You can always AND the netmask with the various IP addresses. I present a different method which I feel is more convenient. It's from Todd Lammle's CCNA prep book (for classful addressing specifically, but the approach extends to this situation).
We have the netmask $255.255.255.224$. I subtract the last octet value from $256$ to get what Todd calls the block size of that octet. It's 32 here. That means that our subnets, starting from $192.168.1.0$, are $192.168.1.0$, $192.168.1.32$, $192.168.1.64$, $192.168.1.96$, $192.168.1.128$, and so on. You get the picture. You keep adding the last octet's block size to get the subnet addresses for all subnets in the network.
Why this makes it easier, is that now I can just compare the last octet values and see which subnet an IP address is in. For example, $X$ is in $192.168.1.96$ and $Y$ is in $192.168.1.64$. Once you find out the subnets of all the given IPs, you'll see that there are only 3 distinct subnet addresses, and so the answer is $C)$.