I have the same devices, except with the wire antenna. I can take one of the devices outside, so the signal has to travel through multiple walls. At distances up to 100 feet, no dropped packets.
So, the antenna type DOES make a difference.
I don't see much difference between the chip and wire antennas. I don't experience dropped packets anywhere in my house with either design; house is about 2300 square feet, three floors. Device in attic has no problems communicating with device in crawlspace. They are series 2 devices, which may make a difference, though I may be seeing some mesh network benefits.
I do use a pair of XBee Pros with RPSMA antennas for remote communications with one of my outbuildings, located about 800 feet from my house. As with my other xbee installations, this configuration just worked out of the box so I haven't done a lot of research into signal strength, etc.
I have a connectPort X2 centrally located in my house which has an internal chip antenna based XB2 Pro running coordinator firmware; I believe my other bees are all running router firmware, though I could probably get away with end device FW because the installation is so small. The ConnectPort runs custom Python code that aggregates sensor readings and transmits it to Pachube and my home automation solution.
Another potential difference - I use Arduino FIOs for most of my external sensor nodes, and they are mounted in plastic enclosures. FIO is nice because it has LIPO charging circuitry and an XBee socket built in, so sensors end up being much smaller.
I have heard some folks complain about terrible range with certain types of enclosures (especially metal).
Rob Faludi's 'Building Wireless Sensor Networks' is actually a great resource for this topic; he covers the implications of different antenna designs in the book, and I believe on his site as well. He also has several posts about extending battery life (plus information on expected runtimes for different battery types) on his blog.