Flaws in hunger survey criticism

I MUST express my appreciation to Inquirer columnist Rigoberto “Bobi” Tiglao, whose Thursday piece, “Flaws of hunger polls,” together with his stature, provided me with an easy topic for today.  The remarks of a veteran journalist, former ambassador and friend (hence I use his nickname) deserve more than a mere dismissal.

SWS sampling has proven its worth.  Bobi’s criticism of SWS sampling, occupying almost half of his column, is the easiest to set aside.

The SWS sampling system is not designed for the topic of hunger in particular, but for all the many topics to be surveyed.  It is the same system used when the questions include voter preferences, which surely vary across provinces much more than hunger does.

Multi-stage random sampling, i.e., selecting provinces first, then towns, then barangays, and so on, is always by probability proportional to size (pps).  Thus populous provinces like Cebu and Pangasinan are in fact the ones most frequently (but not always) sampled.  This system is not peculiar to SWS, but is done by all reputable survey institutions, since single-stage sampling is theoretically nice, but extravagant.

The outstanding proof that multi-stage sampling works is the success of the election surveys.  Pre-stratification by social class is unnecessary.  The respondents who tell us about their votes are the same ones who tell us about their hunger.  I believe they are equally truthful about the two matters.

Don’t belittle Filipino survey respondents.  Bobi’s guess is wrong that respondents do not notice the qualification “and not have anything to eat,” that marks the hunger as involuntary.

Here are examples of surveys using such a qualification. (a) In 2003, the Food and Nutrition Research Institute asked mothers/caregivers in a survey of 6,683 households: “In the last 6 months, were you ever hungry but did not eat because there was no food or no money to buy food?  How frequently did this happen?” Those who said it happened at least once were 24.4 percent—a number that should be halved since the SWS item refers to the past three months only; 2003 was the year of lowest hunger (7 percent annual average).

(b) In April-June 2006, the Bureau of Agricultural Statistics replicated the exact SWS questions on a very large national sample of 12,857 households.  It obtained Moderate Hunger of 15 percent, Severe Hunger of 3.6 percent, and Total Hunger of 18.6 percent, strongly validating the SWS hunger surveys in that year.

Bobi’s cultural card is worthless. In its frequent meetings with foreign counterparts (International Social Survey Program, Gallup World Poll, Asianbarometer, etc.), SWS has never heard allegations of inferiority or superiority, for cultural reasons, of any nationality’s survey responses.

Appreciate new knowledge.  Bobi is wrong in thinking that hunger cannot possibly fluctuate from quarter to quarter.  There are no alternative quarterly hunger surveys that say differently.

Bobi is wrong in thinking it’s impossible for any white collar workers, property owners, or professionals to ever go hungry.  The SWS surveys found some (not many) hungry ones.

An objective person appreciates surprising information learned from surveys, instead of evaluating surveys by their conformity with his preconceived notions.

US statistics are politicized too.  Bobi’s story of the US government deciding, in 2007, to measure “food insecurity” instead of literal hunger provides the probable reason in its note about the political sensitivity of the hunger concept—for good reason, with so many Americans being hungry.

It reminds me very much of the cancellation of the pioneering reference book “US Social Indicators,” a massive collection of statistics on all aspects of the quality of life in the United States, after only two issues, in 1973 and in 1976.  This happened in the time of Republican President Ronald Reagan (1981-88).  It practically broke the heart of the director of the first volumes, Dr. Denis Johnston, my personal friend, who told me that Republicans felt threatened by such statistics.  Denis was my faculty colleague in a number of international training seminars, in the Philippines and abroad.

Is it just coincidence that the cancellation of official US statistics on literal hunger happened during the administration of Republican President George W. Bush (2001-08)?  I must remember to ask my American friends in the Quality of Life biz about this some time.

*  *  *

The SWS surveys on deprivation are now taken seriously by the World Bank.

The following is from its “Philippines Quarterly Update,” released this week (pages 3-4): “Following some rapid improvement through mid-2010, poverty and especially hunger have been worsening anew (Figure 6).  Self-rated poverty and hunger incidence rose noticeably in March.  Self-rated poverty incidence also worsened as 51 percent (10.8 million) of households considered themselves poor against 43 percent in March 2010.  Meanwhile, the latest hunger incidence of 20.5 percent (around 3.4 million families) is 2.4 points higher than in the previous quarter and is nearing the record high of 21 percent reached from December 2009 to June 2010.  The self-rated poverty threshold has also remained flat for several years, indicating that poor families have been lowering their living standards in recent years.”  [The June 2011 figures on hunger and self-rated poverty were not yet available to the World Bank when this was written.]

*  *  *

Contact SWS: www.sws.org.ph or [email protected].