The Georgia County Guide has been published annually by Georgia Extension since 1981. It is a compendium of over 1,000 county-level variables that describe the physical, economic, and demographic aspects of each of Georgia’s 159 counties. Annual sales of the Guide have been around 3,000 copies. To improve its usability, single-user software contained on floppy disks was developed in 1995 to generate graphs, reports and summary statistics. However, because of difficulties with the local software programming company, this method of delivery was costly and unreliable. Delivery via the Internet promised to be cheaper, and it would make the information available to a larger audience.

The software chosen for this project was SAS/Intrnet. Information about this software can be found at www.sas.com/products/intrnet/index.html. Other web-enabling software packages may have been appropriate – candidates included Coldfusion and MS Active Server Pages. However, to the author’s knowledge, no single package contains similar data management, report writing and graphical capabilities for the Unix and Windows platforms as SAS.

In addition, SAS Institute offered a “Pilot Program” with its Intrnet software. For a $8,000 fee, a SAS software engineer worked with the author in Athens for one week in October, 2000. He installed the server software and produced the groundwork for a functioning website. In retrospect, this $8,000 was money very well spent because it pointed the author “in the right direction” and it saved untold months in development time. As part of the Pilot Program, SAS included five free days of classroom software training. The author took several classes including SAS/Intrnet and macro language programming. The author also had to update his programming skills to include html and javascript.

Over the course of the next three months, the author subsequently developed the pilot website into the final product. It went online in February 2001, with the cross sectional and the time series analysis applications. The homepage for www.georgiastats.uga.edu is shown below.

The shift share application was added in September 2001. This module was inspired by another SAS/Intrnet application for shift-share analysis written by Dr. Gary Smith of Washington State University. Dr. Smith's website can be visited at http://niip.wsu.edu/.

No new computer hardware was purchased for this Extension program. The development work is done with a standard Wintel personal computer, and production service is provided by an existing Unix server on the University’s Research Computing Resource.

The most common type of web content is a file written in hypertext markup language (html). Servers can provide these files in two forms: static or dynamic. As the name implies, static files have been produced at some earlier time, and they are stored on the server awaiting client requests for them. A quick review of the websites cited in the other chapters of this web book show that they all use static html files.

With appropriate software, servers can also provide dynamic html files to clients. Dynamic html files are generated “on-the-fly” by the server in response to a client’s request. With this technology, the client enters information into an html “form” and then clicks a submit button. For example, the static html file http://www.georgiastats.uga.edu/sshare1.html allows the client to enter the time period and the counties she wishes to analyze. This page is shown below.

When the “Go To Query” button is clicked, the client’s computer sends the time period, counties, and her internet address to the server computer. The parameters pass through the server’s common gateway interface (cgi) and are used in a cgi script. The most common cgi scripting language is Perl. However, Georgiastats uses SAS/Intrnet.

Georgiastats currently uses three scripts, one for each of the applications listed on the homepage. Each script is a specially-written SAS program. The one for the shift-share analysis of employment changes is the longest, at 12 pages.

Some parts of a script will appear very familiar to anyone who has used SAS in graduate school because Proc Print and Proc Gmap are found there. Other parts are highly specialized. Scripts use SAS macro language programming in procedures such as inserting numerical values from a dataset into a paragraph of text. Scripts also use SAS’s output delivery system to generate text and graphics together in a dynamic web page. An example of this is shown below. This particular graphic demonstrates how inflation has wiped out the gains in real earnings per job over the last 30 years. The Note contains context-sensitive explanatory material, plus a link to additional explanations.

To summarize, an author must produce three different types of files that are needed for a web application such as Georgiastats: (1.) A static html file that serves as the query form for the Extension client; (2.) a corresponding script file containing a SAS program that generates the client’s dynamic html file; and (3.) one or more dataset files that the SAS program accesses and uses to generate reports, maps and graphs. More details about the operation of SAS/Internet are available at http://www.sas.com/rnd/web/intrnet/inettour/index.html.

An on-going task for Extension professionals is to document the effectiveness of their programs. With web-based program delivery this is easy to do. An author can report the number of requests for each item, and she can report the web domain of the requestor. At Georgiastats impact is documented in two ways.

The first way is available to every web author. Each client’s request for a file is logged on the host server. There are several server-side software packages to report and interpret these log entries. The software used by Georgiastats can be viewed at www.rcr.uga.edu/logstats/Georgiastats/.

For example, under “Items/URLs, List” we see that the Georgiastats’s homepage, guide.html, was requested 1,016 times in a month. Under “Client Domain, List” the number of clients from different businesses (for example, disneycorp.com), government agencies (senate.gov) and educational institutions (harvard.edu) is reported.

Note that there is online documentation for this software. It informs about the difference between a “hit” (a single file request) versus a “session” (a user within a 24-hour period), and why the latter is a better indicator of usage. There are several reasons for this, e.g., hits are always inflated by web crawler activity from search engines such as google.com.

The second way to document usage is somewhat unique to delivery methods that use cgi scripts. When a client submits data that she has put into an html form, the server can store each parameter in a dataset for subsequent analysis. To view the results of a SAS/Intrnet program written to analyze the requests for shift-share employment reports, look at: www.rcr.uga.edu/cgi-bin/broker?_program=guidecod.statestats.sas&_service=default.

Other parameters stored in the dataset include the date of the request and the requestor’s domain name. During the first year of operation of Georgiastats, the system processed about 14,000 individual requests submitted through the cgi. In the first nine months of its operation, the shift share employment module generated about 1,000 individual reports.

This chapter has documented the development, operation and client impact of an interactive web site, The Georgia Statistics System. The author has a 25 percent Extension appointment, and the operation of this website employs all of this time for the primary tasks of updating datasets and writing new applications in community economic analysis.

For FY 2003, the new application under development is the impact of an area’s agribusiness sector on the local economy. This application will use data from the IMPLAN™ input-output model. Certain common datasets exist for the 3,100 US counties, such as the Regional Economic Information System and County Business Patterns. The existence of these county datasets means that an economic analysis application written for one state can be quickly rewritten for use in the other 49 states. That was the case in the shift share analysis application at www.rcr.uga.edu/guide/sshare1.html.

The author hopes that the web delivery of Extension programming materials will complement and enhance the public’s perception of CES. Because of their immediate response to client’s wishes, websites that generate dynamic web pages may be better suited for drawing Extension clientele into a richer educational experience. Web delivery also contains its own capabilities for documenting clientele usage of Extension web resources.