You know very well how it works... :)

Science in journalism

As topic


Stories about the edge of scientific research and human knowledge.


Key role in relationship between science and society.


Science journalist can be a watchdog against false science and scientific frauds.

Science in journalism

As method

It would be evident in investigative journalism, because the workflow is similar to police inquiries or scientific research.

Many informations from different sources, accountability problems, hypothesis and proofs, trial and error cycles, and so on.

Not only a story, but also a discovery itself...

Data-driven journalism

A word in a buzzwords era

There was 5 exabytes of information created between the dawn of civilization through 2003, but that much information is now created every 2 days, and the pace is increasing (Eric Schmidt)

A data journalist is no more than a journalist... when his investigation is ultimately based on (or driven by) digital data, he acquires such prefix.

If a journalist want to tell the world, and the world is now made of digital and quantitative informations, he has to acquire skills in management and interpretation of data, or he will miss an opportunity.

Skills involved in data journalism

Teamwork and multidisciplinary


Journalism

Nose for news, public interest, intuition based on contest knowledge


Data science / statistics

Analytical mind, mathematical and statistical skills, intuition based on science of numbers

Skills involved in data journalism

Teamwork and multidisciplinary


Software developing / hacking

Problem solving, hi-tech knowledge in hardware and software, nerd (or geek, if you prefer) mood


Design

Artistic sensibility and intuition, knowledge in User Experience theory and techniques

Workflow of data journalism

Miners, dustmen, researchers, and story tellers


Find the data

Public search engines or deep web? Official 5-stars open data or web spiders and screen scrapers? Monitor and keyboard, smartphone and touch, or boots and mud?


Control, verify and clean

Data should be read by machines and not by humans! Datasets could hide errors, inconsistencies, lies... or show only a part of a story.

Workflow of data journalism

Miners, dustmen, researchers, and story tellers


Analyze the data

Normalizations and comparisons, filtering, grouping, aggregation, correlations, ...


Visualize this!

How to represent numbers and relations among numbers? Yes, with arabic numerals, but  pictures are worth a thousand words...  as long as you keep in mind that there are facts behind the numbers, and facts are sacred (copyright of The Guardian).

From theory to practice



Science in data journalism

In method

You run into a dataset and feel the presence of a possible news...
OR
... you have an interest, an idea, a thesis, so you are looking for data.

Having quantitative data about a phenomenon means that somewhere there is a measurement procedure you have to understand, test, verify... and interpret!

Data themselves can suggest new ways for your investigation or even falsify some hypothesis or assumptions.

Common sense, intellectual honesty, professional ethics

Data journalism about science

Some random examples

Data Journalism Awards 2014

Data flavours in science

Research data, science world, citizen science



Research data

Hard sciences and social sciences

Ok, neither LHC petabytes are for journalists, nor statistical data from epidemiologic surveys.

But synthetic datasets in scientific reports, or (open) data from national health systems, why not?

If you are not specialized in a specific topic or if you lack the knowledge about the framework, you can ask to an expert you trust.

You can also use numbers not in an investigation, but to tell a complex story using infographics and interactive visualizations.

Data about science world

Bibliographies, social networks of scientists, infrastructures


Science is a human activity and an industry (almost) like any other.

How are the European funds invested in scientific research? Where are the centers specialized in the treatment of specific diseases? Why some well known monitoring technologies are not used in some countries?


Data from citizen science

Sensor-based journalism

Cheap electronics and sensors
+
open hardware
+
free information sharing
=
data from stakeholders other than scientists

It's early, but promising:

Data journalism as a toolbox

for science journalism


Not THE way to make today's journalism,
but A way can be useful in some circumstances


If you have data, it's better if you know how to deal with them.

If you think you may find some data, it's better if you use them.

If someone use data, it's better if you can check his claims.

Play with data is funny!

Data sources

Welcome to the jungle!

Data sources

Some examples

  • Public administration
  • International organizations
  • NGOs
  • Civic activists
  • Press offices
  • Leaks
  • Social networks
  • Journalistic sources
  • Single journalists
  • Ourselves...

Open data

Data made public and reusable

"A piece of data or content is open if anyone is free to use, reuse, and redistribute it — subject only, at most, to the requirement to attribute and/or share-alike.” - See more at opendefinition.org

Big Data

Remember the buzzword era?

"Big data usually includes data sets with sizes beyond the ability of commonly used software tools to capture, curate, manage, and process the data within a tolerable elapsed time. Big data sizes are a constantly moving target, as of 2012 ranging from a few dozen terabytes to many petabytes of data in a single data set." - See more at Wikipedia

  • Data from big science experiments (Atlas, Human Brain Project, ...)
  • Google Earth Engine
  • Social networks (Facebook,  Twitter, but also eBay, Amazon, ...)

Maybe it's not for journalists, but it's a hot topic...

Structured data

For machine, not for human

The keyword is metadata!

A metadatum is an atomic information that describes a given datum, and also describes its relationship with the other data. Structuring a data set means making explicit the meaning and relationships between data.

A well-formed table represent a structured data set. A list of facebook comments, articles of a newspaper, a recorded speech are not structured data (and so are not machine-readable).

Getting (reusable) data

It all depends on the format

If we have Gladstone Gander as best friend:
  • spreadsheet (xls, xlsx, ods, csv, tsv);
  • not-so-common good formats (xml, sql, json, shp, kml, ...).

If we are not so lucky:
  • tables or lists in web pages (html);
  • simple tables in well-done pdfs (pdf).

If we have Murphy as worst enemy:
  • scanned images, even if in a pdf wrapper (png, jpg, pdf);
  • digital data behind complex search engines.

And if we have the best data ever, but under closed license?

Washing & cleaning

Well-formed data sets

Numbers are numbers, strings are strings and not numbers, datetime must always have a single format (ie. yyyy/mm/dd), localization is important, no gender values in names' column or similar mixings, every elements should be named with a Unique Identifier (ID).

Data types computer understands:
  • integers (with sign, zero included),
  • floating numbers (with sign),
  • datetime,
  • characters and string (case sensitive),
  • null value (the strange case of a value that states "I'm not a value").


And simple comparisons are strictly equalities, also in strings!

Data management and analysis

"If you torture the data enough,
nature will always confess" - Ronald Coase (1981)


Descriptive statistics

Aggregation, average, normalization, relative difference, distribution, ...

Correlations

A single rule: correlation does not imply causation!

Data visualization

At a glance


Data visualization

With great power comes great responsibility


The basic idea is quite simple: you have quantities expressed in numbers and geometric objects defined by dimensions (ie. radius in a circle), so you just have to decide how connect your quantities to visual dimensions.

There are several (un)common charts and endless combinations: scatter plots, lines, bars, areas, pies, donuts, bubble charts, treemaps, word clouds, alluvional diagrams, dendrograms, networks, streamgraphs, gauges, chord diagrams, motion charts, parallel coordinates, sankey diagrams, maps, choropleth, ...

On d3js.org gallery there is an endless list of examples!

Journalism as public utility


Building a simple dataset or a large and complex database focused on a topic of public interest leads to a valuable product: the database itself, intended as a collection of (linked) data plus metadata.

Can a public frontend to such database, designed for citizens, journalists, stakeholders, be considered a journalistic outcome? If journalism is a public good, it can be a service, not only a product...

"Open journalism is journalism which is fully knitted into the web of information that exists in the world today. It links to it, sifts and filters it, collaborates with it and generally uses the ability of anyone to publish and share material to give a better account of the world" - The Guardian

Tools


Readings


That's all...?



It's your turn, folks!

Hacking + Marathon = Hackathon

ESPAD (European students and drugs):  http://www.espad.org/en/

RASFF (EU food safety):  http://ec.europa.eu/food/food/rapidalert/ 

Rapid Alert System

for Food and Feed


The Rapid Alert System for Food and Feed (RASFF) was put in place to provide food and feed control authorities with an effective tool to exchange information about measures taken responding to serious risks detected in relation to food or feed. This exchange of information helps Member States to act more rapidly and in a coordinated manner in response to a health threat caused by food or feed.

European School Survey Project on Alcohol and Other Drugs


This is the report from the fifth data-collection wave of the European School Survey Project on Alcohol and Other Drugs (ESPAD). It is based on data from more than 100,000 European students. Over the  years about 500,000 European students have answered the ESPAD questionnaire.  A total  of 36 countries and regions have contributed data to the 2011 ESPAD Database. Drugs list includes cigarettes, alcohol, cannabis, other illecit drugs, tranquillants and sedatives without prescriptions.

When data journalism meets science

By Dataninja srls

When data journalism meets science

Introductory lesson about data journalism within science journalism and science communication during the International School of Science Journalism 2014 in Erice (June 10th, 2014).

  • 6,800
Loading comments...

More from Dataninja srls