Surveying the field
SURVEYS are now the stuff of news, even in social media, especially when dealing with preferences for the top two spots “if elections were held today,” and not when it’s scheduled next year in May. Hopefully, by then the pandemic would be over, and you happen to be where you are registered as a voter, and inclined to line up for your turn at the booth, without having to wear a face mask.
Except for Viber groups, which are not representative of the country at all, soliciting informal opinions, I have never been approached as a respondent for nationwide surveys. I am not sure how they randomize the respondents or how I’ve gotten skipped all the past election years. Then again, I haven’t won random events like lotto, raffles, and door prizes either.
Do surveys use stratified clusters based on search engine profiles and interests? What about the non-digital natives who also vote? Are queries still conducted by field surveys in deserted malls and train stations? Sampling can be a tricky business, even without a pandemic. Clue: it’s even harder with masked respondents who avoid strangers with notebooks.
Market researches for consumer products or services are more focused and clearer on what their clients are looking for — why people buy their products, or don’t. For restaurants, it is routine for the waiter handing you the bill to also ask you to fill up a questionnaire with a tiny pencil without an eraser. You are asked for comments on the food and the quality of the service — did you get a back massage? Sorry we no longer offer hot towels.
In social media, surveys are “crowd sourced” and posted in real time. Thus, comments of customers on restaurants and hotels are gathered by ratings agencies who then give their seal of approval with stickers and on-line reviews. Still, bad experiences can be over-represented. Unhappy customers who shout at waiters take the time to let their bile rise with scathing reviews (there was a cockroach waving at me from the chandelier). On the other hand, satisfied customers tend to be lazy with their comments, and just read the review of others.
The reliability of surveys depends on methodology, mainly the sample size and stratification of respondents over the relevant population which determines the statistical margin for error. Thus, a 100% sample, as in actual elections, has zero margin for error (except for rigged machines), making the exit interview the most reliable survey even with a small sample. It is also too late to be of much help to the tail-enders.
Surveys anyway are not always the predictive tools that people expect them to be. The data can be massaged by bias or vested interests. Name familiarity too figures in the choice of respondents. (So what if he is better known as a boxer.)
At this time, surveys should be more diagnostic in nature, providing feedback on a product and how it can be improved. How does the tail-ender raise his profile? How does he appeal to the “don’t knows”? Where is the nearest bathroom? In social interactions, we don’t need surveys to determine the mood of a particular object of study. Text messages (I lose sleep thinking of all the expenses you keep racking up) and facial expressions, after the mask is removed, determine the chances of being alone with her in a dark room where she has nothing on but the radio — retro music?
The importance of surveys on an almost-declared candidate’s ranking, from prompted choices this early in the political season, lies in any or all of the following: a.) claiming bragging rights as a legitimate contender; b.) ability to attract top operatives and financial backing beyond internally generated funds from the importation of vaccines; and c.) getting free media coverage — I still haven’t decided if I’m going to run. Okay, are you taking back the sneakers?
What about the independence of survey companies which routinely hoist near 100% the favorability rating of the leader? Credibility should rest on the transparency of the methodology capable of delivering independent and reliable results.
Anyway, survey companies have clients too. So it is not unusual to reverse the usual survey process. Start with the desired survey results, and then figure out the methodology to achieve these. It’s a lot cheaper to do… and command a higher fee too.
Do surveys use stratified clusters based on search engine profiles and interests? What about the non-digital natives who also vote? Are queries still conducted by field surveys in deserted malls and train stations? Sampling can be a tricky business, even without a pandemic. Clue: it’s even harder with masked respondents who avoid strangers with notebooks.