Measurement Error Idea Applications

ijyliu / ECMA-31330-Project

Econometrics and Machine Learning Group Project

2 stars 1 forks source link

Measurement Error Idea Applications #10

Closed ijyliu closed 3 years ago

ijyliu commented 3 years ago

Brainstorm applications here.

ijyliu commented 3 years ago

What if there's measurement error in defense spending?

The problem is that there isn't really a dependent variable, or other independent variables to build a factor from.

ijyliu commented 3 years ago

@marionoro thinks that instability will be tricky to use as a variable for this idea

ijyliu commented 3 years ago

What are things that are mismeasured a lot?

This page seems to be getting at latent items/loose concepts such as IQ, so maybe we want to not do that sort of area... maybe look for something concrete, but prone to measurement error. https://en.wikipedia.org/wiki/Structural_equation_modeling

Political opinion/ideology (maybe in the defense spending context) seems a bit like a 'loose concept'. Though maybe we could look at ideology's relationship with defense spending?

ijyliu commented 3 years ago

Things prone to measurement error:

Income, especially for high incomes. I just noticed the Forbes billionaires data is on kaggle if we want to use that. I think there are also income and happiness datasets out there? Though that seems like measurement error both in independent and dependent variables...
GDP: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwifx7_vvbjwAhUDZc0KHdoeBq4QFjAKegQIHBAD&url=https%3A%2F%2Fwww.aeaweb.org%2Fconference%2F2018%2Fpreliminary%2Fpaper%2FZfeTyfYG&usg=AOvVaw2-g6SKvkrspI1-4LhhT7G8 consider the informal economy for example
Wages?
Survey data about sensitive topics (ie birth control, domestic violence, child abuse, etc.)
Military size or defense spending itself? https://data.world/stuartltinsley/military-dataset. Would still need a dependent variable.
Unemployment rates? Or, at least, the methodology varies between countries... Maybe we would want to look at a phillips curve type relationship with prices too.

ijyliu commented 3 years ago

I think we need something for which there are a lot of different measurements that we can combine into a factor

ijyliu commented 3 years ago

We can do GDP and anything as the dependent variable.

Coronavirus? Lol

ijyliu commented 3 years ago

@marionoro 's idea of gdp and life expectancy: results here: https://github.com/ijyliu/ECMA-31330-Project/blob/main/Release/Factors_and_Measurement_Error.pdf

ijyliu commented 3 years ago

@marionoro A problem with the World Bank result is that the PCR regression is CLOSER TO ZERO, which is the opposite of what we would expect if our method was reducing attenuation bias due to measurement error

ijyliu commented 3 years ago

Life expectancy and cigarettes per capita and economic indicators as the covariates?

Life expectancy on healthcare systems with economic indicators as covariates?