Printer Friendly

Assuring integrity for federal statistics: focus on GDP.

THE BASIC PURPOSE of official statistics in a democracy is to provide the public and government with information for their general awareness and for their use as a basis for decision or action. To fulfill that purpose, the users must have confidence in statistics. Because few users can replicate official statistics, confidence in official statistics ultimately becomes a matter of confidence in the agency producing them -- confidence in the agency's objectivity and professionalism. The word "integrity" is meant to capture that idea.

In an issue of Business Economics that features the integrity of federal statistics, a focus on gross domestic product (GDP) is appropriate for at least three reasons. First, GDP is the broadest indicator of economic performance, rivaled only by the unemployment rate in the attention it receives from business economists and others. Second, the Bureau of Economic Analysis (BEA) is atypical in that it is mainly an integrator of others' data; with the major exception of surveys of international direct investment, it does not collect data. Therefore, some dimensions of integrity take a different form in BEA than in the more typical statistical agency. Third, and unfortunately, too few business economists are aware of systematic efforts taken by BEA to assure the integrity of the GDP.

This article looks at the integrity of the GDP from two perspectives. The first is the integrity of a published GDP value in the sense of its freedom from political or other influence. Someone, from this perspective, might have asked during the 1992 presidential election: "Was the growth of real GDP, especially in the third quarter, boosted to further the Bush administration's purposes?" (For the record, I reconfirm that it was not, and I think the allegation has been laid to rest.) The second is the integrity of the process of changing the economic accounts of which GDP is a part. Someone, from this perspective, might have asked a year earlier: "Was BEA's switch to GDP from gross national product (GNP) as the featured measure of production made to make the Republican administrations' record of the past decade look better?" (Again, for the record, I confirm it was not.)

Although this article focuses on GDP (following the common usage for GDP as the sum of the expenditure, or product-side, components), the principles and practices it describes apply to BEA's work more generally. BEA prepares personal income and other estimates of the national income and product accounts, capital stock estimates, input-output tables, balance of payments and international investment estimates, regional estimates of income and product, and business cycle indicators under essentially the same conditions. This article does not discuss confidentiality because the situations where it is relevant are not as large a part in BEA's work as in the typical statistical agency. For BEA's international surveys and for certain source data, where confidentiality is relevant, protecting it is an important aspect of the integrity of the statistics.


First, an unequivocal statement: No attempt has been made in at least thirty years to influence a BEA director to push an estimate of real GDP one way or another for political purposes or to deviate from an objective and professional use of the economic information at hand. The thirty years are those covered by the documents of one director and memories of the two others, and span the terms of fifteen Secretaries of Commerce and eight Presidents.(1)

Even the appearance of being subjected to political influence has been an issue only rarely. In twenty-three years, from 1963 to 1985, George Jaszi recounted only two occasions when he had to act to avoid such an appearance. One of those involved GNP, when Jaszi was briefing a Secretary of Commerce before a release. The Secretary, Jaszi recalled, picked up a pen and made a minor editorial change to the release. Jaszi recalled telling him, "You are right, you improved it. But that is none of your business." The Secretary stopped and, according to Jaszi, later wrote an apologetic note.(2) During Allan Young's time and mine, when the "lock-up" safeguards for the final prerelease GDP review gave that meeting a certain glamour, more than one Under Secretary has suggested that they would find it interesting to attend as a "fly on the wall." All such suggestions were firmly turned down.

One could speculate why this track record has been achieved. One possibility is that each administration recognized that integrity of statistics is vital to the functioning of the democratic process. Other possibilities exist, of course; e.g., a substantial political price would be paid if an attempt to influence statistics came to light. However, rather than speculate about the past, it is more useful to look at the present.

Features of the GDP Process

At least three features of the way BEA prepares GDP make it difficult to exert political influence. Although these features help assure integrity, the appearance of integrity, or both, that is not their main purpose.

First, GDP is prepared using published methodologies that open the estimating process to view. The estimation of GDP is bound to be complex in any country, but it is particularly so in the United States. First, the range of the goods and service produced in such a large and diverse economy is wide. Second, the decentralized statistical system yields a vast amount of data, most of which is not necessarily designed to feed into GDP. Although the estimating process is complex, it is not a black-box mystery. BEA's methodologies at several levels of detail help users deal with this complexity according to their needs.

One level of detail appears regularly in the Survey of Current Business, at the time of each annual revision of the national income and product accounts. It shows, in a summary table, the source data and methods used for the current-dollar annual estimates and the extrapolator for the current quarterly estimates.(3) The table shows this information for over forty product-side components and about thirty income-side components. BEA updates the table each time changes are made in methodology.

Since 1985, BEA has published a series of methodology papers that provide even more detail. For example, whereas the summary table just mentioned lists source data and methods for twelve groups of services within personal consumption expenditures, the methodology paper for personal consumption expenditures shows detail for over eighty kinds of services.(4)

Second, GDP is estimated using publicly available source data. BEA integrates a wide range of annual, quarterly, and monthly data from public and private sources to prepare GDP. Amount them are retail sales, manufacturers' shipments and inventories, merchandise trade, and the consumer and producer price indexes, all reported in the news media and tracked by business economists. A few series used as source data, such as some related to motor vehicles, are proprietary, but can be readily obtained by others.

The significance of BEA's working with published source data is underscored each quarter at the time of the advance (first) estimate of GDP. Within hours of the release of the estimate, BEA makes available a table (mentioned on the first page of the news release) that shows the key source data used and the assumptions that BEA makes to fill in months for which source data are not available. For the preliminary and final (second and third, respectively) estimates for each quarter, BEA presents in the "Business Situation" article of the Survey of Current Business a table that shows the revision from the estimate made a month earlier and the source data that caused the revision. This sequence of tables is, in effect, an invitation to verify the use of standard methodologies and publicly available source data. This invitation is backed by a willingness of BEA staff to answer questions -- and I understand that our reputation on this count is a good one.

Third, several stages in preparing GDP involve numbers-in-and-numbers-out comparisons that would reveal tampering. As described in "The Business Economist at Work" in this issue, GDP is built up from detailed components prepared by individual analysts. The source data and components derived from them are reviewed at several levels, by different people at each level, as the total is built up. Then the GDP total, with component detail, is reviewed by a small group of BEA senior staff in a meeting under lock-up conditions the day before the release. Finally, the morning of the release, when BEA analysts are briefed, the release estimates are subjected to scrutiny by the analysts who prepared the component estimates that went into the review meeting.


The three features just mentioned -- published methodologies, publicly available source data, and successive comparisons of the estimates -- are built into the estimating process. In addition, there are safeguards, i.e., explicit precautions or protective stipulations.

Certain characteristics of the structure and authority of a statistical agency fit the definition of safeguard. A recent report by the Committee on National Statistics (CNSTAT) of the National Research Council provided a list of these as "aspects of independence." The list included the professional qualifications of the director and a kind of appointment (referring to career or political, and if political, term of appointment) designed to strengthen independence, the authority to release statistical information without prior clearance, the authority for senior staff to speak on the agency's programs before Congress and public bodies, and the primary authority to select and promote professional staff.(5) Not surprisingly, a very similar list came out of a study, two decades earlier, by a Committee on the Integrity of Federal Statistics sponsored by the American Statistical Association and the Federal Statistics Users' Conference.(6) Although I believe these characteristics of a statistical agency are important, I will not undertake to discuss them. As the CNSTAT report noted, not all characteristics on the list are required for independence. The General Accounting Office's recent report supports the view that BEA's structure and authority have a number of the characteristics that are considered important for the independence that safeguards integrity of the GDP.(7)

BEA provides policy-oriented analysis, but stops short of policy advice. (See the next section for a discussion of why BEA engages in analysis.) BEA's analysis logically falls into the areas covered by the economic accounts; for GDP, these include economic growth, recession and recovery, inflation, saving and investment, and related policy-oriented areas. Stopping short of policy advice precludes any suspicion that BEA's GDP estimates were influenced in order to validate the success of a policy that BEA recommended. More generally, stopping short of policy advice helps maintain the image of a policy-neutral statistical agency.

Steps to avoid premature disclosure comprise another category of safeguards. BEA strengthened these safeguards in the early and mid-1980s, when financial and commodity markets were perhaps too quick to react to rumors about GDP (and other indicators) as though they were the official estimates. The safeguards BEA now maintains are comprehensive.

1. An electronic card access system governs access to the building that houses BEA's computers and its GDP staff, and BEA staff members are required to display ID cards when in the building to help spot unauthorized persons.

2. Passwords and permissions consistent with a "need to know" estimating environment control access to the computers.

3. Final review of GDP and completion of the news release through its "printing" is limited to a few BEA staff, all career civil servants, working under lock-up conditions. These conditions include physical separation of the nearly day-long operation in a lock-up suite, stand-alone computers, and no voice or other contact with the outside by the participants.

4. Unannounced "sweeps" by Commerce Department's Office of Security check the lock-up suite for unauthorized listening devices.

5. Audit trails, telephone logs, and other records emphasize responsibility for security and facilitate after-the-fact review and verification.(8)

Yet, other safeguards are embedded in the provisions of Statistical Policy Directive No. 3 from the Office of Management and Budget (OMB). Some of the safeguards for the GDP were in place independent of this Directive (and its predecessor), but reviewing them in the context of the Directive highlights that they apply to a wider group of important economic statistics. The Directive, as revised in 1985, applies to forty-three series of data or estimates from nine Federal agencies; these series are identified as Principal Economic Indicators. Several BEA series, including GDP, are among them. Four of the Directive's provisions most directly support the integrity of the current estimate of GDP. In the list below, the procedure for GDP follows the statement of the Directive's provision.

1. Release of an indicator is to be prompt, i.e., as soon as practicable after the period to which the indictor refers. Prompt release, besides maximizing the value of an indicator to decisionmakers, reduces the chances of premature disclosure. The GDP release is at 8:30 AM the day after the estimate undergoes final review and the news release is completed in the lock up.

2 The agency preparing an indicator is responsible for assuring against premature disclosure. The agency is directed to provide the indicator to the President through the Chairman of the Council of Economic Advisers when it is ready for release, and the agency is allowed to brief policy officials ahead of official release only to the extent necessary for an orderly review of the indicators and only if there is no risk of prerelease disclosure. BEA delivers the completed GDP release to the Council after 5:00 PM the day before release, and BEA senior staff brief Commerce Department policy officials the next morning, under lock-up conditions until official release at 8:30 AM.

3. Policy officials are prohibited from public comment on an indicator until one hour after official release. This schedule preserves the distinction between the policy-neutral release of an indicator by the statistical agency and its interpretation by policy officials. The Secretary of Commerce often has a press statement or briefing the day of the GDP release, but does so at 9:30 AM or later.

4. Release dates are to be published at the end of the year for the next calendar year. The advance publication of the release schedule, besides assuring equal access by all and facilitating the work of users, assures that release dates are not manipulated for political convenience. The dates, and the times, of the GDP release for the year ahead are widely known. Any change during the year, such as the recent announcement of a change to accommodate BEA's relocation, is announced by BEA with an explanation.

More Than the Sum of its Parts

It is not often that a statistical agency is pleased when the total, or the whole, is more than the sum of its parts. The parts are the features of the process and the many safeguards listed throughout this section. The parts reinforce each other, giving a total that is more than their sum, a result that pleases BEA because it provides strong and systematic assurance of the integrity of the GDP estimates.

It has been only on rare occasions, such as for the third quarter of 1992, when the election's focus on the economy made allegations of political manipulation almost predictable, that even the perceived integrity has been questioned. Still, even rare occurrence is a record to try to improve.

The 1992 episode, painful as it was, raised my awareness of the need to make it more evident to the news media, and through them the public, that the GDP process and the safeguards provide strong assurance of the integrity of the estimate. More specifically, it made me aware of the role that rebuttals to allegations may play in maintaining perceived integrity. Veterans of such skirmishes stress the importance of rebuttals. For example, the "Fundamental Principles" adopted by the Economic Commission for Europe include commenting on erroneous interpretation and misuse of statistics as a practice to which a statistical agency is "entitled."(9) Heads of some other statistical agencies make a point of replying to each misleading statement. I am still considering the pro's and con's of this approach; the con's include the possibility of being viewed as overly defensive and the drain on high-level resources to draft rebuttals. I am leaning toward a more activist approach to responding to allegations.


So far, the discussion has taken a fairly narrow view, one mainly related to freedom from political influence on the published value of an estimate. It has also implied a short-term, static view of GDP and the economic accounts of which GDP is a part. This section takes a less restricted view and mentions several points that are especially relevant during a time of change in the economic accounts. A point of departure is the question: "Will the GDP continue to be prepared within a conceptual framework and using sources and methods that capture, to a generally accepted standard of objectivity and professionalism, the relevant features of the U.S. economy?"

Earlier, I mentioned the switch from GNP to GDP in 1991 as the kind of change that presents a potential problem. GDP had a slightly higher growth rate over the preceding decade than GNP, clear ammunition for charges of political manipulation. A similar potential existed when BEA introduced prices of the current period for the valuation of the direct investment position to replace historical-cost estimates. That change significantly improved the appearance of the U.S. international investment position, although the U.S. was still "the world's largest debtor." To be sure, the status quo is also subject to allegations, including allegations of political motivation. But change, and the process of change, draw attention and thus increase the likelihood of allegations.

Concerns about the integrity during change are particularly relevant now because momentum is building for major change in the economic accounts, change that will upgrade and modernize them over the next decade.

1. The Economic Statistics Initiative, announced in early 1991 as a cross-agency, multiyear effort to provide a sounder statistical base for economic policy, stressed modernizing the economic accounts. The new administration has signaled its intent to carry the effort forward.

2. Major new international guidelines for economic accounts are being unveiled. A revised international System of National Accounts (SNA)--covering subsystems that in the United States are recognized as the national income and product accounts, input-output accounts, flow of funds, and balance sheets -- was approved by the UN Statistical Commission earlier this year. The International Monetary Fund's new edition of the Balance of Payments Manual is expected this fall. These new guidelines lead toward greater comparability of U.S. estimates with those of other countries.

3. The policy questions being addressed are evolving, and policy makers and analysts are asking for more or different statistics. Questions about our "internationalized" economy and about natural resources (and the environment more generally) come quickly to mind.

In the United States and internationally, conferences, studies, panels, and experienced individuals have produced a sizable literature on the principles and practices of statistical agencies. I will draw on that literature to cut short the exposition of general principles and widespread practices so that I can focus on three points related to integrity during change. The first point relates to the role of, and arrangements for, outside advice. I believe the time is ripe for BEA to make some new arrangements for outside advice. The second point relates to research, evaluation, and analysis, and the third relates to documentation. For these, I will state some challenges, some "BEA should...." These changes are substantial, I admit, given the tight resource constraints BEA will face.

Outside Advice

It is widely recommended that statistical agencies obtain outside advice to help identify issues, to suggest improvements, and to reject or validate agency-recommended improvements, among other purposes. Most other countries have advisory committees or boards, especially for their economic accounts, for statistics programs more broadly defined, or for both.(10)

BEA's history shows several models under which advice of different kinds has been provided. In the mid-1950s, a group of nine experts appointed by the National Bureau of Economic Research at the request of the Bureau of the Budget (predecessor of OMB) reviewed the national economic accounts. The emphasis was on their framework or structure. The Bureau of the Budget noted that the various national accounting systems not only serve analytical purposes, but they provide a consistent framework for considering data: "They also reveal what data are lacking and what data are inadequate, inaccurate, or not prompt enough."(11)

The next outside review began in 1973, when OMB commissioned an Advisory Committee consisting of six nongovernmental experts in economic accounting and a four-person working staff to evaluate the underlying data used to estimate the national economic accounts.(12) The Committee made over 150 specific recommendations, designed to be undertaken over a period of years.

Besides these occasional but substantial efforts, the Conference on Research in Income and Wealth (CRIW), sponsored by the National Bureau of Economic Research, has been a long-standing forum for the presentation and evaluation of research related to economic accounting undertaken both in and outside the statistical agencies. Besides conferences that deal with topics related to BEA's work, it has had sessions devoted specifically to critiques of the economic accounts. For example, both academic and government users participated in a "Round Table of GNP Users." More recently, in 1990, the CRIW held a workshop on the revision of the international SNA.

Studies undertaken by the National Research Council's Committee on National Statistics with BEA support have had a narrower focus. For example, a panel was convened to study the adequacy of existing systems for the collection, processing, and dissemination of U.S. merchandise and services trade data and to recommend data to represent U.S. international trade in coming decades.

All these mechanisms are useful; they provided most of their intend benefits. What is needed now? A review of framework is less needed than before; the SNA and the new Balance of Payments Manual provide a future-oriented framework, albeit one that requires adaptation to the U.S. economy and the U.S. statistical system. A review of source data gaps and deficiencies is less needed; the inventory leading up to the Economic Statistics Initiative filled that need for the time being. The need, as I see it, is for a mechanism to obtain outside advice with more continuity and more in-depth focus than in the past. Creating an advisory board to work steadily with BEA toward the long-term improvement of the economic accounts would be a step in the right direction.

Research, Evaluation, and Analysis

Jack Triplett describes the role of research, evaluation, and analysis within a statistical agency by analogy: They serve the same roles as "design and engineering functions" in automobile manufacturing. Without design and engineering, the product falls behind evolving technology and buyer demands.(13)

Evaluation, which I will define as research largely oriented inward, is a continuous process at BEA. For GDP, for example, much of it is embodied in the comprehensive revision of the national income and product accounts. For the last two comprehensive revisions, the Survey of Current Business previewed the conceptual and methodological changes being introduced, explaining their basis to users.

For the period of change ahead, BEA's evaluation efforts will continue to encompass concepts and methodologies. The new international guidelines themselves represent substantial bodies of experience and research. Thus, BEA's research will start with the premise that the concepts in guidelines have much to offer in modernizing the accounts, especially for increasing their international comparability. Methodological research will be needed to implement concepts, including those derived from the guidelines, in the context of the U.S. statistical system.

BEA's evaluation efforts should highlight the relevance of the proposed concepts and definitions to the economic problems known today and envisaged for tomorrow. In some areas on BEA's agenda, e.g., the development of government sector accounts that treat infrastructure as capital in accord with the international guidelines and the development of natural resource accounts on which experts' consensus on some aspects of the treatment has yet to emerge, the role of objective and professional evaluation will be critical.

Thus, evaluation in times of change in the economic accounts has a substantial burden. Evaluation should, objectively and professionally, lay the foundation for change. Further, it should, ideally after discussion within professional organizations and after validation by an advisory board, help provide program justifications in the budget process and provide the basis for documentation of the change to users.

Analysis is research largely oriented to an outside audience. Others, including Janet Norwood, have admirably made the case for the producing agency to analyze its own estimates.(14) Ivan Fellegi, Chief Statistician of Statistics Canada, includes in his arguments for analysis by the producing agency one that bears directly on integrity during change: "An objective and even-handed flow of analytical output can contribute significantly to the image of professionalism and political independence that are so essential for statistical offices. Perhaps more than anything else, this helps to differentiate the public image of the statistical office from that of 'the government.'"(15)

During the resource drought of the 1980s, BEA did less analysis than in earlier years. The capacity to do analysis is still there, as evidenced, for example, by the article "Rates of Return on Direct Investment," which used BEA's new current-value estimates. However, this capacity for analysis should be strengthened. I will aim high.


It is widely recognized that statistical agencies need to be open about the way they work and that documentation is a key to openness. To quote a recent report of the Committee on National Statistics:

"One important criterion to instill credibility and for an agency to operate in a completely open manner. Openness requires providing a full description of its data with acknowledgment of any uncertainly in the data and a description of the methods used and assumptions made. Agencies should provide to users reliable indicators of the kinds and amounts of error to which the statistics are subject.... In addition...., a description of the concepts used and how they relate to the major desirable. Openness also means that the agency describes how decisions on methodology and procedure are made."(16)

If documentation is a key to openness when the status quo is being maintained, it is even more important during change. Further, the scope of documentation broadens during change. During change its elements should include, besides those mentioned in the quotation, presentation of an overall plan for change, merging with the research work just describe, public presentations of the reasons for change.

BEA's efforts to announce our plans widely and well in advance and to explain the reasons in 1991 when GDP became the featured measure followed a model that should be further developed for the period of change ahead. An article appeared in the Survey of Current Business in August, well before the end-of-year change, and BEA staff made presentations explaining the change in a number of settings. The explanation in this case referred to the international guidelines and thus international comparability, the appropriateness of GDP's coverage for certain types of analyses, and the consistency with measurement considerations.

Documentation is not a particularly user-friendly word, but I chose it because other words, such as outreach, may sound too trivial. The groups that are stakeholders in the economic accounts are diverse--public and private users at different levels of sophistication, the media, and source data suppliers. Accordingly, the documentation should be adapted to the needs and interests of these groups.

As BEA upgrades the economic accounts over the next years, BEA should undertake documentation in a way that maintains the integrity. The effort, including application of new presentation skills, will be substantial. Again, I will aim high.


1 George Jaszi, Director of BEA from 1963 to 1985, died in December 1992. His experience is captured in his statement, "Contrary to occasional suspicion, there never has been an attempt to tamper with BEA's estimates." "An Economic Accountant's Audit," American Economic Review, vol. 76, no. 2 (May 1986), page 416.

2 Recounted by Jaszi to Keith B. Richburg, "Retiring Economist Painted Big Picture," Washington Post, January 21, 1985, page A13.

3 See, for example, table 7, "Principle Source Data and Estimating Methods in Preparing Current-Dollar Estimates of GDP" in the July 1992 Survey of Current Business.

4 A list of these papers is in A Users' Guide to BEA Information. The Users' Guide is available in the January 1993 Survey of Current Business, as a separate document available on request from BEA, and on the Commerce Department's Economic Bulletin Board.

5 Committee on National Statistics, Principles and Practices for a Federal Statistical Agency, Margaret E. Martin and Miron L. Straf, eds. (Washington, DC.; National Academy Press, 1992), page 4.

6 American Statistical Association-Federal Statistics Users' Conference Committee on the Integrity of Federal Statistics, "Maintaining the Professional Integrity of Federal Statistics," The American Statistician, vol. 27 (April 1973), pages 60-61. It may be noted that the 1973 report recommended that the heads of statistical agencies be in the career service. The CNSTAT report, in contrast, favors appointment by the President with the approval by the Senate, for a specific term not coincident with that of the administration.

7 General Accounting Office, "Gross Domestic Product: No Evidence of Manipulation in First Quarter 1991 Estimates," March 1993, especially page 22. The report noted that BEA had several of the characteristics, but not others; that this situation does not mean that BEA is insufficiently independent or less independent than agencies that exhibit independence differently; and that the appropriate form and degrees of independence is a complex issue that should be assessed in a larger context.

8 The priority given to security has its costs, most notably perhaps in staff training.

9 "Fundamental Principles of Official Statistics in the Region of the Economic Commission of Europe," adopted during the 47th session of the United Nations Commission for Europe, Geneva, April 14, 1992.

10 Handbook of Official Statistics in ECE Countries (Geneva: Conference of European Statisticians, 1988). My own review a few years ago of the statistical systems in France, Germany, Australia, and Canada drew my attention to this fact.

11 Statement of Raymond T. Bowman, Assistant Director for Statistical Standards, Bureau of the Budget, in U.S. Congress, Joint Economic Committee, The National Economic Accounts of the United States, Hearings, 85th Congress, 1st session, 1957, page 4.

12 Gross National Product Data Improvement Project Report, prepared by an Advisory Committee under the auspices of the Office of Management and Budget, issued by the Office of Federal Statistical Policy and Standards, U.S. Department of Commerce (Washington, DC.: Government Printing Office, October 1977).

13 Jack E. Triplett, "The Federal Statistical System's Response to Emerging Data Needs," Journal of Economic and Social Measurement, 17 (1991), pages 155-177. The same issue carried comments on Triplett's article by F. Thomas Juster, Burton Singer, and Fritz Scheuren.

14 Janet L. Norwood, "Should Those Who Produce Statistics Analyze Them? How Far Should Analysis Go? An American View," Bulletin of the International Statistical Institute, vol. 46 (1975), pages 420-432.

15 I. P. Fellegi, "Maintaining Public Confidence in Official Statistics," Journal of the Royal Statistical Society, vol. 154, part 1 (1991), page 4.

16 Principles and Practices for a Federal Statistical Agency, pages 17-18.

Carol S. Carson is Director, Bureau of Economic Analysis, U.S. Department of Commerce, Washington, DC.
COPYRIGHT 1993 The National Association for Business Economists
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:gross domestic product
Author:Carson, Carol S.
Publication:Business Economics
Date:Jul 1, 1993
Previous Article:The integrity of federal statistics: a case study from the GAO perspective.
Next Article:Perception or reality: can we trust federal statistics?

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters