Trump vs Statistics


  • Author: Carlos Grajales
  • Date: 16 February 2017
  • Copyright: Image appears courtesy of

Government’s relationship with official statistics has always been fundamental for both state and science, since statistics as a discipline began largely due to the necessity of governors for information. The first registered census possibly took place around the year 3800 BC in Babylonia, essentially to estimate how much food they needed to feed their population [1]. The oldest census for which we still have complete data was undertaken in the year AD 2, by the Han Dynasty in China, which recorded a population of 59.6 million. As you can see, it’s been a while since states around the world realized how they could use information to better organize, understand and even control their population. That is why it is still a shock for every statistician to hear of governments who view statistics with disdain, particularly when the numbers disagree with the official discourse.

thumbnail image: Trump vs Statistics

The most publicized recent case came in January 2007, when Graciela Bevacqua, head of Argentina's official consumer price index (CPI) and one of the main statisticians within Argentina’s National Institute of Statistics and Census, was fired [2]. The reason was widely reported to be a disagreement with the Argentina Economic Ministry, who happened to disregard the inflation numbers that had been reported during the past couple of months. Government officials had already suggested “methodological upgrades” to modify the index calculation, something Bevacqua refused to do, as the changes lack the rigour intended for international official statistics. By the time she left office, she had been ordered to revise January inflation numbers, which happened to be “too high” for the government. Her departure was described as “a death blow to the independence and credibility of the institute” [2]. By late 2015, she returned to her former position at the INDEC, after Argentinean voters brought a new party to power. Sadly, that reconciliation lasted only a few months, after she was tasked with building a new Consumer Price Index within a time frame of two months [3]. Due to government interference, the CPI methodology and estimation had been largely compromised, up to the point that a whole revamp of it would be necessary - an impossible task to meet in mere 60 days. She had to leave the INDEC once again, when she refused to meet these time requirements, as they would compromise the index accuracy [3].

Graciela’s situation is remarkable for two factors: first, the direct intervention of the federal government as a means to alter the estimation of official statistics; secondly, because Argentina is widely regarded as a democratic country. This episode brought to the attention of the statistical community how in the 21st century government officials can still undermine, consciously or not, the effort and labour of statisticians. Now, in 2017, the community is worried of such government attacks taking place yet another time: once again in the same continent, but further north.

As of January 20th, Donald John Trump became the 45th president in the history of the United States. Months before he took office there were concerns regarding his unorthodox views about government, concerns that, after the first few weeks of his mandate, have translated into talk of an incoming catastrophe associated with his executive orders, his personal impulses or with the work of some of his staff members [4]. Still, a subject that has received far less attention is the damage he could deliver to statisticians and statistics as a whole.

By the time Trump reached the White House, many veteran government officials resigned. It was widely reported that many State Department veterans left the office just days after he initiated his term [5]. Not so widely reported was the fact that some high-level employees at federal statistical agencies also resigned [6]. And apparently, one of the reasons for departure was the fear of not being able to comply with their jobs properly anymore, due to the effect that a Trump administration could have on official statistics. In an interview with the Guardian, Katherine Wallman, chief statistician of the United States from 1992 and up to this past month, declared “that picking and choosing your numbers to suit your politics is not the way that we ought to be doing it.” [6]

Not that Trump himself will alter official numbers. There are other ways in which the government may interfere with the labour of statisticians. In fact, the most common concern among the statisticians interviewed was the threat of defunding some data collection programs. This is already happening. Two bills currently in congress are attempting to nullify a 2015 housing regulation aimed at addressing racial segregation in urban cities, by altogether cutting the funding for data collection [6]. The new bills, state that “no Federal funds may be used to design, build, maintain, utilize, or provide access to a Federal database of geospatial information on community racial disparities”. Without access to federal resources, the data collection program would surely disappear, thus leaving the country without data to study and correct the racial disparity in the housing market.

Some statisticians are more positive, particularly regarding economic indicators. Economic data is so ubiquitous and useful, that there’s simply no chance any president could interfere with them, at least not in the scale the Argentinean government did. But it is hard to know how a Trump presidency might react to the news of a declining economy or to the news of a poor job market report (particularly when creating jobs was such an important campaign promise for the Republican). Under these circumstances, the threat might not be the direct interference with the Statistical Agencies, but an undermining of their professionalism and accuracy. This is not something unheard of, particularly for Trump. During his primary campaign, Trump repeatedly accused the U.S. Bureau of Labor Statistics of reporting inaccurate numbers for the unemployment rate. "Don't believe those phony numbers when you hear 4.9 and 5 percent unemployment," Trump said. "The number's probably 28, 29, as high as 35. In fact, I even heard recently 42 percent" [7]. Though these claims may seem superfluous, a similar, or more consistent complain can have a deeper impact into the credibility of the Bureau. This can even affect their estimations directly. As it is the standard around the globe, unemployment in the United States is estimated with a survey, the Current Population Survey (CPS) [8]. And as anyone lucky enough to have worked with surveys will know, there’s one source of error that Trump attacks can actually induce: Interviewer non-response. Picture a scenario in which the president is constantly claiming on air (and surely on Twitter as well), that the BLS is producing “phony” numbers. A fair amount of Trump supporters may believe his claim and stop believing in the BLS numbers and estimates immediately. Even worse, if some of the households of such Trump supporters happen to be among the 60,000 sampled for the Current Population Survey, they will likely be uninterested in participating in such a “phony” study. If Trump’s complaints reach a relatively high number of individuals, the CPS could end up with increasingly higher non-response rates, a simple fact that some opponents of the numbers can use to discredit them, even if they remain fairly accurate [9].

Even if the risk of losing funding and being the target of one of the president’s rants wasn’t enough, interviewed statisticians also mentioned another cause of concern: onerous vetting processes to release information [6]. As Katherine Wallman also expressed, this is in no way something new: “I am aware of occasions where policy folks have thought it was appropriate to change the things that were featured in a press release, or take out specific bullets that they thought were unattractive, or change the timing of the release because it might be inconvenient in terms of a policy that a cabinet official wished to announce” [6].

In the Trump administration, this is starting to happen as you read this. A few days ago, the Trump administration mandated that any studies or data from researchers at the Environmental Protection Agency would undergo a deep political review by special appointees before any information is released to the public [10]. The EPA is the U.S. agency that performs most climate change related studies and its relationship with man-made carbon emissions. The review ordered by the Trump administration is so broad that even the contents of the agency’s website and its Facebook pages have to be approved by the White House [10]. Never had the agency been subject of such scrutiny in the past and it is the fear of many data professionals that such restrictions and revisions might repeat in other statistical departments.

Though it is hard to imagine how the statisticians working in agencies targeted by Trump administration could overcome these potential issues, I encourage all my fellow colleagues to raise their voices and concerns over any inappropriate government interventions, either in the U.S. or elsewhere. Working on official statistics is a noble endeavour, one that ensures both the people and the governments base their decisions on the same basic knowledge. As H. James Harrington, a fervent proponent of statistical quality control management once said: “If you can’t measure something, you can’t understand it. If you can’t understand it, you can’t control it. If you can’t control it, you can’t improve it.” It is our duty, as members of the statistical community, to ensure that our countries have the appropriate means to measure and improve the lives of its citizens.


[1] Carlos Gómez Grajales, Eileen Magnello, Robert Woods, Julian Champkin. Great moments in Statistics. Significance Magazine Volume 10, Issue 6 December 2013 Pages 21–28 (December, 2013)
[2] Bermúdez Ismael. Desplazan a la encargada de medir la inflación en el INDEC. Diario El Clarin (January, 2007)
[3] Tarran, Brian Price index row in Argentina costs statistician her job, yet again. Significance Magazine. Volume 13, Issue 2, pages 4–5 (April 2016)
[4] Heilbrunn, Jacob. The Most Dangerous Man in Trump World? Politico Magazine (February, 2017)
[5] Graham, David. Trump's Hollowed-Out State Department. The Atlantic (January, 2017)
[6] Chalabi, Mona. Statisticians fear Trump White House will manipulate figures to fit narrative. The Guardian U.S. (January, 2017)
[7] Jabson, Louis. Donald Trump repeats Pants on Fire claim that unemployment rate could be 42 percent. Politifact Website (February, 2016)
[8] How the Government Measures Unemployment. Bureau of Labor Statistics Website (October, 2015)
[9] Saunders, Catherine. Taking an Interest: What Makes Someone Respond to a Survey. StatisticsViews Website (January, 2016)
[10] Trump administration: EPA studies, data must undergo political review before release. The Guardian (January, 2017)

Related Topics

Related Publications

Related Content

Site Footer


This website is provided by John Wiley & Sons Limited, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ (Company No: 00641132, VAT No: 376766987)

Published features on are checked for statistical accuracy by a panel from the European Network for Business and Industrial Statistics (ENBIS)   to whom Wiley and express their gratitude. This panel are: Ron Kenett, David Steinberg, Shirley Coleman, Irena Ograjenšek, Fabrizio Ruggeri, Rainer Göb, Philippe Castagliola, Xavier Tort-Martorell, Bart De Ketelaere, Antonio Pievatolo, Martina Vandebroek, Lance Mitchell, Gilbert Saporta, Helmut Waldl and Stelios Psarakis.