<<

II. SURVEY METHODOLOGY

Survey design

The commercial vehicle survey was designed as a two-stage survey – a business and vehicle survey and a vehicle travel survey. Numerous businesses were surveyed to verify or correct business characteristics listed for the business and to determine the number and types of commercial vehicles garaged at businesses. This first stage of the survey was completed prior to the selection of any vehicles for the second stage, commercial vehicle travel survey. A travel diary was collected from the use of selected vehicles in which information was collected in the first stage survey. The actual sample unit for the second stage commercial vehicle travel survey was the vehicle, not the business owning the vehicle.

Business/Vehicle Survey

The sample frame for the commercial vehicle survey included all commercial freight and service vehicles owned, leased, or operated by businesses listed in the DRCOG business file and garaged at the place of business. Businesses that operated fleets of vehicles for which travel data were collected via other means were exceptions and are described below. Businesses without employees, self-employed persons, were also exceptions to the survey universe. The business file included all businesses listed within the Denver region for 1996. In total, the file contained 90,558 entries, 89,980 of which were businesses to be included in the survey. A target of 7,500 businesses was to be telephoned. Based upon previous surveys, approximately 3,750 businesses were expected to complete the initial survey.

Information was also obtained for commercial vehicles belonging to the 578 businesses excluded from the first stage business survey sample frame. The excluded businesses operated vehicles including taxies, transit buses, US Postal Service vehicles, police vehicles, rental vehicles, fire station and emergency response vehicles, and waste collection vehicles. The 1992 Truck Inventory and User Survey (TIUS), conducted by the Bureau of the Census, provided some information on the number and types of vehicles owned by self-employed persons.

Table 2 lists the specific Standard Industrial Classification (SIC) codes for businesses excluded from the survey. Although these vehicles had been omitted from the survey for logistical reasons, an attempt was made to elicit sufficient information for vehicle miles of travel (VMT) to be estimated for some of these vehicles. Fire station and emergency response vehicles were omitted entirely, as the VMT contribution by these sources was considered to be negligible. Rental vehicles were problematic since there was no way of identifying the number of miles traveled within the region from the total vehicle miles traveled without actually surveying the people who rented the vehicles. It was felt that even then, there would be difficulties identifying specific in-region miles traveled due to the visitors’ unfamiliarity with the region. For example, an automobile rented at DIA and driven around the region, driven outside of the region to Vail then Beaver Creek and back again and driven around outside the region. An attempt was made to obtain order of magnitude estimates of VMT for rental vehicles from the rental companies.

Table 2 SIC Codes Not Included in Sample Selection

SIC Group SIC Code(s) SIC Group SIC Code Description Description Removed 41 Local and suburban All Varies transit and interurban highway passenger transportation 43 U.S. Postal Service All U.S. Postal Service 49 Electric, gas, & 4953 Refuse systems sanitary services 75 Automotive repair, 7513 Truck rental - no drivers services, and garages 7514 Passenger rental 7515 Passenger car leasing 7519 Utility trailer rental 92 Justice, public order, 9221 Police Protection and safety 9224 Fire Protection 9229 Public Order & Safety

Commercial Vehicle Travel Survey

The second stage of the survey collected travel information for a sample of commercial vehicles selected from the vehicles identified in the first stage of the commercial vehicle survey. A goal of 750 completed, usable travel surveys of commercial vehicles was set with the expectation that approximately 1,000 vehicles would need to be selected for the survey in order to meet the goal.

Nonresponse Issues

Travel surveys are subject to non-response by those persons surveyed. Some of the anticipated reasons for non-response in the commercial vehicle surveys are listed below along with techniques used for mitigation of the non-response.

The willingness of a business owner to participate in the survey could be affected by a perception that the contact or survey is being made for regulatory purposes. Thus, care was used in the preliminary telephone calls to ensure the business owner that the survey was not for regulatory purposes.

Contrary to the household survey, pre-notification letters to increase response were not used. Since DRCOG by nature has the image of a government or regulatory body, such letters were considered counterproductive. Also, the likelihood of the pre-notification letter reaching the “correct” person at a business was much less than would be expected for a household.

In place of a pre-notification letter, an informational letter was mailed with the survey packets, or, if necessary, faxed to potential respondents. The purpose of this letter was to “validate” the survey effort. The letter was endorsed by DRCOG. Since a contact person was already established prior to sending the informational letter, it was much more likely that it would reach an interested party than a blindly mailed pre-notification letter.

Businesses could have also been concerned with providing information that they consid- ered proprietary. During the initial screening calls, businesses were informed that specific information that they provide would remain confidential. Data are presented information by aggregating a number of businesses into groups on not by a specific business.

Perceived respondent burden could also lead to commercial vehicle survey non- response. In order to reduce the burden of telephone reporting for the respondent, the option of completing the survey forms and returning the information via facsimile was provided. This provided the respondent the opportunity to assign support personnel to the completion of the form. In order to reduce the financial burden on the business, the fax number provided was a toll-free number to the firm conducting the survey – CB&A.

Survey conduct

The two-stage business and vehicle travel survey process was as follows:

· Potential businesses were selected from the DRCOG business file (subject to the sampling procedures detailed in a following section) for the first stage business survey. An interviewer conducted a brief survey of the business over the telephone. The interviewer first tried to reach the person responsible for any vehicle owned or leased and then explained the purpose of the survey. During the survey, the respondent was asked to verify information in the business file, the number of “eligible” commercial vehicles in use by the business, and basic information about the vehicles – vehicle type, fuel source, manufacturing year, number of axles and gross weight. Eligible commercial vehicles were defined as vehicles that are owned, leased, or operated by the business that are normally garaged, parked, or based at their address. Vehicles that were driven to and from work by employees (i.e., garaged at an employee’s home) were specifically excluded since they were surveyed as part of the household travel survey.

· Businesses agreeing to participate in the business survey were asked to provide limited information on their commercial vehicles using either computer aided telephone interviewing (CATI) techniques or a vehicle identification form that was sent via facsimile. An option of receiving the information via computer log or printout was acceptable if it was required to secure a survey completion. It was necessary to make follow-up calls in these cases to clarify the information. If the information was obtained using the form, a follow-up telephone call was made to collect the data or arrangements were made for the completed form to be returned via facsimile. The vehicle information form was designed to be straightforward in order to encourage response from businesses with large numbers of commercial vehicles. All of the information requested was necessary, however, to permit estimation of the number of commercial vehicles in the region and expansion of the survey results.

· Commercial vehicles selected for second stage travel survey were selected from the vehicles listed in the first stage survey based on designated sampling procedures. The sampling procedures allowed a single business to have multiple vehicles included in the travel survey. A recruitment telephone call was made to those businesses having one or more survey vehicles. During the recruitment call, the surveyor obtained permission to mail (or fax) the commercial vehicle travel diaries to a contact person. The mailing of the diaries was timed so that they arrived approximately four business days before the assigned travel day.

· A follow-up call or facsimile transmission was made one business day before the assigned travel day reminding the participant of the survey day and date, as well as which vehicle was to be sampled.

· All trips made using the selected commercial vehicle(s) were recorded for an assigned 24-hour period (the travel day). If the business maintained detailed vehicle manifest information, the travel data could generally be obtained from the manifest.

· Approximately one day after the travel day, the business was called and trip diary information was collected using CATI techniques. The surveyor attempted to collect this information directly from the driver, although this was frequently not possible. Survey respondents could also elect to transmit the information by facsimile instead of over the telephone. Follow-up calls were made to collect missing information or correct inconsistent information.

Questionnaire design

Business/Vehicle Survey

Verification of business data included in the DRCOG business file was performed as part of the first stage business survey. This verification portion of the first stage of the survey was performed using CATI and no forms were required. The following business information was verified: name, site address, and activity at the called site (line of business).

In addition to verifying the above information, the number of vehicles owned, leased or otherwise operated by the business and normally garaged, parked, or based at the business address was determined. A vehicle identification form was used to list vehicles owned, leased or operated by the business as shown in Figure 3. The vehicle identification form included the following information:

· license plate number or any other ID · vehicle model year/year of manufacture · type of vehicle (single-unit, combination, car, pickup/panel//) · type of fuel (diesel, gas, or other) · number of axles (2, 3, 4, 5 or more) · gross vehicle weight (less than 8,500 pounds, 8,500-19,500 pounds, 19,500-33,000 pounds, and 33,000 pounds or more)

Gross vehicle weight (GVW) classifications were selected in consultation with the Colorado Department of Public Health and the Environment, Air Pollution Control Division (APCD). These classifications were selected after considering air quality modeling requirements and form limitations.

All forms were printed on 8 ½ " x 11" paper so that they could be easily sent and received by facsimile if necessary. Forms were mailed or transmitted via facsimile based upon the desires of the business contact and the number of vehicles owned by the business. A vehicle type identification sheet was included with the vehicle identification form. Vehicle information was collected via CATI or, at the discretion of the contact person, via mail or fax. Information from forms that were returned via mail or fax was entered into the CATI system, and telephone calls to correct or verify information were made as necessary.

Survey of Commercial Vehicle Travel

The commercial vehicle travel diary is shown in Figure 4. These diaries were distributed by mail to businesses participating in the survey unless the business requested that the information be transmitted via facsimile. The following data were requested on the travel diaries:

· business name, business address, and business type (if other than base) where the vehicle started the day, · vehicle license or identification number (posted on forms mailed to businesses), · whether the driver for each individual trip is from the Denver region, · address (including establishment names) for each place visited on the assigned travel day using the commercial vehicle, · arrival and departure time at each place, · whether or not the engine was turned off at each place visited, · trip purpose or activity at each place visited, · kind of place (land use) at each place visited, · cargo type delivered or picked-up (based on Standard Transportation Cargo Codes) at each place, if applicable.

Figure 3 Vehicle Identification Survey Form

Figure 4 Commercial Vehicle Travel Diary

As with the stage 1 vehicle information, vehicle travel diary data were collected via the telephone and directly entered in the CATI system when possible. If survey respondents preferred, the vehicle travel diaries could be returned via mail or fax. Data from forms that were returned via mail or fax were entered into the CATI system and telephone calls to correct or verify information were made as necessary.

Other survey forms

A survey brochure (Figure 5) and an introductory letter (Figure 6) were prepared in addition to the actual survey instruments shown in Figures 3 and 4. The survey brochure was distributed to businesses receiving the stage 1 vehicle information form by fax and businesses receiving the travel diaries by mail. The introductory letter was distributed along with stage 2 travel diary(s) to businesses having one or more survey vehicles. The letter explained the purpose of the survey, the survey sponsors (DRCOG, CDOT, and RAQC), and a toll-free numbers for data collection (fax) and questions (hotline).

Sample Design and Stratification

The primary sample frame for the first stage business survey included all businesses located in the eight-county region as reported on the DRCOG business file – Adams County, Arapahoe County, Boulder County, Clear Creek County, the City and County of Denver, Douglas County, Gilpin County, and Jefferson County. The sample frame for the second stage vehicle survey was all commercial vehicles owned, leased, or operated by and garaged at businesses participating in the first stage survey. Only vehicles normally garaged, parked, or based at the address of the business contacted were eligible for the second stage survey.

Business/Vehicle Survey

The design of the business/vehicle survey sample began with the regional business file provided by DRCOG. The data provided in the business file included:

· name of business/establishment · business identification number · full street address · latitude/longitude · Federal Information Processing Standard (FIPS) county code · contact name and telephone number · type of business · SIC code · number of employees

The latitude/longitude and FIPS code information was required to ensure that each business selected for the survey could, in fact, be located and geocoded.

Figure 5 Survey Brochure

DENVER REGION 1998 COMMERCIAL VEHICLE TRAVEL SURVEY

You are Part of a Special Group!

Your business has been selected to participate in an important survey to improve transportation in the eight-county Denver region.

Who is being surveyed?

Less than 1 percent of all businesses and agencies in the Denver region have vehicles chosen to be part of this survey project -- that is why your answers are very important and very influential.

Who is sponsoring the project?

This travel survey project is the work of three partners: the Denver Regional Council of Governments (DRCOG), an association of 48 city and county governments in the metro area charged with regional transportation planning; the Colorado Department of Transportation (CDOT), the agency responsible for building, operating and maintaining the major highways in the state; and the Regional Air Quality Council, the agency charged with developing programs to meet air quality standards.

Why is the survey being done?

The region has seen many changes since the last commercial travel survey was conducted in 1971. The region’s population has quadruples, the Interstate System has been completed through the mountains, and double and triple trailers are routinely employed to move goods by truck. The entire Denver region has become a major commercial hub in the nation.

Over the next 25 years, billions of dollars will be spent on highways and roads, public transit, and other transportation work in the eight-county region. We want to be sure these funds are spent on projects to make traveling in the region easier. That’s where you come in.

What will be done with the data?

The data you provide will be summarized to describe commercial vehicle travel in the region today telling us the geographic distribution of trips around the region, travel by time-of-day, and vehicle miles of travel. The survey results will help us understand the varied nature of commercial vehicle operations. This data will be summarized by various cargo and services functions and be used to assess the role of commercial vehicle operations in the region.

Are my answers confidential?

Information your business or agency shares will be kept confidential up to the limits of the law. Nothing will be published that attaches your business name to your responses; reports will only summarize the data.

Your answers will make a difference in directing the future of transportation in the region. If you have questions about this project, please call the toll-free Survey Hotline at 1-888-765-5503.

Thank You

Figure 6 Introductory/Validation Letter

Businesses were selected from this file for the first stage survey using a stratified random sampling technique. The stratified random sampling approach was used because a large majority of the businesses (approximately 80 percent) in the region had less than 10 employees. These businesses were less likely to own a commercial vehicle or, if a vehicle was “owned,” it was probably more likely to be garaged at the business owner’s home (and, thus, included in the household survey). Businesses were classified into one of the following five strata defined by number of employees:

· 1-9 employees · 10-24 employees · 25-49 employees · 50-249 employees · 250 or more employees

Within each stratum the following sampling process was used:

· A sample interval, “n”, was determined by dividing the number of businesses in a stratum by the number of desired telephone samples for the group. · A random number between 1 and “n” was generated to determine the first business to be selected. That business and every “nth” business thereafter were selected for the sample file.

The actual number of samples for each group was determined based on expected completion rates and the expected number of businesses and vehicles required to produce a sample of 750 complete commercial vehicle travel diaries. It was originally anticipated that 7,500 businesses would need to be telephoned to develop a sufficient sample base for the stage two survey of commercial vehicle travel.

The sample list of businesses was divided in replicates that were defined by assigning a random replicate number to each business (by stratum). This removed periodicity problems (e.g., calling certain types of telephone numbers at regular times or intervals) and bias problems (e.g., calling all numbers in a certain exchange and geographic area at one time). Replicates were designed so that each replicate was approximately equal to the number of telephone dialings that could be completed in one day of recruiting. Each replicate was, in effect, a -random sample of the universe of businesses for a stratum. Once a replicate was “opened” for telephone calling, the entire replicate was called to ensure that the biases were not introduced.

Table 3 summarizes the distributions of all businesses in the region including those businesses excluded from the survey universe for various reasons. A total of 578 of the 90,558 businesses listed in the DRCOG business file were excluded from the sample universe because they were involved with the rental of vehicles, or were police, fire, ambulance, bus, taxi, or US Postal Service businesses. The businesses involved with rental of vehicles were excluded from the sample since they operate commercial vehicles that do not necessarily travel within the region (e.g., rental vehicles might be driven to mountain ski areas rather than being used in the region). Police, fire, ambulance, bus, taxi, and US Postal Service businesses operate commercial vehicles that can be efficiently identified and inventoried using alternative means.. Table 3 Number of Businesses by Business Group and Number of Employees

Number of Employees 0 Employee Total Business Group (SIC Codes) 1 - 9 10 - 24 25 - 49 50 - 249 250+ Total Businesses Businesses Production 01 - 39 11,600 1,658 651 467 110 14,486 19 14,505 TCPU/Wholesale1 40 - 51 7,363 1,166 412 306 76 9,323 2 9,325 Retail 52 - 59 10,859 1,781 720 666 64 14,090 12 14,102 Service 60 - 89 42,198 4,013 1,608 1,413 281 49,513 29 49,542 Government 90 - 99 226 133 77 97 63 596 12 608 Classified Businesses 72,246 8,751 3,468 2,949 594 88,008 74 88,082 Unclassified Businesses 432 215 228 607 20 1,502 396 1,898 Total Businesses 72,678 8,966 3,696 3,556 614 89,510 470 89,980 Businesses Excluded from Survey 2 331 98 47 82 14 572 6 578 Total Businesses 73,009 9,064 3,743 3,638 628 90,082 476 90,558 Source: Updated DRCOG business file provided 6/26/98. 1 TCPU/Wholesale - transportation, communication, public utilities and wholesale businesses. 2 Businesses involved in rental vehicles (auto or trucks), police, fire, ambulance, bus, taxi, or US Postal Services (SIC codes 4100-4199, 4300-4399, 4953, 7513, 7514, 7515, 7519, 9221,

In addition to the 578 businesses excluded from the sample universe, 1,898 businesses were excluded from the sample frame (but not the sample universe) because no employees were listed for the business or the business was not classified to an SIC code. Thus, the sample was drawn from the 88,082 businesses in the region with and without telephone numbers in the DRCOG business file. The businesses sampled were intended to represent the sample universe of 89,980 businesses in the region that did not operate vehicles surveyed via other means.

Table 4 summarizes the number of businesses in the DRCOG business file that were lacking business telephone listings. To conserve survey resources, businesses with fewer than 50 employees that lacked a telephone number in the business file were replaced by the next valid business in the business file. This procedure minimized the amount of resources expended searching for telephone numbers. If a business with 50 or more employees and no telephone number listed in the business file was selected for the survey, the business information was transmitted to DRCOG and telephone numbers were determined using telephone directories. Table 5 summarizes the resulting distribution of “eligible” businesses in the sample frame for the business survey.

Table 4 Number of Businesses by Business Group and Number of Employees with No Telephone Number on Business File

Number of Employees Business Group 1 (SIC Codes) 1 - 9 10 - 24 25 - 49 50 - 249 250+ Total Production 01 - 39 2,283 606 278 260 82 3,509 TCPU/Wholesale 2 40 - 51 1,315 421 184 138 59 2,117 Retail 52 - 59 1,881 448 165 166 35 2,695 Service 60 - 89 3,206 903 550 632 205 5,496 Government 90 - 99 128 75 46 52 38 339 Classified Businesses 8,813 2,453 1,223 1,248 419 14,156

Source: Updated DRCOG business file provided 6/26/98. 1 Excludes businesses involved in rental vehicles (auto or trucks), police, fire, ambulance, bus, taxi, or U. S. Postal Services (SIC codes 4100-4199, 4300-4399, 4953, 7513, 7514, 7515, 7519, 9221, 9224, or 9229). 2 TCPU/Wholesale - transportation, communication, public utilities and wholesale businesses.

Table 5 Number of Businesses by Business Group and Number of Employees “Eligible” for Selection1

Number of Employees Business Group2 (SIC Codes) 1 - 9 10 - 24 25 - 49 50 - 249 250+ Total Production 01 - 39 9,317 1,052 373 467 110 11,319 TCPU/Wholesale 40 - 51 6,048 745 228 306 76 7,403 Retail 52 - 59 8,978 1,333 555 666 64 11,596 Service 60 - 89 38,992 3,110 1,058 1,413 281 44,854 Government 90 - 99 98 58 31 97 63 347 Classified Businesses 63,433 6,298 2,245 2,949 594 75,519 Source: Updated DRCOG business file provided 6/26/98. 1 Businesses without telephone numbers fewer than 50 employees were replaced in the sample by the next available business in the sample file. Telephone numbers were obtained from telephone directories by DRCOG staff for businesses with 50 or more 2 Excludes businesses involved in rental vehicles (auto or trucks), police, fire, ambulance, bus, taxi, or US Postal Services (SIC codes 4100-4199, 4300-4399, 4953, 7513, 7514, 7515, 7519, 9221, 9224, or 9229). 3 TCPU/Wholesale - transportation, communication, public utilities, and wholesale businesses.

Table 6 shows the number of businesses that were anticipated to be called (not completed) from each employment size group. The column totals shown in Table 6 were targets only. The expected distributions of businesses by business type within each employment group were based on the relative distributions shown in Table 4. Completed calls by group were expected to be less than the numbers shown, depending on response rates for each group. Overall, it was anticipated that 3,750 businesses, or 50 percent of the telephone numbers called would result in a successful survey (30 percent of the telephone numbers called were expected to be non-working, fax/modem, or disconnected).

Table 6 Expected Number of Businesses to be Called by Business Group and Number of Employees

Number of Employees Business Group1 (SIC Codes) 1 - 9 10 - 24 25 - 49 50 - 249 250+ Total Production 01 - 39 294 334 249 238 93 1,208 TCPU/Wholesale2 40 - 51 191 237 152 156 64 800 Retail 52 - 59 283 423 371 339 54 1,470 Service 60 - 89 1,229 988 707 718 236 3,878 Government 90 - 99 3 18 21 49 53 144 Classified Businesses 2,000 2,000 1,500 1,500 500 7,500 1 Excludes businesses involved in rental vehicles (auto or trucks), police, fire, ambulance, bus, taxi, or US Postal Services (SIC codes 4100-4199, 4300-4399, 4953, 7513, 7514, 7515, 7519, 9221, 9224, or 9229). 2 TCPU/Wholesale - transportation, communication, public utilities, and wholesale businesses.

The target calls shown in Table 6 were based, in part, on assumptions regarding the distribution of vehicles by business size. First, approximately 80 percent of the businesses in the region were in the “1-9 employee” category. Many of these businesses, however, were expected to own few, if any commercial vehicles. In addition, it was assumed that businesses in this category would be more transitory than businesses in other size groupings, resulting in fewer actual contacts per telephone call. Overall, it was assumed that about 25-30 percent of the calls for this group would result in a completed business survey (with or without a commercial vehicle). Thus, about 500-600 completed business surveys should be obtained from this group. With 500 observations, the proportions of businesses in different “classes,” such as businesses owning 0, 1, or 2 or more commercial vehicles, should be able to be determined within ±0.05 at the 95 percent confidence level.

There were only 594 businesses in the 250 or more employee group. Almost all businesses in this group were contacted. Businesses in this group were expected to frequently own or lease commercial vehicles. However, as with the “1–9” group, only about a 25-30 percent response rate was expected. It was assumed that it would be more difficult to obtain permission for the survey from businesses in this group and was likely that some businesses would refuse to participate because of the number of vehicles. While only about 125-150 responses were expected, confidence levels would not be substantially lower than with the “1–9” group due to the impact of “small universe” correction factors for the statistics.

The remaining telephone calls were split semi-proportionately between the remaining three size groups. It was anticipated that businesses in the 10-24, 25-49, and 50-249 employee groupings would be the easiest to contact and the most willing to participate in the survey. It was also anticipated that businesses in these groups would own or lease significant portions of the commercial vehicles that were the subject of this study. Overall, it was anticipated that about 50-75 percent of businesses in these groups would respond to the survey.

Survey of Commercial Vehicle Travel

The first stage survey of businesses/vehicles was completed prior to defining the sample for the second stage survey of commercial vehicle travel. The second stage survey used the list of eligible commercial vehicles drawn from the sample of businesses as a sample frame, again using a stratified random sampling technique.

The commercial vehicle information obtained from the first stage survey was entered into a database of vehicles eligible for survey. The database contained the business name, address, type of business, SIC code, telephone number, contact name, vehicle type, and license plate or other vehicle ID number of the vehicle being surveyed. Vehicles were stratified based on the business type of the vehicle owner and the vehicle type. Businesses were grouped into five strata:

· Production (SIC Codes 1-39), · Transportation, Communications, and Public Utilities/Wholesale (SIC Codes 40-52), · Retail (SIC Codes 53-59),

· Service (SIC Codes 60-89), and · Government (SIC Codes 90-99)

Vehicles were also stratified into five groups based on vehicle type:

· autos, · pickups, , and , · sport utility vehicles, · single-unit trucks, and · combination trucks.

The above groupings yielded 25 cells suggesting a preliminary target of 30 vehicles per cell. However, based on the results of the first stage business/vehicle survey, it was impossible to stratify and select a sample frame that would result in 30 vehicles per cell. Some cells contained very few vehicles (e.g., very few single-unit and combination trucks were owned by government establishments). In addition, in some cells that had reasonably large numbers of potential vehicles to survey, a major portion of the vehicles were owned by relatively few companies. In these cases, selecting the full quota of vehicles to survey (accounting for anticipated response and completion rates) would place an undue burden on selected companies. A rule of thumb was established in the vehicle sample selection process that no more than five vehicles would be selected from any one company. With this information in mind, the sampling procedure was conducted as follows:

· The sport utility vehicles were grouped with the pickup/panel/van and minivan vehicles for each of the five business groups.

· The single-unit trucks and the combination trucks were merged into one category for four of the five business groups. Separate single-unit truck and the combination truck sampling cells were maintained for the transportation/communications/public utilities/wholesale (TCPU/Wholesale) business group.

· The vehicle records from the first stage business survey were subdivided into each of the 25 strata. The vehicle's owner business identification information was retained for each vehicle record.

· Within a stratum, vehicles were sorted by decreasing number of vehicles owned by the business for the stratum. In other words, if a business owned or leased 50 vehicles for a specific stratum, all 50 vehicles were grouped together and listed before the vehicles owned or leased by a business with 49 vehicles. Vehicles were also sorted by vehicle ID (or license plate number) within a specific business for a stratum.

· A sample interval, “n”, was determined by dividing the number of vehicles in a stratum by the number of desired surveys for the stratum (accounting for an assumed agreement rate of 70 percent and an assumed 75 percent completion rate).

· A random number between 1 and “n” was generated to determine the first vehicle to be included in the survey. That vehicle, along with every “nth” vehicle, even if the vehicle was owned by the same business, was included in survey.

In the above procedures, ranking vehicles by size of business (as measured by the number of commercial vehicles) ensured that the small number of businesses with large numbers of vehicles was not under-sampled. Note that the selected vehicles were sorted by business prior to making recruiting calls for the second stage survey. This ensured that a business was not called multiple times to recruit multiple vehicles.

The initial selection of vehicles was reviewed with DRCOG. Based upon this review, it was decided to adjust the sampling process to increase the number of single-unit and combination trucks included in the sample (since those categories of vehicles were most impacted by low numbers of eligible vehicles based on the first stage survey). This was done by relaxing the criterion that a maximum of five vehicles be selected from any one company, and by combining the category with the pickup/van/ minivan category. This combination allowed for the selection of single-unit and combination trucks from some businesses that had originally been “exhausted” for samples originally required for the sport utility vehicle category. As such, single-unit and combination truck sampling cells were merged except for the TCPU/Wholesale trucks.

Table 7 summarizes the eligible commercial vehicles for the second stage survey from the first stage survey, Table 8 shows the number of selected vehicles for the second stage survey and Table 9 shows the expected numbers of completed vehicle travel surveys for each group. The revised sampling process cells are used for display of the data. The expected number of vehicles completing the survey was based on an assumption of a 70 percent response rate to the survey by businesses and a 75 percent completion rate for the selected vehicles.

Table 7 Eligible Commercial Vehicles for the Second Stage Sample of Vehicle Travel

Vehicle Type Pickup/Panel Single-Unit Combination Van/Minivan/ Automobile 2 2 Total Business Group1 (SIC Codes) Sport Utility Truck Truck Production 01 - 39 461 72 205 738 TCPU/Wholesale3 40 - 51 254 64 286 401 1,005 Retail 52 - 59 182 43 71 296 Service 60 - 89 803 236 75 1,114 Government 90 - 97 222 61 47 330 Classified Businesses 1,922 476 1,085 3,483

1 Excludes businesses involved in rental vehicles (auto or trucks), police, fire, ambulance, bus, taxi, or US Postal Services (SIC codes 4100-4199, 4300-4399, 4953, 7513, 7514, 7515, 7519, 9221, 9224, or 9229). 2 Single-unit and combination truck sampling cells were merged except for the TCPU/Wholesale trucks. The total represents all single-unit and combination trucks. 3 TCPU/Wholesale - transportation, communication, public utilities and wholesale businesses.

Table 8 Selected Commercial Vehicles for the Second Stage Sample of Vehicle Travel

Vehicle Type Pickup/Panel Single-Unit Combination Van/Minivan/ Automobile 2 2 Total 1 Truck Truck Business Group (SIC Codes) Sport Utility Production 01 - 39 124 60 141 325 TCPU/Wholesale3 40 - 51 106 57 173 109 445 Retail 52 - 59 109 25 63 197 Service 60 - 89 177 125 65 367 Government 90 - 97 57 44 25 126 Classified Businesses 573 311 576 1,460

1 Excludes businesses involved in rental vehicles (auto or trucks), police, fire, ambulance, bus, taxi, or US Postal Services (SIC codes 4100-4199, 4300-4399, 4953, 7513, 7514, 7515, 7519, 9221, 9224, or 9229). 2 Single-unit and combination truck sampling cells were merged except for the TCPU/Wholesale trucks. The total represents all single-unit and combination trucks. 3 TCPU/Wholesale - transportation, communication, public utilities and wholesale businesses.

Table 9 Expected Commercial Vehicles for the Second Stage Sample of Vehicle Travel

Vehicle Type Pickup/Panel Single-Unit Combination Van/Minivan/ Automobile 2 2 Total 1 Truck Truck Business Group (SIC Codes) Sport Utility Production 01 - 39 65 32 74 171 TCPU/Wholesale3 40 - 51 56 30 91 57 234 Retail 52 - 59 57 13 33 103 Service 60 - 89 93 66 34 193 Government 90 - 97 30 23 13 66 Classified Businesses 301 164 302 767

1 Excludes businesses involved in rental vehicles (auto or trucks), police, fire, ambulance, bus, taxi, or US Postal Services (SIC codes 4100-4199, 4300-4399, 4953, 7513, 7514, 7515, 7519, 9221, 9224, or 2 Single-unit and combination truck sampling cells were merged except for the TCPU/Wholesale trucks. The total represents all single-unit and combination trucks. 3 TCPU/Wholesale - transportation, communication, public utilities and wholesale businesses.

Survey Publicity

The commercial vehicle survey was publicized to local agencies and citizens by DRCOG in several ways. First, a news release was sent to local governments, law enforcement agencies, and news organizations. The news release provided general information regarding the survey as well as persons to contact at DRCOG for additional information. The news release was also sent to the Colorado Motor Carriers Association (CMCA). The CMCA published the news release in their official publication, “The Colorado Carrier”. The second method for publicizing the survey was to establish a hotline to provide “on- line” information about the travel survey. The hotline provided a series of recorded

messages regarding the commercial vehicle survey and other surveys included in the TBI. When a person called the hotline, they heard a general message and were then provided with options to listen to recorded messages about specific surveys and topics, to record a message for DRCOG, or to talk with a staff member if the call was made during working hours.

The news release and the script for the recorded hotline messages are shown in Appendix A.

Commercial Survey Pilot Test

Pilot Test Procedures

A pilot test of the commercial vehicle survey was performed during May and June, 1998. The objective of the pilot test was to apply all aspects of the commercial vehicle survey process to determine any changes or corrections necessary for the success of the full survey.

The pilot test used a random sample of businesses obtained from the business file provided by DRCOG. During the pilot test, 200 businesses were sampled from the DRCOG business file. Based upon previous experience, it was expected that approximately 20 completed commercial vehicle origin-destination surveys would result from a sample of 200 businesses. Table 10 summarizes major dates in the administration of the pilot test.

Table 10 Commercial Vehicle Survey Pilot Test Timetable

Milestone Date (1998) Authorization to Proceed with Pilot Test April 21 Pull Sample of Businesses for Stage 1 Survey May 8 Stage 1 Business Survey May 12-19 Pull Sample of Vehicles for Stage 2 Survey May 26 Stage 2 Vehicle Recruiting May 27-28 Mail Survey Materials May 28 Reminder Calls June 3 Survey Day June 4 Survey Data Collection June 5-10 Clean and Edit Data June 10-16 Geocoding June 8-11 Hard to Geocode File Provided to DRCOG June 12 DRCOG Geocoding June 12-17 End of Pilot Test Data Collection June 17

Business/Vehicle Survey

A sample of 200 businesses was selected from the source file for the pilot test using the stratified random sampling approach. This procedure stratifies the sample business based on the number of employees. During the conduct of the stage 1 business survey for the pilot test, it became apparent that the relatively high proportion of “dead” numbers would result in only a very limited test of the stage 2 survey instruments and procedures.

To ensure that the test results for stage two would be meaningful, a second selection of 200 businesses was performed using the same sampling interval used initially. Specifically, each business in the sorted list following a business selected for the primary selection was selected for the secondary selection. Of the second 200, a final disposition was reached for 198 businesses. The remaining two were dropped without resolution. The findings summarized in this section are based on the results of the combined selection of 398 businesses obtained from the original source file. Commercial vehicles obtained from businesses in the second selection were used, along with those from the first sample, to test stage two of the survey.

An average of 1.8 dials per number was required to reach “final” business survey results. The final disposition of calls for the business survey is shown in Table 11.

Overall, 35 percent of the numbers attempted were working businesses with no qualifying commercial vehicles. This represented approximately 50 percent of the known universe of businesses. Approximately 8.8 percent of the telephoned numbers attempted were working businesses that claimed to have commercial vehicles (35 businesses). This represented approximately 12.5 percent of the known universe of businesses. All but four of the businesses reporting commercial vehicles supplied full information about their vehicles and completed the screening portion of the study. Eligible businesses had, on average, 3.6 commercial vehicles. The disposition of calls showed that the original business selection source file provided by DRCOG and used for the pilot test has a high proportion of numbers (44 percent of the combined sample) which were not valid for the survey.

Survey of Commercial Vehicle Travel

A stratified random sampling procedure was used to select commercial vehicles for stage two from the list of eligible vehicles obtained through the survey of businesses. Because of the pilot test’s relatively small sample size, the business type-vehicle type stratification suggested for the full survey was aggregated into a total of five cells. Table 12 shows the sample stratification utilized in the pilot test. The aggregation of cells for the pilot test was performed before the decision to combine the sport utility vehicle and pick-up/panel van/minivan groupings of vehicles.

Table 11 Pilot Test Business Survey Disposition Report

Percent of Initial Sample Additional Sample Combined Sample Known Disposition of Phone Numbers Business Number Percent Number Percent Number Percent Universe No Answer 5 2.5 4 2.0 9 2.3 N/A Busy 2 1.0 2 1.0 4 1.0 N/A Answering Machine 3 1.5 0 0.0 3 0.8 N/A Fax 5 2.5 1 0.5 6 1.5 N/A Callback 0 0.0 1 0.5 1 0.2 N/A Not In Service/Disconnect 33 16.5 35 17.7 68 17.1 N/A Reassigned (non-business) 9 4.5 14 7.1 23 5.8 N/A Reassigned (business)* 40 20.0 19 9.6 59 14.8 21.2 Language Problem 0 0.0 1 0.5 1 0.2 N/A No Responsible Person Available 3 1.5 1 0.5 4 1.0 N/A Subtotal - Number Not Usable 100 50.0 78 39.4 178 44.7 21.2 Initial Refusal* 13 6.5 20 10.1 33 8.3 11.8 Refused to Participate* 11 5.5 1 0.5 12 3.0 4.3 Subtotal - Unable to Survey 24 12.0 21 10.6 45 11.3 16.1 No Commercial Vehicles* 61 30.5 79 39.9 140 35.2 50.2 Have Commercial Vehicles* 15 7.5 20 10.1 35 8.8 12.5 Subtotal –Surveyed 76 38.0 99 50.0 175 44.0 62.7 Total 200 100.0 198 100.0 398 100.0 100 *These dispositions are included in the known universe of businesses. N/A=Not Applicable

Table 12 Commercial Vehicle Sample Stratification Used in the Pilot Test

Vehicle Type Pickup/Panel Sport Single-Unit Combination Van/Minivan/ Utility 1 Truck Truck Business Group (SIC Codes) Automobile Vehicle Production 01 - 39 12 TCPU/Wholesale2 40 - 51 Retail 52 - 59 6 8 12 Service 60 - 89 12 Government 90 - 97

1 Excludes businesses involved in rental vehicles (auto or trucks), police, fire, ambulance, bus, taxi, or US Postal Services (SIC codes 4100-4199, 4300-4399, 4953, 7513, 7514, 7515, 7519, 9221, 9224, or 9229). 2 TCPU/Wholesale - transportation, communication, public utilities and wholesale businesses.

A detailed disposition of calls was not obtained during vehicle recruitment for the pilot test due to a problem with the call form. Only final call results were captured on the form. The form was corrected by forcing the interviewer to enter the proper dialing result for each attempt before moving to the next record/business.

Fifty vehicles were selected for the survey from the vehicles reported by the 35 businesses from which vehicle information was collected. Twenty-nine businesses totaling 41 vehicles agreed to participate in the survey during the vehicle recruiting stage. Two of the 29 businesses, each with one vehicle, refused to participate in the vehicle survey when the reminder call was made. Their reasons for not participating were as follows:

· Short of help and no time to fill out diaries · Too many emergencies...boss drives truck and has no time to track it.

A final disposition of calls for the stage 2 vehicle travel survey data collection is contained in Table 13.

Table 13 Disposition of Data Collection Results

Number of Percent of Number of Percent of Disposition Businesses Businesses Vehicles Vehicles Refused to Complete 1 3.5% 1 2.4% Refused at Reminder Call 2 6.9% 2 4.9% Timed Out* 1 3.5% 2 4.9% Completed Survey 25 86.1% 36 87.8% Total 29 100.0% 41 100.0% *Business was required to complete information on 2 vehicles and was only able to provide information on 1 of those vehicles. This business was considered “Timed Out” (for both vehicles) in this report.

For the full survey, all vehicles that provide the requested information were considered complete, regardless of whether the business provided information for all recruited vehicles.

Completed vehicle callbacks were achieved with 25 of the 27 businesses remaining after the reminder calls, totaling 36 commercial vehicle survey diaries completed. The 36 vehicles that completed the survey visited 156 places to be geocoded for an average of 4.3 places per vehicle. Of the 156 places, 63 were recorded as “Base”. A summary of the geocoding productivity is provided in Table 14.

Table 14 Summary of Vehicle Trip Geocoding Productivity Geocoding Productivity Number Percentage Address was company base 63 40.40% Address matched exactly 16 10.20% Address matched exactly after editing 54 34.60% Address geocoded by DRCOG 21 13.50% Address could not be geocoded 2 1.30%

Throughout the full survey, the quality control procedures outlined in the Surveyor Manual (Appendix B) and the Editor/Coder Manual (Appendix C) were followed. This included returning for clarification surveys with addresses that can not be geocoded. Addresses that were not further clarified or geocoded were forwarded to DRCOG.

Findings

Selection

A primary finding of the pilot test was that the original business file provided by DRCOG, based on fourth quarter 1996 Dun & Bradstreet information, had a much lower productivity than was originally estimated. The implication of this lower productivity was that more businesses would need to be included in the selection in order to achieve the 750 completed vehicle surveys. The pilot test revealed the following differences between assumptions used to estimate the cost of the survey and actual productivity of the business sample:

· The number of “dead” telephone numbers in the selection, particularly those that were not in service, disconnected, or reassigned, was much higher than originally anticipated. In total, 44 percent of the numbers in the selection were “dead” numbers, compared to the original estimate of 10 percent “dead” numbers.

· The average number of commercial vehicles associated with a particular business was lower than expected. Based on pilot test observations, each qualified business had on average 3.6 eligible commercial vehicles compared to the estimated average of 4.8 vehicles.

Despite the problems with the selection, the recruiting rate for the pilot test was higher than originally estimated. During the pilot test, 83 percent of qualifying businesses agreed to participate in the commercial vehicle survey compared to the 50 percent

originally estimated. Unfortunately, the higher recruiting rate by itself was not sufficient to offset the lower than anticipated productivity of the selection.

The cause of the low selection file productivity was the age of the file. The information contained on listed businesses was at least two years old at the time of the pilot test. The overall implication was that either a larger selection needed to be made for the full survey (to compensate for the higher proportion of “bad” numbers), or a new source for selecting businesses with telephone numbers needed to be obtained.

After further discussion between the consultant team and DRCOG, it was decided that the business selection could be drawn from a recently completed file containing Dun & Bradstreet information updated with ES202 data1. Before recommending that the revised file be used for the full survey, however, an investigation of the productivity of the file was made. The “dead” number problem was assessed by dialing 200 randomly selected businesses telephone numbers to determine the percentage of disconnects and reassign- ments. A revised proportion of not in service/disconnected numbers was obtained. An estimate of the proportion of fax/modems numbers was also made. Since the pilot test results generally indicated that the survey instruments and procedures were adequate, the test was limited to the productivity of the selection. The result of dialing 200 numbers showed that the updated file, particularly with regard to the proportion of reassigned business numbers, had an improved productivity. After further analysis, it was concluded that the productivity of the improved file was sufficient to achieve the desired number of completed vehicle surveys if the original assumptions regarding selection size were used.

Table 15 contains a summary of the original assumptions, observed results, and the estimated results for the full survey assuming that the revised file was used. The first two rows of the table, “Business Survey Calls” and “Business Recruitment Calls” show the expected dispositions of telephone numbers (i.e., businesses). The third row, “Vehicle Data Collection” shows the expected disposition of commercial vehicles recruited for survey.

Survey Instruments

Survey instruments used in the Commercial Vehicle Survey included a Survey Brochure, an Introductory Letter regarding the purpose of the survey, a Vehicle Information Form, and a Trip Diary. The Survey Brochure was distributed to businesses receiving the Vehicle Information Form by fax and businesses receiving the travel diaries by mail. No changes to the brochure or its procedure for distribution were necessary.

1 Colorado Department of Labor database containing quarterly business employment data.

Table 15 Commercial Vehicle Survey - Disposition of Calls Comparison

Telephone Numbers Called 1 Disposition Original Estimate Observed During Disposition of Calls Revised Estimate 2 3 4 Pilot Test Full Survey Pilot Test with New File for Full Survey

Stage 1 Business Survey Calls Disconnected/Non-Working/Fax/Modem 20 (10%) 2,250 (30%) 178 (44.7%) 32 (27.8%) 2,775 (37%) Refusals 50 (25%) 1,500 (20%) 45 (11.3%) 15 (13.0%) 600 (8%) Completed Business Survey – no vehicles 90 (45%) 1,750 (23%) 140 (35.2%) 44 (38.3%) 3,225 (43%) Completed Business Survey – vehicles 40 (20%) 2,000 (27%) 35 (8.8%) 24 (20.9%) 900 (12%) Total 200 (100%) 7,500 (100%) 398 (100%) 115 (100%) 7,500 (100%)

Stage 2 Business Recruitment Calls Refusals 20 (50%) 1,000 (50%) 6 (17.2%) N/A 270 (30%) Recruited 20 (50%) 1,000 (50%) 29 (82.8%) N/A 630 (70%) Total 40 (100%) 2,000 (100%) 35 (100%) N/A 900 (100%)

Stage 2 Data Collection Calls (Vehicles) Completed Vehicle Survey 30 (75%) 750 (75%) 36 (87.8%) N/A 750 (75%) Refusals/Timed Out/Dropped Out 10 (25%) 250 (25%) 5 (12.2%) N/A 250 (25%) Total 40 (100%) 1,000 (100%) 41 (100%) N/A 1,000 (100%) 1 Proportions used in initial costing 2 Based on the original Dun & Bradstreet File 3 Revised file of Dun & Bradstreet and ES202 data 4 Assumes revised file was used for full survey

An Introductory Letter was distributed along with travel diary(s) to businesses having one or more survey vehicles. The letter explained the purpose of the survey, who was sponsoring the survey (DRCOG, CDOT, and RAQC), and also provided toll-free numbers for data collection (fax) and questions (hotline). Based on the results of the pilot test, no changes to the introductory letter or its distribution procedure were necessary.

Due to time constraints during the pilot test, a validation letter was not used. The purpose of such a letter was to increase survey participation levels by reinforcing that it was legitimate. After consideration, it was decided that the introductory letter would not be appropriate for use as a validation tool during the business survey since it contained references to the follow-up commercial vehicle trip survey. Many businesses did not have eligible vehicles, and references to the vehicle trip survey might have discouraged some businesses from participating.

For the full survey, it was recommended that a validation letter be available for use during the business survey. The letter was drafted by the consultant and was reviewed by DRCOG. Since good survey practice dictates a validation letter be available should there be questions regarding the credibility of the survey, a validation letter (Figure 5) was, therefore, distributed to businesses that required further justification.

The Vehicle Information Form was used during stage 1 of the survey to record information vehicles owned, leased, or operated by the business. The information necessary to complete the form was obtained either via the telephone, computer, or fax. Two minor changes to this form were suggested.

During the pilot test, Jefferson County Airport was selected. The airport owned and operated several vehicles on-site including vehicles such as fork-lifts, dump trucks, snow plows, and tractors. Although these vehicles could be driven off-site during emergencies, they were typically used on-site. After further discussion with the business contact, these vehicles were excluded from the survey. Because this circumstance could be repeated during the full survey, it was recommended that the instructions for the Vehicle Information Form be revised to state the following:

“Please fill out the information below about the commercial vehicles owned, leased, or operated by you and usually garaged, parked or based at the address noted above. Exclude company vehicles NOT typically parked at this place during non-working hours. Also exclude those vehicles that are generally ONLY used on-site.”

The second recommendation was that the Gross Vehicle Weight (pounds) categories be corrected to remove overlaps. The revised categories were corrected to:

· Less than 8,500 · 8,500-19,499 · 19,500-32,999 · 33,000 or more

The second form, the Trip Diary, was mailed to businesses with vehicles selected for survey. The diary included provisions for 11 places. The observed average number of places visited in the pilot test was 4.3 places per commercial vehicle and ranged from 0 to 14 places. No changes in the number of places were suggested.

It was suggested that the box for the begin place (Place 0) be revised slightly for further clarification. “The vehicle trip began at:” was revised to “Where was this vehicle at 4:00 AM on the survey day?” In addition, the option “Vehicle not used today (return form)” was then moved below the line “What time did the vehicle leave this place?”

Also on the Trip Diary, it was possible that a respondent could select cargo type “Other” for both pick-up and delivery. However, there was only one blank available for the respondent to specify an answer. It was recommended that the form be revised so that there are two rows, one with a box under Pick-up for “Other (Specify)” and, on the row beneath, a box under Delivery for “Other (Specify)”. If there was not sufficient space on the form, then the interviewer would probe for the correct response. On fax return diaries, a follow-up call was necessary if both “Other” boxes were checked to determine which of the specified cargo types were pick-up and which were delivery.

The Trip Diary used for the pilot test also included the following question: “What is the home county and state of the driver?” The intent of this question was to provide further insight into the potential of double-counting trips that were recorded as part of the household survey. Furthermore, asking this question provided a means to evaluate the magnitude of double-counting and provided a method of correcting for double-counting of trips during model development. Review of the responses to this question from the pilot test showed that all surveyed vehicles had drivers from within the region. Furthermore, the driver’s county of residence was not observed to change during the vehicle’s travel day. This question was still useful for expanding the total of trips made in commercial vehicles (for personnel not involved in transport for a livelihood) from the household survey. This can be verified/contrasted with trips of this sort from the commercial survey. If they did not match relatively closely, one might be “under-reported” (i.e., this is a check for non- response bias). It was recommended that this question be retained in the full survey.

Survey Procedures

Although the survey procedures were generally successful during the pilot test, a few clarifications and modifications were suggested. A summary of the situations either encountered or anticipated along with a clarification on the correct procedure to be applied are shown below:

Situation: A company is selected to have multiple vehicles participate in the study but feels that the number of vehicles we would like to participate is too much of a burden.

Question: Is that company considered a refusal? Or can we ask the company how many vehicles they feel they can handle and allow them to participate? (Of course, replacing “lost” vehicles from the universe not previously selected.)

Answer: If a company offers to provide survey information for fewer than the selected number of vehicles we should agree to accommodate their request. We are responsible for designating which of their originally selected vehicles are to be surveyed and not accept any substitutions (to prevent business from suggesting only the seldom used vehicles to further reduce reporting burden or heavily used vehicles to make it “worth our while”). Since we are actually interested in surveying vehicles, completes should be measured per vehicle, not business. However, in order for a vehicle to be considered complete, all of the business survey information must be provided.

Situation: One of the recruited vehicles was a which means that most of their trips were to locations “on the side of the road”

Question: How should we handle vehicles that do not go to a specific address such as a construction site or bridge? To what level of detail do we need to obtain address information under these circumstances?

Answer: In these cases, we should attempt to geographically pinpoint the location as closely as possible. Mile markers, cross streets or distances from nearest place, exit, cross-street should be obtained. This will permit DRCOG to interactively geocode the location.

Situation: During the pilot test, one company provided complete trip information for only one of the two vehicles they agreed to have surveyed during recruitment.

Question: Do all vehicles recruited from a company have to complete the study in order for any of the company’s vehicles to be considered complete?

Answer: No. The goal of the survey to obtain 750 completed vehicle surveys. There are no requirements on the number of businesses. Therefore we recommend that business be categorized into “complete” (all recruited vehicles reporting), “partial” (some recruited vehicles reporting), or “refused” (no recruited vehicles reporting). However, we are concerned that businesses may be more likely to refuse to complete the trip survey for vehicles making many trips. Therefore, we will collect

information to evaluate the potential bias by asking additional probing questions.

Q1. “Was the other vehicle used on the survey day?” (If response is “Yes”, ask Question 2). If response is “No”, interviewer should ask where the vehicle was at 4:00 a.m. and complete the travel diary in its entirety.

Q2. “How many places did the other vehicle visit on the survey day?” [Record answer]

Q3. “Why is this a refusal?” with the following answer options: 1. Too busy to keep diary 2. Lost diary or diary not available 3. Driver forgot to keep diary 4. Too many different drivers during the day 5. Other (specify)

Situation: A business that was surveyed had an incorrect address in business file.

Question: Would you like to have the original address information collected at the time of the callback as well as any corrected address information required for geocoding? Currently, the file formats only specify one address for each company/place.

Answer: Yes, we will need both fields for both the original address and the corrected address.

Situation: Suppose a company terminates the business survey prior to answering the question regarding the number of vehicles. Also suppose a company responds to the question regarding the number of vehicles but does not follow-up with vehicle data.

Question: What is the disposition of the company in these situations?

Answer: The following contains a summary of possible business dispositions:

1. Initial Refusal: The company refused to answer any questions during screening.

2. Refuse To Participate: The company did not respond to all questions during screening, or did not provide information for some or all vehicles.

3. Complete, Business Screened: The company responded to all questions during screening, and provides information on all vehicles (if applicable).

Non-response to individual questions did not appear to be a concern based on the pilot test. Based upon the observed responses there were a couple of questions that require further clarification:

Number of Axles

There could be some confusion on how the number of axles is measured. The following definition was provided to surveyors:

Number of axles for combination vehicles refers to both tractor and trailer(s) if the tractor is currently hooked to a trailer. If the tractor is not currently hooked to a trailer, number of axles refers to the “normal” configuration for the tractor when pulling a trailer.

Gross Vehicle Weight (GVW)

The following definition of gross vehicle weight was provided to surveyors:

Gross vehicle weight is the maximum allowable, fully laden weight of a truck including its payload. This is the most common classification scheme used by manufacturers and states.

No changes to the forms or questions beyond those identified previously were recommended.

Survey Administration

The full commercial vehicle survey was administered between July 1998 and September 1998. Data cleaning, including finalizing geocoding, for the main portion of the commercial vehicle survey took place between September 1998 and April 1999. Data summaries and analysis were performed during the remainder of 1999. Table 16 shows a summary schedule for the conduct of the full survey.

Survey Quality Control

Quality control was a primary concern throughout the survey process. In summary, the following quality control efforts were employed as part of the survey:

· Prepared a surveyor/recruiter manual (see Appendix B), · Prepared a editor/coder manual (see Appendix C), · Performed surveyor training prior to the initiation of the survey, including mock surveys, · Supervised and monitored surveyors throughout the survey effort by quality control staff,

· Consultant and DRCOG staff had the capability to monitor survey calls from off-site locations, · Trained editors manually reviewed survey data, · Used computer program specifically written for the commercial survey to check the survey data, and · Prepared weekly summaries of survey progress and results. Table 16 Commercial Vehicle Survey Timetable

Milestone Date (1998) Authorization to Proceed on Full Survey Business Sample Selection List of Businesses Without Telephone Numbers Provided to DRCOG Early July, 1998 DRCOG Provides Telephone Numbers to Consultant Training Business Surveys Mid- to Late July, Business Survey Data to DRCOG [not required prior to starting vehicle survey] 1998 Commercial Vehicle Sample Selection Training Vehicle recruitment calls Electronic address information to PTG August to Mid- Mailout Survey Packets September, 1998 Reminder Calls Conduct callbacks Finalize Editing/Cleaning of Survey Data Finalize Geocoding September, 1998 Transmit data to PTG to April, 1999 Transmit data to DRCOG

Information collected during the recruiting calls was summarized to guard against biases in the sample selection. Several data items collected were important in this checking, including business type, business address, business size (based upon employees), and number and type of commercial vehicles. The following weekly summaries were prepared during the business/vehicle survey:

· telephone calls by outcome by dialing attempt, · final disposition of telephone calls, · number of businesses by SIC group and number of employees, · number of vehicles reported by businesses by SIC group and number of employees, · average number of vehicles reported by businesses by SIC group and number of employees, · number of vehicles reported by businesses by SIC group and vehicle type, · number of vehicles reported by businesses by vehicle type and number of axles, and · number of businesses, number of vehicles reported by businesses, and average number of vehicles reported by businesses by county.

During the commercial vehicle travel survey, the following weekly summaries were prepared:

· number of vehicles surveyed by SIC group and number of employees of reporting business, · number of vehicle trips by surveyed vehicles by SIC group and number of employees of reporting business, · vehicle trip rates per vehicle by SIC group and number of employees of reporting business, · number of vehicles surveyed by SIC group of reporting business and reported vehicle type, · number of vehicle trips by surveyed vehicles by SIC group of reporting business and reported vehicle type, · vehicle trip rates per vehicle by SIC group of reporting business and reported vehicle type, · number of vehicles surveyed by vehicle type and number of axles, and · number of vehicles surveyed by county of reporting business.

The business address was used to monitor for biases in the geographic distribution of businesses throughout the region. This monitoring was performed for two different levels of geography: county, and Regional Statistical Area (RSA). There are 54 RSAs defined by DRCOG for statistical and summary purposes in the eight-county area. Summaries to RSAs were made by DRCOG. The summary by county was performed easily without geocoding based on information contained in the business file.

Survey completion criteria were driven by the need to collect high quality data. The following criteria were used:

· Trip information had to be reported for a commercial vehicle to be included in the survey. Data were not imputed for missing vehicles.

· Business type, vehicle ownership, vehicle classification, and business geographic data had to be reported.

· A major emphasis was placed on obtaining geographic location data for trips adequate to ensure that the locations of the trip ends could be geocoded to latitude and longitude.

· While emphasis was placed on the collection of high quality travel data, the collection of full trip information was not a completion criterion. At the minimum, geographic location of the trips at the origin and destination were obtained. Since trip travel time information is generally obtained from computer representations of the transportation network for model calibration purposes, missing data for that item did not result in the vehicle survey being considered incomplete.

Data Coding and Editing

There were three aspects to data coding for the commercial vehicle survey:

· converting open-ended questions to alphanumeric codes · survey editing and correcting for illogical answers, and · survey geocoding.

Each of the surveys was designed to be self-coding to the greatest extent possible. Forms sent to respondents had lists of possible responses with check-off boxes for most questions. Some of the check-off questions included an “other” category for responses that, in the opinion of the respondent, were not covered by the listed responses. The respondents were asked to give a brief description of their response if “other” is selected. Surveyors were instructed to record the “other” category and the description; data editors had the option of converting the response to one of the listed responses based on their review of the description.

As described in the survey design section, surveyors directly entered the data into a CATI system either when they collected the information from survey respondents over the telephone or from survey forms returned via fax or mail. In order to reduce the amount of manual editing, the CATI system included logic to test for responses that were outside of valid ranges. Data editors worked directly with data files and printed reports from the data files.

Geocoding is the process of converting address information to latitude and longitude. The consultant geocoded all data, including trip-end addresses. Assistance in coding hard-to-find addresses was provided by DRCOG. Final procedures mimicked those used for the household survey. Performing the geocoding soon after the collection of the survey data enhanced the likelihood of obtaining addresses that could be geocoded since this allowed for follow-up calls to respondents to clarify responses.

Exact address matching was performed where possible. In the survey, trip-end addresses were reported as an exact address, intersecting streets if exact addresses were not known, or, at a minimum, place names. Garage was also an acceptable address since the address information was available from the owner garage (“home”) data. Place names, such as McDonald’s, King Soopers, or the Denver Regional Council of Governments, were requested for all non-”home” addresses. Thus, exact addresses could be determined from telephone listings for locations identified by intersecting streets. It was anticipated, however, that the commercial vehicle drivers would be able to provide very accurate address information from their driver logs. The geocoding technique used for each address was recorded. Possible geocoding techniques included:

· exact address match from respondent report, · exact address match from telephone book look-up of respondent information, · match to intersecting streets, · geocode determined by “pointing” to location within the GIS based on “directions” provided by the respondent (e.g., “four miles up Turkey Creek Canyon from the Tiny Town turn-off”), and · insufficient information to geocode address.

Use of one of the first two techniques was emphasized. If an exact match could not be found based on the address coded using one of the first two techniques, match criteria were relaxed. The following relaxation sequence was used: (1) nothing relaxed (exact match) (2) suffix direction only relaxed (3) street type only relaxed (4) prefix direction only relaxed (5) suffix direction and street type relaxed (6) suffix direction and prefix direction relaxed (7) street type and prefix direction relaxed (8) suffix direction, street type, and prefix direction relaxed (9) zip code only relaxed

Geocoding efforts were as follows:

(1) The consultant posted “fill” address for geocoding based on reported information including look-up of address based on establishment name. (2) The consultant geocoded the address from (1). (3) Addresses were checked for non-geocoded records and steps (1) and (2) were repeated. (4) The consultant used the “relaxation” sequence noted above. (5) Ungeocoded records were transmitted to DRCOG for geocoding.

Automated Data Editing

Two SURVCHK computer programs were written for the commercial vehicle survey. The first, which was run during the initial survey of businesses, checked the stage 1 business and vehicle survey data. The second, which was run during the survey of vehicles from selected businesses, mainly checked trip data, but referred back to the business and vehicle survey data for consistency checks. The programs performed the following types of data consistency checks:

· range checks and intra-record checks on business data (business survey) · range checks and intra-record checks on vehicle roster data (business survey) · inter-record checks comparing business and vehicle roster data for consistency (business survey) · range checks and intra-record checks on vehicle trip data (vehicle survey) · inter-record checks comparing vehicle trip data with business and vehicle roster data for consistency (vehicle survey)

For the business survey SURVCHK, if any part of the data for a business failed an edit check, all business and vehicle information for that business was rejected. Only businesses with no errors in any of the data were accepted as clean data. Warnings were also printed for data items that were questionable. Businesses with warnings were accepted as clean.

For the vehicle survey SURVCHK, if any part of the data for the vehicle trips or inter- record checks against business and vehicle data failed, all vehicle trip information for that vehicle was rejected. Note that the business and vehicle data from the business survey were not rejected, since they had already passed the business survey SURVCHK. Only vehicle trip data with no errors in any of the data were accepted as clean data. Warnings were also printed for data items that were questionable. Vehicle trip records with SURVCHK program warnings were accepted as clean. Appendix D contains the business survey data checks and the commercial vehicle travel survey data checks.