U.S. Department of Transportation Roadway Transportation Data Business Plan (Phase 3): Data Business Plan Development for State and Local Departments of Transportation
Appendix E. Survey Practices
Pilot sites used a stakeholder survey to obtain information needed to conduct the assessment. Surveys can be a very effective tool for a successful development of a data business plan, helping with:
- Setting the right mindset for stakeholders and get conversation started regarding current situation and needs.
- Gathering information regarding what types of data are collected or used by the agency and by whom.
The first bullet can be addressed by having an inviting introduction that speaks to the importance of a DBP and encourages the survey taker to envision a future where data integration and collaboration is streamlined. Lead agencies may find it helpful to encourage stakeholders to get together to submit one survey for a particular office or division. For the MARC pilot, several divisions decided to do this voluntarily in order to submit comprehensive responses that are representative of the entire division.
There are two crucial aspects the survey designer can tackle to address the second bullet:
In developing the survey, it is important to make a distinction between limited-choice questions and open-text ones. Open-text questions offer respondents the flexibility and freedom to use their own words in answering questions. By the same token, however, they will generate a set of virtually unique answers for each question, making the task of a systematic analysis much harder to conduct.
By design, limited-choice questions offer a much cleaner set of data for survey analysis. They are concrete, avoid misinterpretation from either end, and prevent unusual or unexpected answers, thus enabling quantitative analyses on the dataset.
For the purposes of a DBP survey, it is best to restrict the amount of open-text questions to a minimum, and reserve them for qualitative, exploratory questions (which are certainly very important). Limited-choice questions should be used whenever there are a finite set of anticipated answers. If there is a mostly finite set of anticipated answers but there may be the possibility of a response outside that set, the question can simply have “other” as the last option, accompanied with a textbox to specify.
Another lesson learned is that surveys should avoid asking questions that may refer to different data types. For instance, a question asking to “rate the dataset(s) you work with” to a stakeholder that collects data for a bike/pedestrian dataset but uses vehicular volume data will be unclear to the respondent, and the answer unclear to interpret. Instead, it may be helpful to provide answer grids, such as the one shown below. This will allow respondents to give different answers to each data type, and will enable the survey analyst to evaluate answers based on multiple criteria, such as data type, or interaction with data (e.g., collectors indicating data quality is good but users indicating otherwise). An equally valid alternative is to ask the same questions for each data type that the respondent works with.
|I am responsible for collecting or updating the data||I use and/or analyze the data||I generate metadata and/or resolve data quality issues||I am an IT professional responsible for technical application||I am an administrator and/or designer for databases and systems||Other (please specify)|
|Transit On-time arrival|
|Transit Speed or travel time|
|Vehicular Speed or travel time|
|Freight Speed or travel time|
Additional lessons learned on the stakeholder survey include the following:
- The survey language and terminology should fit the local construct.
- They survey design should account for the wide array of potential stakeholder roles. For example, survey questions will differ depending on whether the stakeholder is an owner or user of mobility data. Could add something about tailoring the language to fit with the client. And also doing some thinking at the outset on the wide array of users that you should include—doing that thinking up front allows us to be more specific with our questions.
- Lead agencies should test the survey questions to make sure they make sense from the perspective of different stakeholder roles.
- Should strive to strike a balance between not demanding too much time/effort from respondent and at the same time collect valuable info.
- It is best to keep open-text responses at a minimum.
- They survey should include questions to assess staff expertise and available resources.
- They survey should include a question that asks if data is publicly shared.
- Pilot sites found the survey helpful to take stock of current.
- Stakeholder willingness to participate.