Thread by @PJACaytonPhD

Pulse Asia's survey methodology is typical of a private survey firm's approach to sampling given limited resources but still able to make some measurable conclusions.

I am not an agent of Pulse Asia, but Pulse Asia has technical notes able to be understood by people exposed to survey sampling methods.

pulseasia.ph/ulat-ng-bayan-…

Let break this down:

1st, Pulse Asia's domain of concern in studying to have measurable accuracy with sampling is the four geographic areas: NCR, Luzon w/o NCR (termed Balance Luzon), Visayas, and Mindanao, each of the four alloted 300 samples each, thus 1,200 samples.

Why 300? Well, it's a good round number for a desired 6% margin of error, or around that.

For very large populations of which we would like to be about 95% confident that the true proportion of approving citizens is within 6% margin of error is 266.777... or 300 is round enough.

The regional breakdown of margin of error? That is computed for the people post-hoc, and regional estimation is not really the targeted level of the survey to be accurate.

They computed it still to be open to those who would like to see it.

I would need a class setting to explain the other aspects of the sampling methodology, but whatbI can say is nothing too out of the ordinary is in the technical notes. I kinda remember the survey operations class in undergrad HAHAHAHA

Could they have done better in sampling methodology? Sure! But sample size determination is a statistical and a cost decision, and they are a business. They have limited resources. I don't know the financial situation of Pulse Asia to say why exactly they used this method.

Thanks for attending the TED talk on survey sampling!

I know statisticians who can explain it much better than me hahahaha!