Two Group, Two Measures, and Ordinal Data- Wilcox it?

I'm proposing the following methodology (a cross-over design variation?), consisting of a sample of respondents equally and randomly assigned to one of two groups (i.e., video materials or written materials). The participants will then complete a ranking survey task and ranking selection-task (counter-balanced order), both measuring the same ordinal DV. (see image; sorry for the large size). Diss_Design.jpg
We're trying to determine, do rankings vary between a survey-ranking versus a selection-ranking task; and do rankings differ between video and written groups. We're trying to determine 1) does the measurement task matter (survey versus selection-tasks) and 2) does presentation format (video or written) matter?

I believe the above design would answer these questions, but I'm not solid on the statistical tests to run. I assume I'd run a Wilcoxon signed-rank test, but can this be used when the groups are spilt between written and video? Does the group assignment feel like it would mess with the assumptions of the W s-r test? Should I be considering GEE, OLS, or something instead? Any thoughts would be helpful. Thanks!