In their paper titled Predicting Material Accounting Misstatements (2011), Dechow et al. define the unconditional expectation of misstatement as "equal to the number of misstatement firms divided by the total number of firms". They calculate it for their sample as 0.0037.

If I am applying the F-score on a different sample in a different country, should I still apply the unconditional expectation of misstatement of 0.0037 or should I update it to use a country/sample-specific unconditional expectation?

As my sample is not based in the USA, it made sense to me to update it, however, studies by Hang, Ha and Binh (2017) argue for the use of the 0.0037 in Vietnam as it was collected over a long time and thus is likely to reflect reality. My country is generally considered to have higher levels of manipulation compared to the USA.