Fifty years ago at the University of Tennessee, Knoxville, Gregg Robinson helped co-author 1975-2000 population projections for Tennessee. That project led to a distinguished career at the U.S. Census Bureau and recognition as one of the country’s leading demographers.

Before his 40-year career at the Census Bureau, Gregg Robinson’s love for demography began in 1972 at UTK. His college roommate signed up a class on demographic techniques, and this piqued Robinson’s interest more than the master’s degree in statistics that he was pursuing.

Robinson made one trip to the campus library to leaf through the course textbook, and the rest is history.

“It was fascinating,” Robinson said. “Demography is a combination of applied statistics, history, geography, sociology; it’s all there together. I said, ‘I’ve gotta learn this.’ ”

Robinson, an Oak Ridge native who graduated in 1970 with an Industrial Management degree from UTK, decided to audit his roommate’s class. That eventually led him to ask Donald “Chip” Hastings—the sociology professor teaching the course on demographic techniques—to be his mentor. Hastings put Robinson in touch with Dick Engels, a research associate at the Center Business and Economic Research (renamed the Boyd Center for Business and Economic Research in 2016) in UTK’s College of Business Administration (now known as Haslam College of Business).

As luck would have it, Engels had the perfect opportunity for Robinson to hone his demography skills.

“Dick Engels just happened to be working on (population) projections,” Robinson said. “I was fortunate that he said, ‘Gregg, you look interested in this. Why don’t you help me with the work on the projections?’ ”

Looking back, Robinson said some of the challenges in demography are very similar to the 1970s when he was producing the “Projections of Population and Labor Force, Tennessee, Regions, and Counties: 1975-2000”. Aside from understating the number of people in the Nashville area, Robinson said he and Engels didn’t do a bad job of projecting Tennessee’s population through the year 2000.

“It comes down to migration. It’s difficult to estimate migration,” Robinson said.

Estimates for net migration, the difference between the number of people moving in and moving out of the state, reached a record 389,000 people from 1970 to 1980, that after falling by 45,000 people between 1960 and 1970. Those levels fueled a 17 percent population increase that decade, the largest increase in the state’s history. 1990 to 2000 saw the second largest increase at 16.5 percent.

“There’s fertility, which is pretty stable short term, but sometimes long term you miss it because projections at the Census Bureau in 50s expected it to be higher than it was just 10-15 years later,” Robinson continued. “Likewise with mortality—you have the data, you make some assumptions and usually it holds, but there can be differences like we had in the past year or two (with COVID).”

BCBER and the Tennessee State Data Center have continued this research, and the most recent 2020-2070 Tennessee population projections are available on the TNSDC website.

Robinson joined Engels for a trip to the TNSDC, then housed in Nashville but now at BCBER, and Robinson said his research there for the population projections convinced him to pursue a graduate degree in demography at the University of Pennsylvania.

Forty-Year Career Focused on Census Quality

Luck struck again his first year in Pennsylvania, when his professor, pioneering demographer Ann Miller, secured him an internship at the U.S. Census Bureau in 1973. He graduated from the University of Pennsylvania with a master’s degree and Ph.D. in demography, and Robinson’s work at the Census Bureau over the following four decades mainly focused on evaluating the accuracy of census and demographic data.

Some of his most important work was related to the development of Demographic Analysis estimates of coverage error for the decennial census. Since 1960, this approach has been used to evaluate the quality of the census using vital statistics, migration data and other administrative data independent of the census itself. This approach helps to highlight disparities in coverage error by age, race and ethnicity.

Robinson’s work also has been forward-looking and included efforts to improve census response rates through his involvement in developing the “Hard-to-Count” score. This tract-level gauge uses low self-response scores from the prior decennial census, along with detailed demographic statistics for small areas, to help identify neighborhoods that could prove difficult to enumerate. Those scores, which are incorporated into the Census Planning Database, were used by communities across the country in the run-up to the 2020 Census.

The keys to a better census response rate, Robinson said, are publicity, advertisement and recognizing diversity and language needs.

“This helps response, but the challenges of this census with COVID and all the other situations—it was really a perfect storm in the wrong way,” Robinson said. “The Bureau did an incredible job of accomplishing what they did. The products were delayed because of the difficulty, but I was really interested in, how did that affect the patterns of response?”

Robinson retired from the Census Bureau in 2015, but his research work continues.

Following the 2020 Census, he teamed up with two other researchers to produce “Hard-to-Count Scores and 2020 Census Response Rates: A Case Study of California”. Robinson and his co-authors found that patterns for hard-to-count areas were still there from previous decades, but some areas performed worse than expected.

“The response rates went down some in the hard-to-count tracts, and they actually rose more in the so-called ‘easy-to-count tracts’, or more affluent areas,” Robinson explained. “Part of this may be because of—we need to study—the new internet option.”

Essentially, Robinson said, internet access may have made it easier for more affluent areas to take part in the 2020 Census. In order to get a more accurate count, Robinson said he believes the Census Bureau needs to “really push the envelope and do something different” ahead of the 2030 Census to get the word out.

Despite the difficulties of the 2020 Census, Robinson said the undercount wasn’t as bad as some predicted.

“At the national level, (2020 Census data) shows our estimates by minority and Hispanic groups. As in the past, the undercount for Hispanics, Blacks and American Indians are higher than the total population,” Robinson said. “The estimates for the non-Hispanic whites and Asians actually showed a net overcount, so the thing is the differential.

“Unless we do something differently, (the undercounts) will be here in the future. There’s always a lot of research and work ahead.”