Hello! Recently, as I was reading a paper, I came across the term kernel-weighted local polynomial smoothing. The regression model itself was a very simple logit model. Can anyone easily explain what exactly that technique is?.. Here is an excerpt from the paper:
We construct Ethnocentrism(j,t) by standardizing survey responses such that 1 reflects pro-immigration sentiment and 0 reflects anti-immigration sentiment. We use kernel-weighted local polynomial smoothing to aggregate the anti-immigrant sentiment for each state year. This technique attenuates measurement error arising from different sample sizes across states and combining responses across different questions.