In light of the current COVID-19 pandemic response, researchers around the world have been evaluating ways to support all aspects of disease identification, monitoring and tracking. The idea of using audio-based processing methods to evaluate cough events, one of the most common symptoms of COVID-19, in terms of their frequency, severity and characterization has become a promising possible solution. In addition to physical distancing measures, the vast majority of the health authority also recommends the adoption of face coverings (i.e. masks) while in the presence of others and covering one’s cough with a bent elbow. The covering of cough events may present an issue when evaluating recordings using pre-existing cough analysis tools. This paper presents a modeling approach used to characterize the effects of both coughing while wearing a mask and coughing into a bent elbow. These two models were then applied to an existing dataset for evaluating the influence of the face coverings on selected data features that have been used for differentiating wet and dry cough types. It was found that one of the features (number of peaks in the energy spectrum) did not change after mask and elbow modeling, however the second feature (power ratio) was greatly affected and was unable to differentiate between the cough types. The application of these models are therefore recommended when using classification tools that were designed using uncovered clear cough sounds in order to ensure that they will be robust to the presence of face coverings.