I would expect the longer incubation times are a factor of lower initial viral load. If the ID50 is 1-10 viral organisms, starting with that lower spectrum of load in the patient it will take longer for the virus to replicate and disperse throughout the body. If the initial viral load consisted of millions of viral organisms, what was mentioned as being the viral load in droplets for instance (500,000 to 1,000,000 organisms) could constitute a significant advantage for the virus and a 'leg up' resulting in shorter incubation.
I would suspect the route of infection to make a difference: through a wound vs. droplet inhaled or ingested vs. possibly transdermal contact (sweaty patient/infectee), vs. (possibly) fomite contact.
This is the end of the spectrum we least understand: it is the end where how little it takes to infect someone matters, and the route of infection may not be clear.
Ahem...Dallas is an urban area, as were the various places in Ohio and elsewhere contacts of Ms Vinson visited.
We can say the data is bad, but we can't say anything more.
For “Bad how?” see Smokin’ Joe's post above.
We don't know and the public health people who should be asking are too deep into “Risk Messaging” cant to want to go there.
At least absent of a looking down the barrel of a public riot dumped on their collective heads.