How?

So how exactly is this information collected and analyzed? This page discusses the methodology of our research and details the process through which web designers can participate.

How does it work?

1. Webmasters wishing to participate in this project sign up for a free account.
2. Then, they install a special JavaScript code on their websites.
3. When visitors view pages on this website, the JavaScript file collects basic configuration information like browser, operating system, plugin versions, etc.
4. This JavaScript file sends this information to the Web of Variables server through a "beacon" or data-encoded image request.
5. The information is collected and analyzed on our server.
6. The webmaster can view statistics for his or her website's traffic and make the necessary code changes.

Methodology

This project will attempt to answer two questions:

1. Which end-user configurations are the most popular overall?
    
2. Do the most popular configurations vary enough between websites to warrant individual site testing and accommodation of different configurations for each site?

To answer these questions, this project will monitor Internet visitors on various websites. Each website will voluntarily participate in this research by installing a small JavaScript file on their pages. When these pages are accessed, users will be sampled at a random interval, and information concerning their operating system, browser, plug-ins, and user settings will be collected and sent to a mySQL database. Only basic and readily available information will be collected - no personal information will be involved with this project.

Collecting internet statistics by random sample should be the best sampling method because 1) an adequate sample size should generalize to the entire population and 2) investigating only a percentage of the site's traffic should minimize additional bandwidth requirements.

Nonetheless, there are some drawbacks that should be considered when using sampling methods like the one selected for this project. 1) Random sampling might record the same user more than once. 2) Collecting information without requiring user input might result in the collection of search engine spiders. 3) Different network nodes between a website and the end-user might temporarily cache (store) a page, resulting in a reduction of visitor hits and a manipulation of the sample. 4) End-user computers might cache files as well, meaning repeat visits to a cached page will not be included in this research (Haigh, 1998).

Luckily, scripts already exist to collect the data needed for this research. This project will simply adapt them to the needs of the research. Many of these scripts have been compiled by BrowserSpy (http://gemal.dk/browserspy/). This website introduces users to the various types of information a website can collect about an end-user's configuration.

The collection and cataloging process will require cookies, JavaScript, PHP, and mySQL. Because the analysis will primarily involve nominal values and measures of central tendency, most of the math can be calculated with PHP and mySQL, without the help of a statistics program like SPSS.

The primary results of this project should be continual, so webmasters will be able to check these stats at any time. By offering this service to webmasters for free, it should be easier to convince third party webmasters to voluntarily participate in this study.

Overall, there are several steps necessary to make this project a success:

1. create the mySQL database to collect the information
2. create the PHP scripting program to manage the information
3. write a JavaScript file that webmasters can incorporate into their websites
4. market the script as a useful tool in order to get webmasters to use it.