Congress Is Threatening Your Market Data
This is part two of the Q&A. Part one looked at how the Census impacts your media planning. This section will get a little geekier, but when talking to the head of the Census, how else would you expect an AdAgeStat Q&A to roll?
AdAgeStat had the opportunity to chat via phone with the director of the U.S. Census Bureau, Robert Groves, as he was riding between buildings near the Capitol. While he says "I'm not in [advertising], but I have in the past speculated about that business" he offered plenty of insight for marketers, addressed concerns tweeted by @adagestat followers, and explained why we need to lobby Congress now for better data.
AdAgeStat: On your blog, you sometimes critique media pieces about the Census, such as a recent piece labeling some counties as "dying." What are some big pitfalls of interpreting this data?
Mr. Groves: It's always good to talk to people who study a phenomenon over a long period of time. If you talk to demographers, this labeling of a city or county that lost population as a "dying" county really is kind of historically myopic. But we have many of our cities losing population over this 10-year period. Labeling them as "dying" just doesn't hold up historically. Cities go through cycles. Businesses change, industrial base changes and so on. You're assuming a trend that you've seen will continue in the future and that 's just a fallacy that applies to all sorts of things with data. That's the principle behind the flaw of "dying cities."
AdAgeStat: One Census question that has been problematic is the interpretation of the differences between race and ethnicity. In Census 2010 you're studying different changes to that question, correct?
Mr. Groves: We built in, as we do every decade, some experiments. These are randomized experiments inside the Census to try out different ways of doing the Census. One of them was different ways of asking the race and ethnicity questions. I haven't seen those results yet but some of them are motivated by concerns among some Hispanics that asking them to clarify their race after asking their ethnicity poses a real problem and they don't what to choose on race and hence choose "some other race." We're trying to explore whether we could handle that better.
There were other issues that arose with regards to race and ethnicity measurement, for example, the use of the word "negro" is something objected to, and the concerns of the Arab-American community that they really don't know how to handle the race question and they don't know what to do with it.
We will mount this decade a fairly formal open and broad effort to reevaluate the questions that attempt to get input from a lot of different groups.
AdAgeStat: Will that make decade-to-decade comparisons troublesome?
Mr. Groves: That's the counter-argument -- if you keep changing the categories, how can I compare across time. Any attempt to improve a measure for one reason or another will break the time series and that 's a trade-off that has to be expressly on the table.
Just a footnote to get this right, this is a process that would actually be governed in some sense by the office of Office of Budget and Management's Statistical and Science Policy group. They are actually charged with defining the official categories of race and ethnicity in the U.S. under law.
AdAgeStat: Depending on which Census survey you look at [the ACS, the Current Population Survey, etc.] there are a lot of different numbers for occupied housing units. That's a really important number in our industry. Can't you just pick one already?
Mr. Groves: Every sample survey would generate a different estimate. There's a philosophical issue that you're raising: Should one be blessed as the official number? We do have an official unemployment rate, right? There are actually a lot of different unemployment rates, but we have decided through processes to bless one as the official one. So that would be one way to handle that . It would be an interesting conversation, but I have never been engaged in a conversation on that before.
AdAgeStat: The ACS five-year estimates have some pretty sizable margins of error in the smaller geographies.
Mr. Groves: Yup. You're right.
AdAgeStat: Things like, "0 plus or minus 250." How concerned should we be about that ?
Mr. Groves: The old long-form [Census] was a 1 in 6 sample roughly. I guess the first answer is : don't look at the estimates without looking at the margin of error. And just a historical footnote that you'd like: The long-form estimates had standard errors attached to them, too but we didn't really publish those the way we do with the ACS. In a real way, giving you the margins of error on ACS is a more honest delivery of the information because it cautions you about appropriate uses. The long-form standard errors were buried in some technical documentation that some people saw but it wasn't in your face as a user.
The second thing to note is that this is purely a cost thing. If we had unlimited budgets, we'd have smaller standard errors because we'd have bigger sample sizes. We're trying to maximize the use of every dollar we get to have big sample sizes and the ACS is an expensive survey but for small area uses it's unstable for certain kinds of estimates. They only antidote to that is bigger samples.
AdAgeStat: So we should all go lobby for more money?
Mr. Groves: As you may know, there's some concerns about the ACS. The Republican National Committee, this is a matter of fact, passed a resolution in the summer to abolish the ACS. I think understanding the trade-off between the burden we give a person to filling out the form and the wealth of statistical information it gives small businesses and mayors of small towns -- that kind of tradeoff we need to make sure people understand in making decisions about the future of the ACS.
AdAgeStat: Where else could small-business owners turn to get this kind of insights?
Mr. Groves: I don't know of another data source on all of the attributes we have that could substitute for this. We just don't produce it for a country. Local jobs are connected to intelligent use of ACS and it can be used to really help American business in a variety of ways.
AdAgeStat: I asked the @adagestat Twitter followers if they had questions for you and one thing that came up was a plea for the Census to standardize some of the age ranges it uses both internally and to align with common industry ranges, like TV demographics.
Mr. Groves: I see, I see. Actually this is a wonderful piece of feedback for us. This is the kind of feedback that we need to listen to. We need to be really sensitive to major users groups to what kind of cuts fit best.