Kindof a side-bar — In most cases i strongly prefer median, any dataset with outliers or that is skewed in some direction is usually better represented by the median. And in my experiences most real-world datasets do have one or both of those.
All news/public relations/research publications without details about the distribution of the data are clickbait. It would cost nothing to provide deciles, or even quintiles.
That still would not really clarify the main question, which is whether the average transaction price may be rising due to consumer shift, with those of less means choosing to not buy.
