[CTPP] Census Response to Chuck Purvis's message

Last week, Chuck Purvis wrote a long, useful message to all who are working on the Work-UP program to help improve the workplace coding of the Decennial Census. We at the Census Bureau have some additional thoughts to share with you. Chuck wrote: 1. The TIGER files *can't* be corrected as part of this WORK-UP project. So, if the TIGER/ArcView files are missing key streets, or address ranges on streets, then I think the best solution is to code to the closest available block-face. The idea is to get the employer location accurately coded to the nearest census block. Once the Census Bureau is able to geocode to the nearest block, then they will be able to aggregate to the CTPP Travel Analysis Zones created as part of TAZ-UP. If, on the other hand, the employer is within a large new subdivision that *isn't* included in the current set of TIGER/ArcView files, then I don't know what happens. I haven't come across this problem, though I have had problems with streets-without-address ranges. Census Bureau Response: Item #1: You are correct in saying that these TIGER files cannot be corrected. The files that are being used as a part of Work-UP are Tiger Line 98. In the past year, a lot of work has been done to improve TIGER. Most of these changes are now in 'live' TIGER and will be reflected in Tiger Line 99 and 00. The Census Bureau is not prepared to process any changes that result from the Work-UP program. The Work-UP software is designed to allow you to map spot employers, taking into account insufficient address coverage and missing streets. As long as you keep the correct address we can match respondents to that employer and then they will be coded to whatever Tiger Line segment that you put them on. In cases where the street is missing just place the employer on a nearby street. BUT YOU NEED TO MAKE SURE THEY ARE STILL WITHIN THE SAME TAZ!!!!! As long as the line segment you associate the employer with is in the same TAZ your work place data and flows will be correct. Chuck wrote: 2. The alias street name problem is one I've been battling with as early as this morning. The TIGER file will probably have the older name (e.g., North Main Street), but the "newcomers" to the area may be used to the new street name (e.g., Last Chance Gulch). [This is very common in the City of San Francisco where they rename streets after famous labor leaders or writers, e.g., Army Street is now Cesar Chavez Street, Ivy Street is now Lech Walesa St., Stark St. is now Jack Kerouac Alley, etc.] In answering Census 2000, respondents will use what they want (North Main or Last Chance Gulch). Maybe the Census Bureau Geography Division has a "street name alias file" that can be used for improved workplace coding???? (If not, they should!) My recommendation may be to add duplicate employer records that have the alias street name, e.g., if the employer record provided by the Census Bureau is Joe's Bar & Grill, 100 North Main Street, then I would add a new record (at the same precise geocoded location) as: Joe's Bar & Grill, 100 Last Chance Gulch. (Now, if there were 500 establishments along this Last Chance Gulch, then I'd have second thoughts about adding alias employer records.) Census Bureau Response: Item #2: Alias street names are a problem. To the best of our knowledge there is not a street name alias feature in TIGER. We are talking here at Census about a possible list of street name aliases that can be used to figure out that respondents may report "Last Chance Gulch" but it may be shown in TIGER as "Main St. " We will put together a suggested format of such a list and send it out to the list serve soon. If you have any questions or comments, please call or email me. (email: clara.a.reschovsky(a)ccmail.census.gov or phone: 301-457-2454) In the meantime, I would encourage you to add a second record for large employers on streets that have alternative names. I understand that you probably cannot do this for a lot of smaller employers. Chuck Wrote: Strategy: I've been working on WORK-UP for the past week, and I've completed initial work on five of my nine counties. In the Bay Area we have 300 thousand geocoded employer locations, and 32,800 un-geocoded employer locations. Of the 32,800 ungeocoded, 1,102 are for employers of 50+ employees. In my large counties, I'm only reviewing the ungeocoded for the 50+ employees (10+ employees in Napa County). The goal in my initial work is to "eliminate the big red dots" (ungeocoded workplaces of 50+). I'm also reviewing *all* of the geocoded records for large (250+ or 500+) employers to make sure they're in the right locations. They mostly are, though there were some problems (UCBerkeley was located in downtown Berkeley instead of on the campus; Marine World USA amusement park was on the wrong side of the road....) And some of the research is actually kind of fun (finding the correct location for Domaine Chandon & Niebaum-Coppola wineries; "Skywalker Sound" in Marin County....) Unfortunately, a lot of the ungeocoded records are shopping centers (they're a real mess.) So, my first phase strategy is to correct the ungeocoded large employers (10+ or 50+ employees); and to review the existing geocoding for very large employers (250+ or 500+). I haven't yet devised a second phase strategy. Census Bureau Response: Strategy: Chuck, your strategy sounds good. It is what we had figured many MPOs would be doing. The shopping centers are important. We often have responses with the store name as the employer and the shopping center name as the only address information. If the shopping center is geocoded to the correct location, then we can successfully geocode these responses. Another special type of record is public schools. Usually people who work for school districts will put down the name of the school district as their employer. Sometimes they put the school name where they actually work in the address field and sometimes they give us partial address information. If you can add records for each school in your area and a geocoded address it will help in coding these respondents to the correct TAZ. Chuck wrote: 3. My question: would it be useful to include the acronym of the employer in the employer name alias field? For example, I doubt that our friends at AASHTO will fill in American Association of State Highway Transportation Officials, but rather just use AASHTO.... Census Bureau Response: Item 3: On the issue of acronyms, YES PLEASE, PLEASE, PLEASE!!!! Adding in acronyms and abbreviations as aliases or separate records is very useful, because some of these acronyms are nationwide in scope while some are highly localized. In what we have seen in testing our coding systems with Dress Rehearsal Data and ACS data, we get a lot responses that are acronyms or local abbreviations. We have a clerical staff that attempts to match these responses with employers in the employer file. There is a learning curve on figuring out what some of these abbreviations really are that often make it very difficult to resolve these cases. If you have any questions about any of the information above, please call or email me. Clara Reschovsky 301 457 2454