Researching relationships between malware families is critical because it makes malware identification and clustering easier.
Any single malware sample is rarely totally unique. Just like in non-malicious software development, a new malware sample will typically reuse code and/or techniques from older samples.
Malware researchers can look for these shared elements with signatures. These signatures will cluster (or "group") separate malware samples which have similar characteristics. You could, for example, write signatures to identify samples which contain techniques like DNS Poisoning and Browser Settings Hijacking. You could also write signatures to identify members of a specific malware campaign such as Zeus or Conficker.
Such signatures can (1) be used to protect users, by detecting that a given file contains malware, and (2) be used to track how the malware changes and spreads over time.
This is made possible when researchers understand relationships between malware samples rather than trying to study each sample in isolation.