Economist Zvi Griliches wrote the following paragraph in the third volume of The Handbook of Econometrics in 1986:
Econometricians have an ambivalent attitude towards economic data. At one level, the “data” are the world that we want to explain, the basic facts that economists purport to elucidate. At the other level, they are the source of all our trouble. Their imperfection makes our job difficult and often impossible. Many a question remains unresolved because of “multicollinearity” or other sins of the data. We tend to forget that these imperfections are what gives us our legitimacy in the first place. If the data were perfect, collected from well designed randomized experiments, there would be hardly room for a separate field of econometrics. Given that it is the “badness” of the data that provides us with our living, perhaps it is not all that surprising that we have shown little interest in improving it, in getting involved in the grubby task of designing and collecting original data sets of our own. Most of our work is on “found” data, data that have been collected by somebody else, often for quite different purposes.
Our goal at Better Data Initiative is to try to create an Asymptopia, where data are perfect and abundant for social science research.