X hits on this document

PDF document

The Web Structure of E-Government - Developing a Methodology for Quantitative Evaluation - page 2 / 10





2 / 10

nodality in the virtual realm, thereby weakening one of its key tools. We can hypothesize that a ‘healthy’ government domain, if we can establish appropriate characteristics to define such a thing, will help government to become more nodal. If a domain has more incoming links, for example, it is likely to be more visible to search engines and more easily found by citizens searching for government-related information.

Current research in computer science, political science or communications research tells us little about the Web structure of e-government. As outlined below, there are numerous qualitative analyses of government online and international rankings of e- government. However, there are no quantitative attempts to analyze the underlying link structure of a government domain, to assess the ‘health’ of such a domain, or to compare domains. We believe that such an undertaking will provide valuable information about this new element of government and ultimately could be used to aid the redesign of governments’ online presence. Although there are a number of studies developing the field of webmetrics (discussed below), there have been virtually no attempts to apply them to government.

Existing Web metrics may be categorized as either user-based or structural. User-based metrics measure a variety of characteristics of a user’s usage of a site, e.g. page impressions. Acquisition of user-based metrics can be difficult since the data is only available to the owner of the website, and the data may be difficult and costly to obtain. Furthermore, cooperation might not be forthcoming, if the evaluation is performed by a neutral third party intent of publishing comparative data between sites. And data may be unavailable for reasons of user privacy or contract confidentiality (where external providers maintain the site).

In contrast, structural metrics measure properties such as the average distance between two random pages and the interconnectedness of sites. They are readily available to anyone capable of crawling websites. They would seem to offer the potential for establishing the ‘health’ of a domain. For example, if a government domain is highly inter-connected, then citizens are much more likely to find information (such as how to make an application for a visa) by traversing the link structure of the site.

There are numerous technical and methodological problems involved in creating such metrics at the domain level. The numbers of pages, nodes and links involved can be very large. For this reason, we have chosen to test our methodology on a single agency example – the national audit office – which may be compared across several countries. The UK National Audit Office, the US General Accounting Office, the Canada Audit Office, the Controller and Auditor General of New Zealand and the Supreme Audit Office of the Czech Republic all have roughly comparable roles and responsibilities. Thus, we believe that performing a comparative evaluation of their online presence will take us some way towards developing a methodology for the much larger task of comparing whole government domains. In addition, the audit offices can be deemed to constitute some part of each country’s overall e-government effort, so there will be some value in comparing existing evaluations of each country’s place in e-government rankings.

In Section 2 we briefly survey prior work in the field. Section 3 then discusses our experimental methodology. Section 4 describes experimental results and Section 5 provides a discussion.


There have been extensive efforts to assess the quality of e- government throughout the world. An overview of this work is provided in Section 2.1. Similarly, there have been numerous studies within the computer science community to assess and characterize the structure of hyperlinked environments, as discussed in Section 2.2.

2.1 Qualitative Assessment of E-Government

There have been numerous attempts to assess e-government internationally, in the form of rankings of countries carried out or commissioned by international organisations (such as UNPAN [15], European Commission [4]), private sector consultancies (particularly by [1] Taylor Nelson Sofres [14] and Graafland-








[8][16][13][6]). While some are widely cited and eagerly awaited by governments which score well, all are of methodological questionability and rely, ultimately, on subjective judgments. Most make some form of assessment of government websites according to content (eg. [16]) and availability of services (eg. [4]), while Accenture’s widely known annual study is largely a qualitative analysis based on researcher assessments of websites and available e-services and a limited number of short visits to the 22 countries covered.

All these studies fail to collect either user-based or structural Web metrics. None have been able to overcome the ‘cooperation’ problem, noted above, with regard to collecting user data for significant numbers of websites, although some use survey evidence (Taylor Nelson Sofres [14] in particular, while Accenture included a user opinion survey for the first time in 2005 [1]) to estimate the extent to which a population as a whole have interacted with their government online. West [16] gathered content-related data from approximately 2,000 websites in nearly 200 countries and LaPorte and Demchak developed measures of ‘interactivity’ and ‘transparency’ for tracking the diffusion and use of the Web in nearly 200 governments around the world (now discontinued). However, none of these studies have considered the link structure of e-government sites.

Methodological variations across these studies are evidenced by the different rankings that the countries achieve. The table below gives the scores attained by the countries covered here in the most recent reports by the UN [15], Accenture [1], Taylor Nelson Sofres (TNS) [14], and West [16]. The table shows that (for example) the UK scores well with the UN but close to the median for Accenture’s 22 countries, well below Canada and the US for West and lowest of these six countries for TNS, the latter one reporting percentage of population using government online in the last 12 months.





0.927 0.806 0.542 0.710 0.814 0.927 0.009

st st

1 1 n/a n/a 10 2 22

th nd nd

63 51 23 45 18 44 1

46.3 42.4 33.8 35.5 37.7 45.3 16.0





Table 1. Selected results of important e-government rankings

Maximum Canada Czech Republic New Zealand United Kingdom United States Minimum Countries Surveyed

Document info
Document views15
Page views15
Page last viewedWed Oct 26 23:15:18 UTC 2016