The Question consists of three parts
1: Decline of insect populations.
Is there any substantial and robust decline of insects in a large enough space and timespan to consider it a general decline?
The overall abundance of butterfly species in Great Britain has declined by approx 70% in 20 years. http://science.sciencemag.org/content/303/5665/1879/tab-figures-data
Other insect groups have declined as well in biodiversity, however not that much as butterflies. http://science.sciencemag.org/content/345/6195/401.full
This just means there is a decline of species richness in some key groups, which while alarming, does not mean population decline in general. This study found over a 30- year period a 75% decline of insect biomass in various sampling locations in Germany, however. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0185809#sec001
I could not find follow up studies or similar ones, yet this study seems solidly done and over a large area and timespan account to for variability. So yes, there seem to be a general decline in insect populations.
2. Are there fewer insects smashed on car windshields?
Is there substantial evidence for a sufficiently large enough area and timespan?
There is no evidence to compare insect numbers on windshields. BBC did an experiment covered in Wikipedia, which is a good summary of the "Windshield Phenomenon". https://en.wikipedia.org/wiki/Windshield_phenomenon. This experiment is however only of anecdotal value. Humans are highly biased, maybe we remember just those days when our windshield was heavily covered in stains?
So far, inconclusive.
3. Does the population decline cause fewer smashed insects?
As we have seen in 2, it is inconclusive if there are really fewer smashed insects. However to prove this causality could be difficult anyways. We would need the numbers of smashed insects of areas that do not have populations decline at least for a correlation. There are a lot of factors to control for, too. But maybe we could show a correlation if the difference is pronounced enough.
Overall, the answer to the question is a NO in the sense that it is not valuable to give an answer as there is no evidence for one of the presumptions.