You should know that imports are not universally welcome in OSM; imports are not impossible but they are not our usual way of data acquisition. OSM values the work of its volunteer contributors highly, and would not usually replace or preempt such work through a data import. Roughly speaking, an import is a social as well as a technical challenge. You will first have to build a consensus about whether OSM wants the data imported at all, and if that happens, then you will have to think about how to import the data without damaging or duplicating that which is already there. (Simply deleting all existing road centerlines in Arkansas and replacing them would be totally unacceptable to the community.)
On a "whole state" scale, even the preparation for such an import would very likely take several months.
You will find more about the topic of imports if you type "import" in the search box above and read some of the responses, or subscribe to the [talk-us mailing list](http://lists.openstreetmap.org/listinfo/talk-us) and raise the issue there.
It is usually easier, and certainly less of a contentious issue, if instead of importing a data set into OSM outright, you make it available in a form that our mappers and other volunteers can then use; this might be a WMS from which to trace missing bits or correct those that might be wrong, or it could be a publicly accessible shape file that volunteers can then automatically match against OSM and thereby spot differences. Not every such difference would necessarily point to an error in OSM, and the best way to deal with them would be to survey the area or at evaluate suitable aerial imagery to find out who's right.
