3.0 Web Usage Data Analysis
This analysis examined both public and ITD maintenance staff use of the Road-Weather Integrated Data System (RWIDS) webpage. The methodology, results and conclusions of the analysis are presented in this section.
3.1 Methodology
3.1.1 Public Website Usage Analysis
The RWIDS webpage was added to the ITD Road Report traveler information website in November 2002. The focus of this analysis is to determine how the addition of the RWIDS information to the Road Report website influenced usage of Road Report, including the overall volume of usage and usage of the various types of RWIDS information. Two analyses have been performed to address this question, a before-and-after comparison of the number of website user sessions and an after-only examination of the various types of RWIDS information viewed.
Pre-deployment (before) data was analyzed for the two winter seasons that immediately preceded the November 2002 RWIDS launch. Those two winter seasons encompass the periods November 1, 2000 through May 31, 2001 and October 2001 through May 31, 2002. Post-deployment data from one winter season was analyzed, encompassing the period October 2003 through May 2004—the second winter of RWIDS operation. As with all of the evaluation analyses, it has been assumed that the second winter season provides a more accurate representation of the RWIDS system. By the second winter the public and ITD staff had time to become familiar with the system and the system itself had been “broken in” (the first winter of operation included some prolonged periods where RWIDS data was not updated as frequently as intended due to server outages).
The data utilized in this analysis consisted of ITD web server log files. Data manipulation was accomplished via custom Microsoft Access database routines. Off-the-shelf website analysis tools like WebTrends did not provide the needed capabilities. Public users were differentiated from ITD users based on the Internet Protocol (IP) address of the individual web pages viewed. All Road Report web pages accessible by the public include “public” in the IP address.
This analysis focuses strictly on user sessions and does not consider the users themselves. It was not possible to analyze changes in the number of discrete public website users because data on individual users is not available (the “incoming” or user IP addresses in the log file are dynamic across sessions for most users and are not directly linked to a specific user).
The post-deployment analysis RWIDS information types examined the number and percentage of total user sessions that included viewing of various types of information. Omitted from this analysis are several types of information, including avalanche warnings, composite radar, isobars, and Pacific Loop radar for which usage is not tracked in the ITD server logs. These data are resident on other organizations logs and when selected, a separate web browser window is opened, directed to the other organization’s server, thus leaving no trace on the ITS server.
3.1.2 ITD Website Usage Analysis
The focus of this analysis is to determine how much ITD maintenance staff used the RWIDS webpage and which RWIDS information was viewed most frequently. Two analyses were conducted to address these issues, one examining the volume of RWIDS utilization and one addressing the type of RWIDS information accessed. This analysis only considered the post-deployment period because the focus is on the RWIDS information, which was not present in the pre-deployment period.
Like the public website usage analysis, this analysis considers only the 2003-2004 winter season only (October 1, 2003 – May 31, 2004), the second full winter of RWIDS webpage operations. The analysis did not consider the first winter of operation because the system was still being refined and training was still being conducted.
Data for this analysis came from the same source used for the public website usage analysis—ITD web server log files. The ITD maintenance staff RWIDS utilization presented here represents a conservative estimate, that is, somewhat less than actual usage. This is because it was not possible to differentiate one particular type of ITD maintenance user session from general public sessions: those conducted from home or from one of the 37 maintenance sheds that do not have access to the ITD wide area network (37 of the total 71 sheds statewide) and where the publicly accessible version of RWIDS, rather than the ITD-only version, was viewed. (The public version contains the same information in the same format but excludes the “alerts” feature.) Any user session from a wide area network (WAN)-connected computer, or who viewed the ITD-only version of the site were discernable in the server log files, and those sessions have been analyzed.
Unfortunately, it is not possible to estimate the number of “missing” ITD maintenance RWIDS sessions, and therefore estimate how many additional ITD sessions may have occurred in addition to those captured in this analysis. Based on feedback from ITD maintenance staff who were interviewed and those who participated in the e-mail survey used to identify interview candidates (results presented in Section 2.0), it appears that this could be a substantial number of sessions. A number of staff indicated that they were either unaware of the ITD-only version of the RWIDS page, or wished to avoid the user log-in required to access it, and therefore used the public version of RWIDS.
Although it is not possible to estimate the actual number of missing sessions, a likely upper limit to the number of those sessions can be estimated. Specifically, that the actual number of total ITD sessions is no more than twice the number captured in this analysis. If every user session from non-WAN connected sheds—about half of the total sheds—viewed the public version of RWIDS, and assuming the level of RWIDS usage for this group is not significantly more than among WAN-connected users (a reasonable assumption given that WAN connections are faster), the missing sessions would represent no more than about the same number of sessions captured in this analysis. Thus, the total number of sessions is unlikely to be more than twice the number described here.
Once ITD user sessions were differentiated in the ITD server log file, the total number of sessions per month was calculated. The number of sessions featuring viewing of the various types of RWIDS information was also calculated, based on the web page IP address information in the log files. As was the case with the public usage analysis, ITD access of a number of types of information could not be calculated because the information is actually resident on another server, appears in a separate window when selected, and leaves no trace in the ITD server log.
3.2 Results
3.2.1 Public Website Usage Analysis
Usage Volume
Figure 3-1 compares pre- and post-deployment monthly Road Report website sessions. For most months, usage of the Road Report website increased dramatically in the post-deployment period. Overall, the average number of monthly user sessions increased by 169 percent, from 41,000 sessions per month to about 109,000 sessions per month. In both the pre- and post-deployment periods, use of the website peaks in the middle of the winter season when the most adverse winter driving conditions occur, and is much lower in the early and late portions of the season. This is consistent with the fact that the website is focused on winter traveler information.
Figure 3-2 decomposes the post-deployment Road Report user sessions by type, showing the relative proportion of sessions in which only “traditional” Road Report content was viewed (content of the type included on the pre-RWIDS version of the Road Report website); sessions in which only RWIDS content was viewed; and sessions in which content of both types were viewed. It is clear that the large increase in overall usage included sizable gains in traditional Road Report usage, which increased by about 90 percent, and which constitutes the bulk of total post-RWIDS Road Report usage (about 71 percent of all sessions). However, it is also clear that the RWIDS webpage was utilized considerably and the addition of the RWIDS webpage accounts for about half of the total increase in Road Report website usage. Further, most (85 percent) of the RWIDS-related usage increase featured sessions where only RWIDS sessions were viewed. These sessions account for about 25 percent of Road Report website sessions. It seems clear that a significant portion of the growth in Road Report usage is related to the introduction of the RWIDS webpage.
Figure 3–1. Public Monthly Road Report Website Usage
Figure 3–2. Post-Deployment Public User Sessions by Content Viewed
As indicated in Figure 3-2, very few Road Report website user sessions—about 4 percent—include viewing both traditional Road Report content and RWIDS content. Both types of information are popular, and so much as the introduction of RWIDS has not reduced traditional Road Report sessions, the two information sources can be viewed as complimentary (they do in fact provide very different types of information). Complimentary though they may be, the two information sources do not appear to be synergistic. If in fact travel planning decisions for a given trip are usually made based on a single user session, than on a decision by decision (i.e., trip by trip) basis, it appears that decisions are usually influenced either by the traditional Road Report data or by the RWIDS data, but seldom by both.
Although it is not clear why RWIDS and the traditional Road Report information would not be used in combination, there are a couple of possible explanations. It may be that users consult the sources for different types of travel decisions: the traditional Road Report for route-specific roadway conditions—when they want to know about conditions along the route to their destination, and RWIDS for point-specific information on overall weather conditions—when they want to know what weather they expect at their destination. Aside from pavement condition information, which is not listed on the RWIDS main page (it is a subitem under the weather menu, which must be selected in order to see the pavement option), the RWIDS webpage provides weather information (e.g., air temperatures, cloud cover, winds) as opposed to information on roadway conditions.
An alternative theory is that many users consult RWIDS primarily for the camera images, a premise supported both by anecdotal information from ITD headquarters staff that the public is traditionally most interested in camera images and by that fact that camera images are the most popular type of RWIDS information. After becoming familiar with which routes include cameras and which do not (like most states, camera coverage is still sparse on a statewide basis) users only visit RWIDS for the trips that they know include cameras along their routes. When they know in advance that there are no cameras of interest for a particular trip decision, they view only the Road Report information. This theory does rely heavily on the debatable assumption that frequent users come to remember specific camera locations.
Usage by RWIDS Information Type
Figure 3-3 presents the percentage of total RWIDS sessions (both RWIDS-only and RWIDS plus traditional Road Report) using various types of RWIDS information. Omitted from Figure 3-3 are a number of information types where the information does not reside on the ITD server, and where users are directed to another organization’s server and the information appears on the Road Report website in a separate, inset, web browser window. This includes information on avalanches, composite radar images, isobars, jet stream, and Pacific Loop radar.
Expectedly, the RWIDS homepage, through which any public user would necessarily pass if not accessing an RWIDS subpage via a previous bookmark, is the most commonly viewed portion of the RWIDS webpage, included in almost 95 percent of all sessions. In addition to providing the gateway to various specific information, the homepage also contains a map of Idaho showing temperatures and cloud cover (e.g., sunny, cloudy, partly cloudy, etc.) for locations throughout the state.
Figure 3–3. Public RWIDS Information Viewing by Data Type
Among the other specific information available on the RWIDS webpage, the most popular are camera images, included in about 20 percent of all sessions. Cameras are followed by individual ESS, included in slightly over 10 percent of all sessions. Users can select a specific station from a long scroll-down list of locations and view several types of information, including temperatures, wind, dew point, and cloud cover. Following ESS, at slightly under
10 percent of all sessions, is the statewide weather map—the same map that appears on the homepage. These selections are most likely made by users who have been navigating through the RWIDS webpage and do not want to hit the “back” button repeatedly to return to the homepage (there is no “home” button on most of the subpages) or who have not entered RWIDS via the homepage (that is, have entered via a bookmarked subpage). Usage of the remaining information types falls off quickly to 3 percent or less of total sessions.
There is clearly no data type aside from the homepage which is accessed most of the time. Even cameras are accessed, on average, only in about one of every five user sessions (20 percent), and frequencies for the remaining data types drop off quickly. The lack of a popular single data type besides the homepage could mean that the statewide temperature and cloud cover information on the homepage is the primary RWIDS attraction among public users. Alternatively, it could be that a lot of the public user sessions are “one-time” sessions, with the user investigating the RWIDS homepage, never to return.
3.2.2 ITD Website Usage Analysis
Usage Volume
Figure 3-4 presents ITD maintenance staff RWIDS sessions by month for the 2003-2004 winter season (October through May). The data in Figure 3-4 constitute a conservative estimate of ITD maintenance staff usage in so much as it excludes sessions from non-WAN-connected maintenance sheds in which the public version of the RWIDS webpage was accessed. Although the number of these missing sessions cannot be estimated, it is likely they represent no more than the same number of sessions reflected in Figure 3-4.
Figure 3–4. TD Maintenance Staff RWIDS Sessions by Month
Monthly ITD RWIDS sessions ranged from about 300 during the late winter/early spring months up to about 2,550 sessions during December and January. For reference, if usage during the peak month, December 2004, was spread evenly across the 71 maintenance sheds, it would amount to about 36 sessions per shed, an average of a little more than one session every day.
Usage by Information Type
Figure 3-5 identifies the percentage of total ITD maintenance staff sessions in which particular RWIDS data types were viewed—an indicator of the relative popularity of the diverse RWIDS data. Omitted from Figure 3-5 are several data types where usage could not be tracked from the ITD server log files, consisting of composite radar images, isobars, NWS avalanche alerts, jet stream weather images, and Pacific Loop satellite images.
Figure 3–5. TD Maintenance Staff RWIDS Information Viewing by Data Types
The most frequently accessed portion of the RWIDS webpage, viewed in nearly 90 percent of all sessions, is the homepage. This could be because most users pass through the homepage before accessing more specific RWIDS data (the alternative would be to access that specific data directly, via a bookmark, without passing through the homepage). Or it may be because the statewide weather conditions map (temperatures and cloud cover) included on the homepage represents a popular data type. Aside from the homepage, there is no single type of RWIDS data that is accessed during the majority of sessions. The next most commonly viewed data types are cameras and individual ESS, both viewed between 40 and 50 percent of all sessions. Although there are clearly more and less popular data types, the results suggest that there is no “go-to” data type that is always or almost always viewed. Rather, different types of data are viewed in different sessions, which may suggest that some data types are more useful for some situations and the associated decision-making.
The top four most popular data types among ITD users are the same as among public users. There are two major distinctions between ITD staff and public users’ access of RWIDS information. First, public users aren’t very interested in anything other than the homepage. Aside from camera images, none of the other data types are accessed very often (all are accessed in less than 10 percent of all sessions). ITD users, on the other hand, demonstrate interest in nearly the full range of RWIDS data—only two data types are viewed in less than 10 percent of all sessions. The second distinction is that even though there is no “silver bullet”, nearly always-viewed data types among either set of users, the most popular data types among ITD are much more popular than the data types most frequently viewed by the public. Camera images—the second most popular data type among both public and ITD users—are viewed in only about 20 percent of all public sessions. By comparison, camera images are viewed in a high percentage of ITD user sessions, about 46 percent. Even though there is no “always viewed” data type for either group, there are several very frequently accessed data types among ITD users. Public users, on the other hand, do not show high levels of interest in anything other than the homepage.
Although ITD users seldom view, in RWIDS, the road closure information available from the Road Report, they do in fact use the “traditional” Road Report data (that is, the data that was on Road Report pre-RWIDS, and still is). Approximately 19 percent of all ITD user sessions view only traditional Road Report information, indicating that, as is the case with the public, both types of data have value to at least some ITD users, that is, they are complimentary. The number of ITD sessions that included viewing both traditional Road Report and RWIDS data may be even higher. About 45 percent of all ITD sessions included a presence on both the traditional Road Report webpage and the RWIDS webpage. However, it is known that some percentage, perhaps a very large percentage, of those Road Report visits were restricted to the homepage, which the ITD user visited only to access the ITD RWIDS page log-in.
Figure 3-6 identifies the popularity of various types of traditional Road Report data types among ITD users. Expectedly, the most popular information is, in fact, the primary Road Report information—brief text descriptions of the travel conditions on individual roadway segments. That information is one of only two types of real-time, regionally relevant, roadway information on the webpage (the other being semi-real time information on roadway detours and closures). None of the other types of information are frequently accessed by ITD users. That information includes static (non-real time) general interest information such as press releases and an overview of ITD winter maintenance practices (“Winter Road Maintenance”). It also includes “Additional Road Information,” brief descriptions of maintenance, and construction-impacted travel conditions. Following the Idaho road condition information, the most popular selection on the Road Report webpage among ITD maintenance users is the RWIDS employee log-in.
Figure 3–6. Traditional Road Report Data Accessed by ITD Users
3.3 Conclusions
- Overall public Road Report usage in the post-deployment period is strong and increased significantly in the RWIDS era. Usage peaked at nearly 300,000 user sessions per month in mid-winter 2004. Usage in the second winter after RWIDS deployment is
169 percent higher than pre-RWIDS levels. Much of this increase probably reflects a strong general upward trend in Road Report usage and is independent of the RWIDS addition—
71 percent of the post-deployment sessions included viewing of only traditional Road Report information. - The RWIDS webpage was utilized to a considerable extent and the addition of RWIDS accounts for a significant portion of the overall growth in Road Report usage. There was an average of about 218,000 sessions each month that featured only RWIDS information access, accounting for about 20 percent of the total growth in Road Report usage. These figures are particularly positive in light of the fact that the ITD conducted essentially no marketing in support of the RWIDS roll-out. The only way the public would find their way to the RWIDS webpage would be by stumbling into the link on the Road Report webpage.
- The public values both RWIDS and traditional Road Report information (it appears complimentary) but they don’t use them in concert (they are not synergistic). Very few public Road Report sessions, about 4 percent, include viewing both RWIDS and traditional Road Report content. This suggests they use the different types of data under different circumstances to address different travel planning needs. The role of the different information types is not clear. RWIDS may be of interest to users seeking information on their destination—discrete (not summarized into an overall advisory) information—which is how RWIDS presents information. They may use the traditional Road Report for synthesized, advisory information on their travel route (the information is organized by roadway segment). RWIDS information is diverse and unconsolidated, lacking a specific “bottom line”. Users are left to synthesize it on their own, and therefore may use it when they are only interested in a specific piece of information, like what it’s like at their destination (thus explaining the popularity of camera views and individual ESS data). Road Report information is consolidated and although not necessarily prescriptive, it provides clear summaries of roadway conditions—little synthesis is required of the user.
- The public doesn’t appear to be frequently interested in most specific RWIDS information. Most public user sessions penetrate no further than the RWIDS homepage, either because they are only interested in the statewide temperature and cloud cover map presented there or they are making an exploratory visit to RWIDS and are not looking for specific information. The most popular specific RWIDS data type (subpage) is camera images, which are only accessed in about one in five sessions. There is little or no apparent interest in pavement conditions, National Weather Service watches and warnings, weather radar images, National Weather Service forecasts, weather satellite images, or road closure information. This may suggest that the public is not sure how to synthesize and derive benefit from the various discrete information types on RWIDS, or that they are not interested in investing the effort to do so.
- ITD maintenance personnel demonstrated significant interest in the RWIDS webpage; they used it fairly frequently. Average monthly sessions range from about 300 during the late fall and early spring months to slightly over 2,500 sessions during mid-winter (January and February).
- TD maintenance users are interested in a wide variety of RWIDS information. Only two of the ten RWIDS data types for which usage could be analyzed are consulted in less than about 10 percent of sessions: weather satellite images and road closure and detour information. This is consistent with the desire for a very wide variety of information expressed by maintenance personnel during the design of the RWIDS webpage and in the post-deployment interviews. The breadth of ITD user interest is in sharp contrast to the public who show only limited interest in one or two types of information. This suggests that institutional users are more willing and able to synthesize a variety of discrete information—none of which provides clear direction or advice. It may also be that the much wider range of decisions faced by maintenance users calls for a wider range of data. Public users, usually facing a “go/no-go” or “how much time do I allow?” travel decision may prefer a single data type, consisting of an easily interpreted advisory message, such as that found on the Road Report.
- ITD maintenance users are more interested in RWIDS information than are most public users. In addition to being interested in a wider variety of information, maintenance users are simply more interested in the data. Of the ten data types analyzed, maintenance users accessed 8 of them in 10 percent of their sessions or more; the two most popular (cameras and individual ESS data) in more than 40 percent of sessions. This suggests that RWIDS information is more critical to maintenance decisions than general travel decisions, or its manner of presentation is less attractive to public users.
- ITD maintenance users are also interested in traditional Road Report information. In a significant number of user sessions, about 19 percent of all ITD sessions only traditional Road Report information was accessed (does not include any RWIDS information access). There appears to be value in the traditional Road Report information for both the public and institutional users and that for at least some maintenance personnel, RWIDS has not made the Road Report information obsolete. The most popular information is the primary, traditional focus of the Road Report: advisory information on the weather-influenced travel conditions.
- Relatively low ITD user interest in pavement information may be due to limited coverage. ITD maintenance personnel would seem to have an interest in pavement condition information, as an input into winter roadway treatment decisions—certainly more so than the general public, who might be expected to have more difficulty in interpreting such microscopic data. ITD users did demonstrate more interest than public users (about
18 percent of sessions versus about 2 percent), but the limited number of pavement sensors and very large (almost district-wide) gaps in coverage would seem to limit the usefulness of pavement data overall. - Cameras are very popular among all users. Consistent with the sporadic anecdotal feedback received by ITD from public users and the opinions expressed by ITD maintenance interviewees, camera images appear to be the single most useful type of information. This is probably because they provide fairly comprehensive and easily interpreted information. Maintenance personnel indicated that they utilize the cameras to observe precipitation and visibility, to some extent to determine certain pavement conditions (e.g., depending on lighting and visibility they can determine if the pavement is wet or dry and whether there is snow on the pavement) and maintenance effectiveness (e.g., whether the road appears to have been plowed, whether freeze-point depressant chemical appears to be working, as evidenced by wet pavement, etc.).
2.0 Key Informant Interviews | Table of Contents | 4.0 Accident Data Analysis