eKonometrics

Thursday, July 29, 2010

Poor grammar, i.e., the grammar of graphics

The front page lead story in the Globe and Mail on July 29 highlights one of the biggest challenges faced by the human beings today, i.e., global warming. The Globe offers graphic evidence as time series charts, which depict rising temperatures on the planet over the past 100-plus years.

While the Globe should be commended for presenting graphic facts to educate its readers rather than publishing photographs of drowning polar bears, one must still point out the basic lacunae in the graphs. They are missing legends and labels for abscissa and ordinate, which diminishes their ability to inform the reader.

Without legend one does not know what different coloured lines stand for. Without axes labels one does not know if the temperatures are depicted in Fahrenheit or Celsius, or when exactly did the peaks in ocean temperature, which are visible in the graph, occur.

Since the ordinate labels are missing, one may erroneously assume that the graphs laid out in the same column, (e.g., sea level, ocean, sea-air and land-air temperatures) experienced a proportionate change over time, which may not be true.

A graph can be worth a thousand words, provided one follows the ‘grammar of graphics’, which is perhaps as difficult to follow as the struggle with split infinitives and dangling modifiers.

A look at the original publication by the National Oceanic and Atmospheric Administration reveals that the different coloured lines represent monitoring by different agencies or authors. The other important fact to know is that most graphs represent the anomalous deviation from some expected value and hence the horizontal black line in the above graph represents the absolute zero deviation. However, this is not true for all graphs and hence the axis labels become even more important.

The correctly illustrated graphs are available at:

http://www.ncdc.noaa.gov/bams-state-of-the-climate/2009-time-series

An example of the correct graph is reproduced below, which shows that land surface temperatures increased sharply as of late seventies, which could not be deciphered from the graph published by the Globe.

A sneak peak at the new census questionnaire

While the controversy over the 2011 census in Canada continues, Statistics Canada, the agency responsible for holding Census in Canada, has released the draft questionnaire for review.

There are a few changes in the questionnaire that may make transport planners happy.

Here is a brief list of the changes that I could spot:

Question 33 about unpaid work or “care economy” has been removed. It appears that the National Statistics Council had already decided upon this question, but made it look like a new suggestion in their press release two days ago. This needs more probing. The exact timing of NSC’s recommendation to exclude question about unpaid work should be established.
Questions about childcare expenses have been included.
Also included are questions about child support paid to former spouses.

This development is interesting. The question about unpaid work, which predominantly dealt with women’s non-monetised contribution to economy and society, has been removed and a question about child support, that affects men predominantly, has been included.

Transport planners should be delighted to note that in addition to the questions about the mode of travel to work, and work location, the questionnaire in 2011 will collect info on the following:

Average travel time for home to work trip
Departure time for home to work trip
Car pooling, i.e., number of people travelling together to work in the same vehicle
The mode of travel question has also been enhanced by asking details about public transit, i.e., differentiating between buses, subways, street cars, and regional rail.

The question is phrased poorly as it will most likely report lower mode split for buses for multi-modal trips.

And yes, the survey will be voluntary, not mandatory!

Lastly, I think it is not too early to start thinking about putting up a legal challenge to the government’s decision to make the survey voluntary. It is indeed not going to be easy because the courts would not like to interfere/override the authority of the minister (executive) as enshrined in the constitution. However, there are ways to make this happen and universities with faculties of law and businesses with legal departments may want to start looking for those whose welfare will be adversely impacted by this decision and help such aggrieved parties litigate.

STATISTICS ACT of Canada

According to the act it is not merely the refusal to respond to Census, but wilfully falsifying a response to Census or any other info sought under the act may lead to a prison term.

The Statistics Act states:

False or unlawful information

31. Every person who, without lawful excuse,

(a) refuses or neglects to answer, or wilfully answers falsely, any question requisite for obtaining any information sought in respect of the objects of this Act or pertinent thereto that has been asked of him by any person employed or deemed to be employed under this Act, or

(b) refuses or neglects to furnish any information or to fill in to the best of his knowledge and belief any schedule or form that the person has been required to fill in, and to return the same when and as required of him pursuant to this Act, or knowingly gives false or misleading information or practises any other deception thereunder is, for every refusal or neglect, or false answer or deception, guilty of an offence and liable on summary conviction to a fine not exceeding five hundred dollars or to imprisonment for a term not exceeding three months or to both. 1970-71-72, c. 15, s. 29.

For the complete text of Statistics Act of Canada, please visit:

http://www.statcan.gc.ca/about-apercu/act-loi-eng.htm

From Coats to Sheikh

Last week Dr. Munir Sheikh resigned as Statistics Canada’s chief statistician to register his protest at the government’s attempt to misrepresent the agency about its views on converting the mandatory census into a voluntary one.

Dr. Sheikh told the parliamentary industry committee that the “fact that in the media and in the public that there was this perception that Statistics Canada was supporting a decision that no statistician would, it really casts doubt on the integrity of that agency, and I as head of that agency cannot survive in that job."

It is only appropriate to remember Robert H. Coats who was the first chief statistician of Canada who laid the foundation of Statistics Canada after being appointed as the Controller of Census in 1915.

Wikipedia recalls Robert Coats as follows:

Robert Hamilton Coats (July 25, 1874 – February 7, 1960) was Canada's first Dominion Statistician.

He was born in Clinton, Huron County, Ontario in 1874, the son of Charles Coats, who came to Canada from Scotland. In 1896, Coats received a B.A. from the University College in Toronto. He worked as a journalist for the Toronto World and then the Toronto Globe until 1902 when, at the request of Prime Minister Mackenzie King, he became editor of the Labour Gazette; King himself had been the first editor of this publication which included statistical information related to labour. Coats was named Chief Statistician for the Department of Labour in 1905. In 1915, he was appointed Dominion Statistician and Controller of the Census. Coats helped establish the Dominion Bureau of Statistics, now Statistics Canada.

He also served on statistical committees with the League of Nations. After he retired in 1942, Coats served as statistical advisor to the government of Ontario and the United Nations. He served four years as visiting professor of statistics in the Department of Political Economy at the University of Toronto.

Coats, one of the first residents of Rockliffe Park, died in Ottawa at the age of 85. He had been married twice, first to Marie Hollbeister, and then, after his first wife's death, to Maida Skelly.

The R.H. Coats Building in Ottawa was named after him.

References

Tuesday, July 27, 2010

“Munir Sheikh shows us what integrity and leadership looks like”

"Let me first of all say that it is the right of the government to make decisions, which if lawful should be implemented by any department of the government…

“The fact that in the media and in the public that there was this perception that Statistics Canada was supporting a decision that no statistician would, it really casts doubt on the integrity of that agency, and I as head of that agency cannot survive in that job."

Dr. Munir Sheikh

This story cannot be told any better than John Ibbitson of Globe & Mail in today’s online edition:

“Munir Sheikh’s testimony before the Commons industry committee reminded us of something that too many forget: He did not resign as deputy minister responsible for Statistics Canada for the wrong reason; he resigned for the right reason…

“It is one thing to quit your job because you don’t agree with the boss. It is something quite different to quit your job rather than see the integrity of the people you lead compromised. There aren’t many of us who would do such a thing. But Munir Sheikh would, and did.

Self-selection bias and census

What happens when one tries to analyse women’s labour force participation using data collected from a voluntary survey. There will obviously be a self-selection bias, that is, we will observe labour force participation choice made by only those women who chose to fill out the census form. As for those women who opted not to fill the census form, we would have no knowledge about their decision to participate in labour force or otherwise.

Now armed with a data set that suffers from self selection bias, we would have to face the challenge of analysing wages for only those women who chose to fill out the form and then chose to participate in the labour force. There are now two self-selection biases in play: one dealing with filling out the census, and the other dealing with participating in the labour force.

This poses interesting challenges in econometrics, as pointed out by Mathew Kahn in the Christian Science Monitor. He writes:

“The only winner here is Jim Heckman [recipient of the the 2000 Nobel Memorial Prize in Economic Sciences (with Daniel McFadden)] as his cite count will go up as smart researchers in sociology, demography and economics will have to come up with a selection correction to model who bothers to fill out the form and then a researcher studying women's labor force participation would need to estimate a second selection correction. I'd like to see somebody work out the "three step" standard error formula in this case!”

What Mathew means is that a researcher has to account for not just one, but two self-selection biases to be able to estimate unbiased coefficients in an econometric model that explains wage differentials between men and women. Obviously, this fine point will be lost on those who never made it beyond Statistics 101. For them, Mathew Kahn offers a simpler example:

“Suppose that highly motivated busy people don't bother [filling out the Census], then the "average" person in Canada will look "lazy" because the sample who fills out the survey will omit the high achievers.”

Put simply, the voluntary census long form will make Canada lazier than it really is!

Lastly, a paper published on census response rates suggested that racially and socio-economically diverse communities reported lower response rates at the county level. This implies that racial minorities are more involved in census when there numbers increase within the community.

Furthermore, the US Census Bureau in 2003 compared the response rates for the mandatory American Community Survey by sending a portion of the respondents the same survey but indicating that their response was voluntary. The change from mandatory to voluntary survey resulted in a huge 20.7% decline in response rate. At the same time the Bureau observed that the highest decline in response rate occurred in communities with high response rates recorded for the Census conducted in 2000.

These issues have serious implications for converting the mandatory long form into a voluntary survey.

Community Composition and Collective Action: Analyzing Initial Mail Response to the 2000 Census

Jacob L. Vigdor (Duke University)

The Review of Economics and Statistics, February 2004, 86(1): 303–312

Abstract
This paper analyzes how community heterogeneity influences resident decisions to undertake actions generating public benefits. The decision in question is completing and returning the 2000 Census questionnaire, an action which secures a significant amount of federal grants for the community. The model developed to explain this action allows members of societal groups to differentially value public benefits that accrue to other group members. Racial, generational, and socioeconomic class heterogeneity all predict significantly lower response rates at the county level. The potential for endogenous sorting into heterogeneous counties implies that the magnitude of true behavioral effects exceeds these estimates. Copyright (c) 2004 President and Fellows of Harvard College and the Massachusetts Institute of Technology.

Monday, July 26, 2010

Do immigrants perform more unpaid work?

Image by Ian Muttoo via Flickr

The National Statistics Council (NSC) of Canada has suggested to axe Question 33 from the Census long form in an effort to resolve the standoff between Statistics Canada and the Conservative government of Stephen Harper.

This is the same question that Statistics Canada was unwilling to include in the long form and was overruled by the Parliament in 1995. More details on this are available in my earlier blog.

The decision to keep or eliminate Question 33 from the Census should be based on proper debate and deliberation including all stakeholders. The NSC has made an ad hoc decision to eliminate the question about unpaid work from Census that has undermined the due process of managing Census in Canada.

The inclusion of unpaid work in Census allows us to analyze the welfare of the most disenfranchised in the society. See the following tabulation that I performed on 2001 Census for Toronto CMA. It reveals that immigrants are far more likely to engage in unpaid work than non-immigrants. Furthermore, 44% immigrant women worked more than 15 hours per week in unpaid assignments. A much lower proportion (36%) of non-immigrant women worked unpaid for more than 15 hours per week. At the same time, a smaller proportion of men (immigrant or otherwise) worked unpaid for more than 15 hours per week.

These insights would not have been possible without Question 33 about unpaid work in the Census. Are there other sources for this type of data? May be there are. But are such data readily available to those who advise policymakers? Definitely not.

Immigrants unpaid work	female	male	Total
none	9.5	14.9	12.1
less than 5 hours	17.5	30.0	23.4
5 to 14 hours	29.2	34.6	31.8
15 to 29 hours	23.4	14.1	19.0
30 to 59 hours	14.1	5.0	9.8
60 hours or more	6.4	1.5	4.0
Total	100.0	100.0	100.0
*15+ hours*	*43.9*	*20.6*	*32.8*
Non-Immigrants unpaid work	female	male	Total
none	8.5	14.5	11.4
less than 5 hours	22.7	35.0	28.7
5 to 14 hours	32.6	33.5	33.1
15 to 29 hours	21.1	12.6	16.9
30 to 59 hours	10.5	3.6	7.2
60 hours or more	4.5	0.9	2.8
Total	100.0	100.0	100.0
*15+ hours*	*36.1*	*17.1*	*26.8*

Pages