Econometrics — the art of pulling a rabbit out of a hat
from Lars Syll
In econometrics one often gets the feeling that many of its practitioners think of it as a kind of automatic inferential machine: input data and out comes causal knowledge. This is — as Joan Robinson once had it — like pulling a rabbit from a hat. Great — but first you have to put the rabbit in the hat. And this is where assumptions come in to the picture.
The assumption of imaginary ‘superpopulations’ is one of the many dubious assumptions used in modern econometrics, and as Clint Ballinger highlights, this is a particularly questionable rabbit pulling assumption:
Inferential statistics are based on taking a random sample from a larger population … and attempting to draw conclusions about a) the larger population from that data and b) the probability that the relations between measured variables are consistent or are artifacts of the sampling procedure.
However, in political science, economics, development studies and related fields the data often represents as complete an amount of data as can be measured from the real world (an ‘apparent population’). It is not the result of a random sampling from a larger population. Nevertheless, social scientists treat such data as the result of random sampling.
Because there is no source of further cases a fiction is propagated—the data is treated as if it were from a larger population, a ‘superpopulation’ where repeated realizations of the data are imagined. Imagine there could be more worlds with more cases and the problem is fixed …
What ‘draw’ from this imaginary superpopulation does the real-world set of cases we have in hand represent? This is simply an unanswerable question. The current set of cases could be representative of the superpopulation, and it could be an extremely unrepresentative sample, a one in a million chance selection from it …
The problem is not one of statistics that need to be fixed. Rather, it is a problem of the misapplication of inferential statistics to non-inferential situations.
Leave a Reply
—– look inside —– $5.93 / 14.99
—– look inside —– $15.99
Now available as a Kindle – $4.99
—– look inside —– $5.93 / $12.99
—– look inside —– $5.99 / 12.99
—– look inside —– $5.94 / $20.00
—– look inside —– $4.97 / $9.90
—– look inside —– $5.99 / $11.00
—– look inside —– $9.99
—– look inside —– $4.90 / $8.00
—— Ugarteche, Puyana and Madi ——
Gerson Lima / Maria Alejandra Madi
Edward Fullbrook and Jamie Morgan
————— Michael Hudson ————–
Maria Alejandra Madi / Jack Reardon
————- Edward Fullbrook ————-
—————— Steve Keen —————–
————— Richard Smith —————
————– Gustavo Marques————
– Victor Beker and Beniamino Moro –
————– Lars Pålsson Syll ————-
—————– Stuart Birks —————-
Edward Fullbrook and Jamie Morgan
———— Armando Ochangco ———-
Shimshon Bichler / Jonathan Nitzan
————— Mauro Gallegati ————–
————— Herman Daly —————-
————— Asad Zaman —————
—————– C. T. Kurien —————
————— Robert Locke —————-
Econometrics generally apply to reductive/selective research and observations. True and actual paradigm changes however are always characterized by conceptual opposition, inversions of temporal reality and entire pattern changes…as that is what a paradigm is, a mental and temporal pattern change.
Paradigm changes are the closest thing to magic that we can experience short of what is referred to as cosmic consciousness, which experience is simply a direct integration of the normal egoistic consciousness with a higher level of awareness of the electro-magnetic milieu that is the present moment.
At one time epidemiologists focused on the individual biological and behavioral risk factors in the spread of infectious diseases. The era of ‘do not visit prostitutes if you do not want to contract an STD.’ Now they tend to focus on the social context of risk and vulnerability to infectious diseases.
Proximate behavioral risk factors that shape infection, occurrence, and severity of the disease include individual and group performance related, for example, to hygiene and sanitation, sexual behavior, food consumption, or movement (including travel, migration, and displacement). But social scientists emphasize the ‘cultural logic’ underpinning individual behavior and argue that behavioral patterns, and, consequently, exposure to, and distribution of infectious disease risk, are the expression of larger-scale forces such as poverty, social inequalities, armed conflict, and other shared ways of life. For social scientists, basic notions such as ‘risk group,’ ‘patient compliance,’ and ‘community’ do not adequately grasp the complex cultural reality of populations. Consequently, among social scientists, the social construct of vulnerability has widely replaced the epidemiological construct of ‘risk,’ providing a direct link to how those who might become infected construct their situation. Providing a background for practical interventions that try to consider how specific social contexts influence individual identity-constructions, while avoiding the risk of group discrimination.
Vulnerability to infectious disease results from several major overlapping socioeconomic, biological, and environmental factors. Macro-level social processes such as globalization and trade liberalization, unplanned rapid urbanization, widespread poverty, and inequalities lead to vulnerability for some but not every person possible.
Epidemiologists have been quite successful in this work. One of the reasons infectious diseases are better understood and better controlled in terms of human impacts than at any other time in human history. In this work, epidemiologists employ descriptive statistics and qualitative models. Inferential statistics is not used for precisely the reasons pointed out here.
An essential aspect of the use of mathematical models in infectious disease epidemiology is validation of model results against real data, validation here referring simply to the ability to satisfy oneself that the model results are consistent with the available data relating to the population which is being modeled (or in its absence, data from a population sharing similar characteristics). It is not necessary for this process to be one of fitting, in the statistical sense, as we are dealing with model results which are qualitative rather than quantitative, and it is usually therefore more important that the model can reproduce the change in shape over time of the data being used for validation rather than absolute values. Thus, the process is more one of inspection and fitting by eye rather than statistical fitting. This involves work with a specific population (e.g., residents of New York City).
Given that we can establish the validity of the model results in this way, it is important to know how sensitive these results are to the values of the model parameters. Depending upon the structure of the model and the role played by each parameter in the model, small changes in the values of some parameters can lead to large variations in model results (infection rates); for other parameters the situation may be reversed and relatively large changes in values may have little impact on the results. It is the role of sensitivity analysis to establish how variation in parameter values might impact on model outputs; if the model should prove to be too sensitive, the results will be useless. This can either be done simply by varying the values of each parameter in turn and observing what effect each has, or more systematically using “Latin hypercube” sampling (cf. Latin square), which involves specifying a distribution for the values of each parameter (which could simply be a uniform distribution if the true distribution is not known) and sampling at random and without replacement parameter values from the combined set of distributions (Seaholm et al., 1988). Once the sensitivity of the model to its parameters has been established, it is necessary to consider how much ambiguity there is in our knowledge about the true values of each parameter in relation to the population being considered. In uncertainty analysis we are considering what the plausible range of values for each parameter might be and exploring the variation in model results which arises when parameter values are varied within this set of ranges; the variation in model outputs then provides an indication of uncertainty about the likely way in which, for example, prevalence or incidence of an infection may change over time in the population being studied.
I have used this framework for years in epidemiological and crowd studies. It seems to provide useful results without a high error rate (e.g., 5-10%).