(Q should be moved to Products methinks.)
The strictest answer is that there is no difference in presumed spamminess between an original email and one that claims (via the References: header or the "poor man's threading" of a common subject line) to be part of an ongoing conversation. Such characteristics are easily forgeable and so cannot be used to confer legitimacy.
However, the wider answer is that if, by virtue of being part of an actual conversation, an incoming email is a response to a earlier outgoing email to the same address, then an intermediate mail scanner may have auto-whitelisted (or auto-counterweighted) the sender based on previous outbound communications. Determining whether said outbound was legitimate is a little tricky since you have to exclude automated responses. But it is certainly done. A scarily long time ago I wrote a utility called AUTOWHITE that did exactly this, for the then-popular IMail SMTP server.
So the overall answer is that being part of a real convo can be helpful for transmitting mail. Not so much because of "threading" but because of prior behavior.