For the most part you are safe assuming that crystal malt will ferment the same as your base malt, especially considering that the crystal will be a relatively small portion of your grain bill. Additions of lactose and sucrose will effect final gravity much more, but when it really comes down to fine tuning your recipe calculations you might want to consider how fermentable crystal malt is. It could change your final gravity by a couple of points.
A few years back some fairly extensive tests were run by Nilo over the course of several months. The results of these tests are a look into how fermentable crystal malt is when mashed with a base malt.
The Average attenuation from Nilo's data:
80 % for 2-row
77% for 50% C10, 50% 2-row
70% for 50% C40, 50% 2-row
67% for 50% C120, 50% 2-row
If we assume that the 2-row ferments the same regardless of being mashed with the crystal the 100% crystal fermentability numbers can be derived.
74% for C10
60% for C40
54% for C120
If you really want to fine tune your recipes you can use the difference in fermentability to adjust your final gravity calculation. These would be:
-6% for C10
-20% for C40
-26% for C120
For example, say you were making a dark beer with 9 lbs of 2-row, and 1 lb of crystal. Based on your mash temperature and type of yeast you can determine your base malt fermentability as described on the "Final Gravity in Recipe Formulation" post.
Let's say that came out to 75%
OG would be (note that mash efficiency has been rolled into the pppg numbers)
9 lbs of 20row * 24 pppg = 216 GU
1 lb of C120 * 23 pppg = 23 GU
216+23 / 5 = 48 GP or a specific gravity of 1.048
FG would be as follows
216 GU * (1-75%) = 54 GU
23 GU * (1-75%-26%) = 4 GU
54+4 / 5 gallons = 11.6 GP or a specific gravity of 1.012
If the fermentability of the crystal was not considered then the contribution would be 6 GU instead of 4. Because this change is very small, the FG of the five gallon batch would still be 1.012
So, as you can see, most of the time the change in attenuation caused by crystal malt is insignificant.
Here is Nilo's thread: (post number 108 has the data)