|
About These Statistics By Daniel Klein, Editor, Econ Journal Watch Here you see, for each file, the number of cumulative accesses since the file appeared at econjournalwatch.org. The EJW site is constructed and hosted by Lexicom Internet Services of Calgary, Alberta. (File posting and editing within the "shell" is done by EJW Managing Editor Kevin Rollins.) Lexicom does not use cookies to track website user information on any of the websites it constructs and hosts. Without cookies, the only meaningful measure of readership is file accesses. Lexicom provides a stats package that gives us monthly totals for file accesses. The "Cumulative Accesses" shown here simply sum up the monthly numbers. To view monthly data, click on the Excel file link at the bottom of the Stats Page. In the interest of representing EJW readership as accurately and transparently as possible, we explain the various factors that could inflate or deflate the file-access numbers. In composing this statement I have worked closely with Lexicom's top technical staff. The factors listed here would be typical of any site's file access/download numbers. Our best judgment is that the file-access numbers are a fairly accurate representation of bona fide downloads. The file-access numbers probably inflate the true numbers somewhat (perhaps 5-15%), but no more than is normal at sites that provide file access/download totals. Factors that could inflate the file-access numbers: Software failures: The typical case here is someone clicking on the EJW file link but not getting the file, and then clicking again, creating multiple "accesses" as far as the file-access number is concerned. Also, the software itself may, because of software failures, automatically re-request the file. This problem can arise because of failures in Windows, server software, dirty Internet connection lines, and busy networks. The important thing to note here is that false signals can be sent to the stats server or signals can be lost. Also included as part of this is improperly configured browsers and ISP servers. (It is possible that this factor could deflate the file-access number, but it is much more likely to inflate it, and hence is listed here.) Anti Virus and Firewall Software can block a signal, again stopping a genuine download even though the access-number increased by 1. This should only happen if the EJW file were infected. Search engine robots: Bots or spiders, for example, from Google, go out periodically and search the Internet for links, sites, and pages within sites. When a search engine indexes a site it can hit every page and make it look like there is inflated traffic on a particular day. This is generally done at random and it is impossible to say whether a search engine would index all the pages on your site at one time or just one or two. Refreshing browser: A browser that is set to automatically refresh every few minutes or is refreshed manually will inflate statistics. Each refreshing would add 1 to the file-access number. But few people have their browser set to automatically refresh. Programming accesses: In mastering the file, the programming team might access the file repeatedly to check on adjustments, etc. This factor is very minor. Factors that could deflate the file-access number: Browser Caches: If your browser is set to cache a site, then, when you access an EJW file on Monday, your computer stores it, and if you click again on the file link at the EJW website on Tuesday, then your browser simply displays the cached page from Monday, rather than sending a new signal to the EJW site. Thus, what should arguably be counted as two viewings is counted as only one. Proxy Caches: This is like Browser caches, except that the caching is happening at the level of, say, your school or company. When your colleague in the same building accesses the file, it is cached for viewing by anyone else under that proxy who might also click on the link. Again, multiple viewings would show up as only one file access. Why We Think the File-Access Numbers Are Fairly Accurate: Even the best technicians have no way of knowing exactly how widespread the factors are. If users' institutions use caching to reduce network traffic and improve speed, deflating the numbers, there would be no way of knowing that. We doubt that the file-access numbers significantly inflate the bona fide number. First, the site is produced using software to make the files as accessible as possible; we do not receive many user reports about software failures and access problems. Second, the access pattern, day by day, follows the typical pattern of explosion upon initial posting/announcing, and then falls off to low steady numbers until some prominent notice occurs. Search engine bot activity would not coincide with posting/announcing, so it is unlikely that such activity figures prominently in access totals. Lexicom's technical staff members tell me that they think it reasonable to say that most likely the file-access number inflates the bona fide access only somewhat, perhaps in the range of 5 to 15 percent. This inflation is typical of any site that provides file access numbers, so the numbers here would be directly comparable to those at other sites. |
|
|