"The U.S. has monitored weather conditions somewhat systematically since the late 1800s and the National Center for Environmental Information (NCEI) archives these observations.
Scientists at NCEI identified a set of stations with long- term observations that are useful for answering questions like that posed above.
These stations form the United States Historical Climate Network or USHCN.
The USHCN stations are still archived, though the network does not formally carry the status it originally had.
These stations reported the highest (usually occurring in late afternoon) and lowest (usually around sunrise) temperature for each day.
These values are often referred to as the daily extremes and denoted as TMax and TMin respectively.
A subset of these stations having at least 105 years of data from 1895 to 2020 were examined for the question posed. There were 737 (738 for TMin) stations which met the criteria.
...The results indicate some clear changes have occurred since 1895.
... The 1930s stand out. In fact, of the 15 years with the most record high TMax values, 14 occurred before 1955 and six in the decade of the 1930s alone.
Record hot days are not occurring more often than before.
The story for the coldest nights is more substantial regarding long term change.
Here we see a distinct and highly significant downward trend with fewer and fewer stations recording record low values of TMin since 1895.
Of the 15 years with the fewest record values, 14 occurred just since 1997.
This combined situation - record hot days are trending slightly downward (thanks to a hot 1930s) while record cold days are almost disappearing
The answer is no, they are in fact decreasing – and by a significant amount ... "