II. Methodology

A. Data Collection

The DLL used customized searches powered by Google as its primary tools for collecting the data analyzed in this study. Customized searches were introduced by Google in October 2006. As the reader will see from the description on Google's Website, a customized search can be directed to confine its searching to a specified set of Websites. Whereas Google usually searches all of the Websites in the world, a customized search can be confined to a specific list of Websites. This study used three highly focused customized searches:

1. Customized searches of all HBCU Websites enabled us to identify the Websites that were most frequently referenced by all HBCU Websites, i.e., the "Top 10"
   
   
2. Searches of the "Top 10" Websites enabled us to identify the sites most frequently referenced by the "Top 10", i.e., their "Closest Associates". We then combined the "Top 10" and their "Closest Associates" into a "Core" group.
   
   
3. Searches of the "Core" enabled us to determine their connections to the Websites of a selected group of nationally acclaimed academic leaders, Websites to whom Google had assigned PageRanks = 10, 9, and 8.

(Note: Buttons linking to the customized searches for all HBCUs and for the Core will be found on the Gateway's customized HBCU Search Page. A button linking to the customized search of the "Top 10" Websites will be found in Section IV of this report.)

B. Subjects of Study

Given that our ultimate objective was to identify a "Core" group of Websites that facilitated the entire HBCU community's access to the outside world and vice versa, we quickly realized that we had to include a few other organizations besides HBCUs in our study. In particular, we had to include the Websites of five well-known organizations that provide important services to HBCUs: the United Negro College Fund (UNCF), the Thurgood Marshall Scholarship Fund (TMSF), the National Association for Equal Opportunity in Higher Education (NAFEO), the HBCU Library Alliance, and the HBCU Faculty Development Network.

C. ReferenceRanks As per this paper's Introduction, our objective was to construct "ReferenceRanks" that would be analogous to Google's "PageRanks". We therefore attempted to mimic Google's strategy as closely as possible. Google's PageRanks are not only based on the total number of links that a Website receives, but on the "quality" of the links they receive. In other words, links received from some Websites are worth more than links received from others. But whereas Google's methods for assigning PageRanks are based on a patented set of algorithms that are implemented automatically by computer programs, our ReferenceRanks will be heuristic and hand-crafted. And whereas Google assigns PageRanks from 0 to 10, we originally intended to only assign ReferenceRanks from 0 to 3.

• As will be seen in the next section of this report, we began our analysis by determining the total number of references that an HBCU's name received from the Websites of all other HBCUs using a "Customized Search" of all HBCU Websites.
  
  
• We originally intended to identify the top ten Websites that had received the most references. However, four more Websites received almost the same number of references as the Website in tenth position. Accordingly, we also included these four sites in our initial group. Hence we place the term "Top 10" in quotes throughout this discussion.
  
  
• After identifying the "Top 10", we constructed a second "Customized Search" that only searched the Websites of the "Top 10". We wanted to identify the "Closest Associates" of the "Top 10", i.e., the Websites that were most frequently referenced by the "Top 10". We then combined the "Top 10" and their "Closest Associates" into a "Core" group and assigned ReferenceRank = 3 to each member.
  
  
• Returning to the table of references made by all HBCUs, we divided the remaining Websites into two groups: those that scored in the upper half of the range (from 154 to 305), and those that scored in the lower half (below 154). We assigned ReferenceRank = 2 to the upper group, and ReferenceRank = 1 to the lower group. No HBCU received zero references, so none was assigned ReferenceRank = 0.
  
  
• As the reader will see in a subsequent section, the UNCF's reference count so far surpassed the number of references received by every other HBCU and service organization that we elevated it into a higher category, i.e., we assigned it a unique ReferenceRank = 4.
  
  
• Finally, we constructed a third customized search that only searched the Websites in the Core. We used this search to determine the number of references this central group made to selected well-known academic Websites in the outside world to whom Google had assigned PageRanks = 10, 9, and 8. Other HBCUs that referenced the Websites in this group would thereby provide their users with indirect access to the Websites of the nation's academic leaders.