With recent changes weakening moderation and fact-checking policies on major social networking sites, understanding the threat of misinformation to democratic societies is more crucial than ever. Previous research has found a link between political attitudes and the sharing of misinformation, especially the congruence between the reader's political attitudes and the message's content. These studies have almost exclusively focused on sharing misinformation on public channels like Twitter/X or Facebook, while personal messaging apps like WhatsApp have received less attention. In addition, little is known about how partisans interact with misinformation beyond sharing it, in part because measuring such behaviour on private channels is challenging. We developed a novel method for measuring interactions with misinformation on WhatsApp. This method allows us to capture not only sharing intention but also genuine sharing, flagging, and reporting behaviour. We apply this method to gather data from an online field experiment with partisan news articles in the U.S. The analyses show that while there are no differences in intention to share between congruent and incongruent messages, actual behaviour shows a clear pattern: incongruent misinformation is shared less than congruent misinformation. Similarly, incongruent misinformation is more likely to be flagged as inaccurate and reported as offensive. Our study establishes that political congruence affects sharing, flagging as inaccurate, and reporting as offensive, even though focusing solely on an intention measure would not reveal this. These findings imply that careful measurement is needed to unpack the link between intent and behaviour.