Not everything is a number | Statistical data

What do public statistics measure in Santa Fe and Rosario, why do the data complement direct experience, and what recommendations does the report propose? Report from September 2016.

A regional overview of what is being measured, what we can learn about ourselves, and what the state of statistics is like

Yesterday, the Argentine Chamber of Deputies approved the law on access to public information with 71,1% of the vote. This law affects national government agencies and obligates the various branches and institutions of the national government, state-owned companies and corporations, concessionaires and licensees, and entities that receive public funds to comply with the law.

Although not yet fully implemented, the regulations require the listed departments to respond to information requests submitted by any citizen or institution within 30 days. The regulations also include a system of exceptions, the most notable being the exemption for social welfare organizations from providing information.

In the report available for download, we review the current state of regional statistics (province and municipality). We then discuss their importance, both for decision-making and in relation to development. Finally, we conclude with a series of recommendations that, at first glance, appear easy to implement.

📄 Download the report (PDF)

The title of this article is not accidental. Common sense tells us that economists tend to reduce everything to a number. It's worth clarifying that, depending on the context, this criticism can be valid. But other times it isn't. For those who are hesitating about downloading the report, we offer a defense below. We hope to find it interesting and—above all—to improve the debate.

Not everything is a number

Is measurement really important? Everything seems to indicate that it is: well-understood statistics provide an unbiased overview that firsthand observation cannot. However, to achieve this overview provided by data, the process sacrifices a wealth of concrete experience regarding the specific problem being studied.

An example: the volunteer and their biases

To understand the above, let's look at an example. Imagine a scenario where someone decides to do volunteer community service for an NGO working in a shantytown. Also imagine that this individual comes from a home where the socioeconomic situation is radically different from the place where they will be doing their community service.

In this situation, biases will naturally be revealed on the part of the volunteer, as a result of the contrast of the mental model that he formed in another context of greater opulence. That is, the contrast between the situation at home and the situation of the place where he is carrying out his volunteer work.

When this occurs, two types of (pure) responses can arise: either the force of preconceived ideas prevails and the individual abandons the activity, or a gradual adjustment occurs as a result of the new information. In other words, a learning process is triggered.

But the story doesn't end there, because the creation of a new bias is inevitable. According to the analytical framework of behavioral economics, all individuals reduce the processing of complex tasks to simpler operations through a feature substitution mechanism called a heuristic.

When an individual evaluates a specific attribute of an object (in our example: the relationships of a family in poverty), replacing it with some other property of said object that is easier for him to interpret (the relationships established in his own home), he is operating a heuristic.

In this way, the processing costs that are avoided by giving a quick response to the situation or problem that arises reappear in the form of serious and systematic errors, called biases.

From learning to a new bias

Understandably, this is not in any way an assertion that one mental model is ethically superior to another. Making such a comparison would be a mistake (one that is made very often). At most, certain mental models, in certain contexts, work better operationally to achieve a specific goal.

And this is one of the issues that public policy planning must address: Why is what is being done the way it is? And how can it be done in a socially desirable way?

But let's go back to our example. At this point, the volunteer has persevered, made an adjustment, and thus triggered a learning process. The concrete experience of poverty in the shantytown is now rich in nuances, and the level of interpretation that person now has about the problem is definitely more complex than his previous level.

Now imagine our volunteer, sitting at the family table with lifelong friends. None of them have ever worked in a shantytown. During the conversation, the topic comes up, and our volunteer speaks at length about the problems of poverty, the direction the country is taking, and—with good intentions—how to solve them. But here's where a problem arises: a new bias has emerged.

Extrapolation of a specific experience probably explains part of the general problem being addressed, but it does not provide an overall view. At least not at the country level.

What are statistics used for?

This is where measurements become essential. Not as a substitute, but as a complement. Statistics provide a comprehensive view because they are representative of a population. Furthermore, they offer a suitable framework for comparing situations, both over time and across different regions.

On their own, they don't provide a complete picture either, because each number—each indicator—needs interpretation. And a nuanced interpretation is definitely shaped by experience (and experience by some kind of theory).

Furthermore, another advantage provided by statistics is that they delimit the field of discussion of the different interpretations that contribute to the explanation of a fact, making the process of agreement between different actors less arduous or, in the worst case, preventing it from dissipating into inactivity and a sterile debate.

It is clear that in the social sciences, no fact of reality can be measured with absolute precision. Behind every measurement tool lies an intention to achieve precision, and a system of checks and peer review to ensure its validity. But the result will always be probabilistic.

In other words, there will be approximations and, of course, there will be errors. But so far, this is the best that can be offered from the state of the art (and that is no small feat).

Open access, for use and citation

The report is available freely and without charge, under a Creative Commons license (CC BY-NC-SA 4.0):

No todo es un número. Informes del Observatorio UNR N.º 29 — Serie Informes de Coyuntura N.º 11, septiembre de 2016. ISSN 2683-9067 / 2683-9075. Autores: Germán A. Tessmer, Luciano A. Jara Musuruana, Patricio H. Almeida Gentile y Norberto Martín.