skuggvariabel

A skuggvariabel, or shadow variable, is a concept used in statistical modeling, particularly in the context of instrumental variables or when dealing with unobserved heterogeneity. It represents a variable that influences both the treatment or exposure variable and the outcome variable, but is itself unobserved by the researcher. The presence of a skuggvariabel can lead to omitted variable bias, making it difficult to establish a true causal relationship between the observed treatment and outcome.

In instrumental variable estimation, a skuggvariabel is often the reason why a direct observational study might be biased. An instrumental variable is a variable that affects the treatment but does not directly affect the outcome, except through its effect on the treatment. The existence of a skuggvariabel that affects both the treatment and outcome, and is correlated with the instrument, can invalidate the instrument.

The term is also relevant in fields like econometrics and epidemiology. For example, if researchers are studying the effect of education on income, intelligence could be a skuggvariabel. Intelligence might influence an individual's level of education and also directly contribute to their earning potential, independently of their educational attainment. Without accounting for such unobserved factors, the estimated effect of education on income might be biased.

Addressing skuggvariabler often involves employing methods like instrumental variables, fixed effects models, or difference-in-differences, which aim to control for or isolate the effects of unobserved confounding factors. The goal is to approximate a randomized experiment where the skuggvariabel would, in theory, be balanced across treatment groups.