Confounding and control

Abstract

This paper deals both with the issues of confounding and of control, as the definition of a confounding factor is far from universal and there exist different methodological approaches, ex ante and ex post, for controlling for a confounding factor. In the first section the paper compares some definitions of a confounder given in the demographic and epidemiological literature with the definition of a confounder as a common cause of both treatment/exposure and response/outcome. In the second section, the paper examines confounder control from the data collection viewpoint and recalls the stratification approach for ex post control. The paper finally raises the issue of controlling for a common cause or for intervening variables, focusing in particular on latent confounders.

The perception of colombians about science and technology according to their education level: professional and non-professional population

“…Some demographic variables such us age, affect the perception of humans in many issues [25,26], in consequence and for the purposes of comparison needed in the research, the idea presented in the methodology section, about using a subsampling technique to try to reduce the possible bias that could be present due to the two dissimilar age structures found at the beginning (see Table 1), was proved to be convenient: Maintaining the population as it was in the preliminary sampling, would not allow showing the effect of age in the opinion, neither the effect of the technological advances along with education, creating what is called a "possible confusion" or "confounding" [27]. The idea of "matching" the two groups of population through a re-sampling exercise and therefore avoiding the bias or the possible confusion was very accurate and improved the validity of the assumptions made.…”

Statistical inference, exchangeability and multilevel approach

“…Cet article a permis de montrer comment le paradoxe de Simpson pouvait être correctement résolu en utilisant un modèle multiniveau adéquat [Wunsch, 2007] [Lindley, Novick, 1981 ;Courgeau, 2004(a)]. Mais le fait que l'échangeabilité se réfère explicitement à des variables aléatoires, alors que le concept de population ne peut s'y référer, enlève une partie de l'intérêt de l'approche suivie par Fisher.…”

Inferring Causality through Counterfactuals in Observational Studies - Some Epistemological Issues

“…In order to assess the causal effect of education on self-rated health, we should control for possible confounders, taking into account the heterogeneity of the population. From the existing literature, we know for example that social support and physical health could confound the relationship between education and self-rated health as they have an impact on the latter and are also associated to the former (see Wunsch 2007 for a discussion of confounding). Social support and physical health should therefore be added to the model, albeit as control variables.…”

“…it assumes that there are no latent confounders influencing the assignment of units between the treatment and control groups (for a discussion of the concept of confounding see e.g. Wunsch 2007 ). This requirement is less demanding in experimental studies where the units are randomly assigned to the treatment and control groups.…”