Hi all,

This question is in regards to options for analysing a dataset. I have a dataset where we have 23 video clips that have been treated using three different algorithms to change how each clips looks. Each treatment has then been viewed by a number of scientists to try and identify the species in each clip. These reviewers were asked to classify the species in each clip at the order, family, genus and species level and provide an associated confidence level (%) for each classification - see example below.

So: - 23 clips - 4 treatments per clip (3 algorithms plus raw clip) - 5 reviewers - 0-100% confidence for each classification (very high number of possible classifications)

I am way out of my depth on this one and would be very grateful if someone could point me in the right direction to making sense of these data. Our question is; do any of the treatments increase confidence levels associated with classifications over the raw footage? And is there less variation in classifications at the order, family and genus level across all 5 reviewers for processed or raw footage?

Am I right in assuming that these two questions will need to be addressed using two separate analyses?

Any assistance would be gratefully recieve