Survey says!

There’s a TV show entitled Family Feud that we’re all familiar with.  The contestants try to guess the top answers to a topic or question on the board with the host revealing the answers by shouting – Survey says!

My lectures in Research Methodologies includes Surveys and Questionnaires.

I will not make this blog complicated.  It is not a discourse in mathematics and statistics.  While there are people that will pretend to know about this topic, I am writing this for you.  Sharing it among the misinformed can help guide them in the interpretation and utilisation of surveys, as well as educate them on the the advantages and disadvantages of relying on surveys as a mathematical basis for outcome analysis.

What is a survey?

It’s a research method used to collect information about a population of interest.

It is a reflection of information of a single point in time. TODAY. In short, it means that a survey today cannot be extrapolated with absolute certainty to something that has not happened yet.  Many circumstances that change will with certainty affect outcomes.

Let us use the last presidential election as a classic example on how fickle minded surveys are.  The Presidential race was a shoo-in for Vice President Jejomar Binay.  When Grace Poe threw the gauntlet, there was a shift of tide. The dark horse Rodrigo Duterte won that election. You could tell that when the people marched to the polls, Poe was going to slip further down the road.  Roxas landed a far second from Duterte. But the Liberal Party knew that from the get go.  Unless Poe gave up her candidacy, there would be no way Roxas would win over Duterte. And Poe should have known that a neophyte would never win this election.

The Vice-Presidential race saw independent candidate Chiz Escudero at the forefront at the start of the surveys.  Subsequently BongBong Marcos began to climb slowly to eventually lead the pack.  A month before the Vice-Presidential debate, the young Marcos was ahead. During the debates, Allan Peter Cayetano practically delivered the Vice-Presidency to Robredo on a silver platter. His maligning the young Marcos and resurrecting the Martial Law regime and the billions that they stole provided the setback needed to push Robredo to the forefront.  That one and only debate was the final nail on the coffin.  The aftermath of that debate was a winning moment for a dark horse like Robredo.  Even the surveys agreed. A week before the elections, Robredo and Marcos were statistically tied. Marcos should have learned from the last presidential election where Roxas lost to Binay.

It is important to know that there are various types of surveys. There are also various ways of conducting a survey. And finally, there are many ways a sampling method is done.

Because of the complexities in making sure that a reliable survey is well designed, a poorly done survey, is a biased one. It is important to remember that bias sampling is the best way to skew data to favour desired outcome.  Appropriate randomisation in order to clearly reflect the true population being surveyed is vital to the reliability and accuracy of any survey.

Remember – a survey is made up of a sample from the population.  Unlike a census, where each and every person inclusible is included, a survey utilises only a sample.  That sample SHOULD be highly representative of the population the survey intends to address.

Surveys (and questionnaires) can be self-administered. That means that I give you a piece of paper or ask you a question online and you just provide the answer.  It’s usually answerable by multiple choice(s).

Then there are those that use enablers or enumerators.  In short, there are people that are sent to the field and these people conduct the interviews by asking questions to the respondent.  Using enablers is an additional layer of bias.  How the interviewee replies to the question and how the respondent is approached by the enabler can affect the response.  How the enabler chooses the respondent when the actual respondent targeted is not there is the largest bias called convenience sampling.

Leo Laroza, senior survey research and communications specialist from the Social Weather Station provided the methodology the SWS uses in undertaking their surveys.  From choosing 1,200 (because it’s an internationally accepted number sample size?!? is a lame answer) as the number of respondents of voting age. From the way the stratified random sampling is conducted, the likelihood of the survey results being accurate and reliable should be considered.  And they peg their errors at +/- a certain percentage.  The basis for the percentage error should be disclosed, particularly with how many hits and misses they had for an enabler assisted survey.

Their constraints on areas that are not readily reachable compared to the more accessible ones introduces bias already.  Another bias is on dividing the 1,200 into 4 areas – Metro Manila, Luzon, Visayas, and Mindanao.  Then appropriating 300 subjects for each of these four areas, and further subdividing these 300 into 60 barangays (which means 5 people per barangay for Metro Manila, the rest of Luzon, the whole of the Visayas, and the whole of Mindanao).

Interpreting surveys

Surveys are information guide. Taken now. A snapshot of time.

Reflective of data from a sample that may be representative of the sentiment of the population depending on whether the sampling process was appropriately conducted or not.

The results can be interpreted to fit what what one wants to sell.

Let’s not rely too much on surveys to check how the landscape of the political climate in the country is evolving. The better indicator is our moral compass.

It is barely 7 months into the midterm election.  We need to focus on the ball.  Those numbers, remain unreliable until the final candidates are announced.  Remember, it is not the opinion alone that matters.  It is going to the polls and showing that there is strength in your vote that will change the landscape of politics.

Our votes are the only opinions that count.