Surveys and UP education

This piece is about the University of the Philippines, in view of the coming annual alumni homecoming (June 23, 2012; www.upalumni.ph), which I shall attend as a golden jubilarian. It was in 1962 that I got my bachelor of arts degree, major in economics, from UP.  How time has passed; good grief.

The day after graduation, I started work in UP as a research assistant, then became assistant instructor, and so on, in what became the School of Economics in 1965. I was the San Miguel professor of economics when I transferred to the Development Academy of the Philippines in 1981, after 19 years on the UP faculty. I had written a long letter to then UP president Edgardo Angara, explaining the survey research opportunities promised by DAP.

I am in the middle of three generations of Mangahases on the UP faculty. My mother Ruby was for 30 years with the College of Music, where she eventually became the dean. My daughter Maria is now associate professor of anthropology. Our respective doctorates (economics, University of Chicago; Indian music, University of London; anthropology, Cambridge University) were aided by scholarships granted on the expectation that we would serve UP.

My first acquaintance with survey research was not as an undergraduate, but as a graduate student, on secondment to the International Rice Research Institute for a year, doing my master’s thesis, for UP, on the price-responsiveness of rice farmers. It was not so much what I used in the thesis (i.e. official crop and price surveys), as what I saw being generated on the IRRI experimental farm and in field trials, and presented in weekly seminars. Whereas how to analyze surveys can be taught in the classroom, how to do surveys is mostly learned on the job itself.

Survey-doing. The UP has no unit specializing in survey research, but has many faculty members capable of analyzing surveys and teaching analytical methods, and some with experience in doing surveys themselves. In particular, Enrique T. Virata (statistician; former UP executive vice president) and Jose V. Abueva (political scientist; former UP president; SWS fellow) were Filipino pioneers in election surveys.  Dr. Abueva got his Ph.D. from the University of Michigan, host of the formidable Institute of Survey Research, and the Inter-University Consortium for Political and Social Research, the world’s largest survey archive.

A social survey is a team effort that includes (a) conceiving of how to resolve a research issue through a survey, (b) designing the questionnaire, (c) designing a sampling procedure and implementing it, (d) conducting interviews, (e) encoding and processing responses, and (f) analyzing data and drawing conclusions.  Intellectual authorship belongs to whoever is responsible for steps (a), (b) and (f); it is not lost in case steps (c), (d) and (e) are outsourced.

Steps (c) and (f) entail knowledge of statistics, but not necessarily statistics degree-holders.  Thus, within a university, survey research is not the prerogative of an academic unit for statistics (UP Diliman has a School; UP Los Baños has an Institute). Such research is done by many academic units, from time to time.  The UP Population Institute deserves special mention for having participated in many Philippine demographic surveys.  UP Los Baños has a stronger tradition of survey-based research than UP Diliman.

On matters of survey quality, subject-matter experts are more authoritative than statisticians. Thus the UP’s best teachers and researchers about trends in health, poverty and aquaculture, using statistics, would be in the College of Medicine, the School of Economics, and the College of Fisheries and Ocean Sciences (at UP Visayas).

Basic statistics training. It does not require a formal degree in statistics to be able to teach a basic course in it. I think all social science degree programs include some statistics. Those in the life sciences and natural sciences also need it. I hope all lawyers took it within their undergraduate programs, prior to law school.

To me, there are only two things that general students need remember from basic statistics.  First,  the usability of a sample requires that its elements be drawn at random. Second, the error margin for a sample-based proportion is approximately 1 divided by the square root of the sample size. For example, the error margin for a proportion coming from a sample of 1,000 persons is 1/(sqrt 1000) = 1/31.6, or about 3 percent, plus or minus.

The formula holds regardless of the population of persons from which the sample is drawn.  So I will imagine that anyone who claims that “a survey of only 1,000 persons cannot possibly represent the Filipino population of over 90 million people” is not a UP graduate, or failed in statistics.

Survey archiving. Like any good book that yields new insights to each new reader, the value of a high-quality survey is not depleted by the analysis of the original researcher. Its full dataset, which is a giant matrix of the answers of each respondent to each question, should be carefully archived for public use, to allow new discoveries based on analysis in greater depth, or in conjunction with other data, or for addressing new issues.

There is no central archive where researchers can locate and acquire the raw data from the high-quality surveys done by UP researchers over the years.  I think the creation of a survey archive would be an excellent new offering of a modern, digitized, UP Library.

* * *

Contact SWS: or mahar.mangahas@sws.org.ph.

Read more...