The concept is that while you can't do so directly, the degree to which you value these things can in principle be quantified monetarily by the monetary value of the things you can buy with money that you are willing to forego in order to attain the things that cannot be bought with money, allowing money to be used as a tool to quantify utility even for things which it can't buy directly.
Its certainly consistent with preference being transitive (since it is intimately tied in with cardinal utility, which is a much stronger position which includes transitive preference.) Transitive preferences in a poset aren't sufficient to support money-as-general-utility-scorecard, however.
I meant that it isn't consistent with people's observed preferences being transitive; if they are, then there is no need to make the inferences you suggest. But of course you were talking about things that can't be traded for money, not things that people refuse to because they value them more, so ignore my criticism.
Actually, that makes me realize that I need to change the edge criterion in my previous comment.
A principle assumption of many of the models underlying economic theories (including, but not limited to, rational choice theory) is that there is a unidimensional quantity, "utility", that aggregates a persons preferences and which individual decisions increase or decrease by definable amounts. Because money is widely fungible, and utility cannot be measured directly, money (e.g., the money people are willing to pay to get something they want, or the money they demand to do or give up something) is often used as a proxy for utility by economists, with the idea that all decisions have an (potentially zero) utility impact, and that utility impact can be stated in monetary terms.