I have found that Dynamo generally hates performing analysis on mixed data types. By converting only some of the data to a Boolean (true/false) you create this mixed data set. This is similar to when a dataset contains null values or an empty list. Sometimes I have been able to get around this with lacing and level manipulation, but that is dicey to say the least. As @Dimitar_Venkov pointed out, simplifying your code to perform the function on the data while it is a single type is certainly the simplest method here.
Another good test here is to ensure the correct data is being fed in before you proceed. Run the data into a List.UniqueItems node and then into a few other tests or List removal functions to produce a single Boolean for ‘all data is matching types’. That Boolean can then cause Dynamo to throw up a popup to say something like ‘The given dataset has unsupported values. Please review the dataset and try again.’