When displaying and combining spatial datasets, consider
1) the scale for which each dataset is valid, and
2) the scale of the map product you are viewing.
In this example, the road data isn't wrong in itself but rather it is being used well beyond its intended scale as well as the scale of the NAIP image (National Agricultural Imagery Program). This is similar to zooming into a digital photo beyond its intended resolution and revealing only pixilated blocks.
Dataset Scale Accuracy
GIS Road data 1:100,000 50.8 meters
Background image 1:3,780 5 meters
Map display 1:3,780 5 meters
Consider scale when displaying any data and combining differing datasets.
For example, when collecting a particular trail amidst desolate rural miles, 'close enough' with a recreational grade GPS is sufficient for navigation. Yet, how will that same data appear when displayed atop imagery that is accurate down to 5 meters?
...now that we've covered scale, what about datums and coordinate system?