Skip to main contentSkip to solution

As crises continue to erupt worldwide, the ability to address their adverse effects remains elusive. While the jury is still out on best practices for anticipating and responding to crises, the growing salience of what has been dubbed as ‘evidence-based policymaking’ has led to an increasing demand for data-driven analyses (DDAs). This is arguably because statistics are usually deemed apt at capturing, and processing, the high-frequency nature of crises. DDAs can provide a digestible way of measuring the temporal evolution and spatial distribution of crisis drivers like social unrest, macroeconomic imbalances, armed violence, food insecurity and population displacement. Through graphical representations, numerical data can convey otherwise complex information that usually requires pages of text to summarize. A visual can sometimes be worth a thousand words. But while the quantitative approach has its notable advantages, data is often taken at face value, usually creating the illusion that statistics are unquestionably reliable. Because of the hazardous and even politicized method of collecting and disseminating information across crisis-affected locations, data must always be consumed with caution. Indeed, data is not impervious to errors, bias, misrepresentation, or manipulation. When dealing with issues that require contextual knowledge, data can only tell us so much. For this reason, DDAs should be envisaged as complementary tools to qualitative insights. This is particularly true in crises, where widespread disruption can cause much confusion due to the multiplicity of events taking place simultaneously, some of which cannot necessarily be quantified. When analyzing different datasets, careful consideration must however be paid to the caveats that data providers tend to identify, notably regarding the process of collecting and coding quantifiable information on crisis events. This is especially important when the sourcing of the information is incomplete, contested or even manipulated. In turn, poor data limits the potential to develop more ‘sophisticated’ forms of DDAs, notably the ones that are growing in popularity and that aim to make use of Artificial Intelligence. But so long as crisis data remains largely small in sample size, DDAs in this realm will remain limited as large samples are required for statistical validity.

To make DDAs more reliable, we need:

Solution

✅ Correct Option: 1

Question asks: What is needed to make DDAs more reliable? So we are looking for the step to take, not just a description of the goal.

From passage: every reason the passage gives for unreliable DDAs:

→ "hazardous and even politicized method of collecting ... information across crisis-affected locations"

→ caveats "regarding the process of collecting and coding quantifiable information"

→ sourcing that is "incomplete, contested or even manipulated"

→ crisis data that is "small in sample size"

All four problems happen while the data is being gathered: unsafe places, political pressure, weak sources, too few data points. Fix the collection, and the data (and so the DDAs) becomes reliable.


Checking the options:

Reliable data collection → CORRECT, the one stage the passage names again and again as the source of the problem.

Reliable data → trap. DDAs are data-driven analyses, so "reliable data" is just another way of saying "reliable DDAs". It describes the goal, not the step that achieves it. The question asks what we need to make DDAs reliable: the passage's answer is better collection.

Reliable data dissemination → mentioned only once, joined with "collecting", and the danger described there is about gathering information on the ground. None of the detailed concerns is about sharing data.

Reliable data measurement → the passage never says the way crises are measured is faulty; the problem is the quality of what is collected.

Hence, the correct answer is Reliable data collection.

Keyboard Shortcuts

  • Left arrow: Previous question
  • Right arrow: Next question
  • S key: Jump to solution
  • Q key: Jump to question