International Political Analyst Lectures Richard Heydarian on "Margin of Error" Issue (Video)

International political analyst and veteran blogger Sass Rogando Sasot lectures GMA's resident political analyst Richard Heydarian on his interpretation of the "margin of error" in the latest Social Weather Station (SWS) survey results. 



Richard Heydarian explained the margin of error by saying "First of all dapat i-take into consideration natin iyong margin of error -- 3-5% overall. Sa mga sub-groups 5-6%. Ang ibig sabihin niyan eh dapat magkaroon ng lagpas 12% generally para magkaroon ng statistically significant change. So iyong 10 points ay parang hindi pa abot sa statistical relevance," Heydarian said.

Sass Sasot burns Heydarian by saying that the resident analyst of GMA News got it wrong and his interpretation of "margin of error" is indeed wrong and explained further the meaning of "margin of error" in the latest SWS survey.

Sasot stated that once you conduct a survey, you cannot asked all the population therefore you only get a sample. Margin of error and its concept of confidence level are indicators to determine how it represents the whole population in terms of the survey results.



Here's How Sass Sasot Lectures Richard Heydarian:

ANO BA ANG MARGIN OF ERROR?

cc: Richard Heydarian

Kamakailan tinanong ni Kara David si Heydarian tungkol sa pag-taas muli ng SWS satistifaction rating ni PRRD ng 10%. Sabi ni Heydarian, and I quote VERBATIM (https://youtu.be/2y9PM0yh3s0):

"First of all dapat i-take into consideration natin iyong margin of error -- 3-5% overall. Sa mga sub-groups 5-6%. Ang ibig sabihin niyan eh dapat magkaroon ng lagpas 12% generally para magkaroon ng statistically significant change. So iyong 10 points ay parang hindi pa abot sa statistical relevance."

As Heydarian said that, Kara David just mmm, as if having a eureka moment.


Mali ang pagkakaintindi ni Heydarian ng margin of error. So ano ba ang margin of error?

Kapag nagpapa-survey ka hindi mo lahat matatanong ang buong population. So you will just get a SAMPLE. Ngayon ang margin of error at ang kaakibat nitong konsepto na confidence level ay indicators para malaman mo kung gaano ka-representative ng buong population ang resulta ng survey mo.

So 95% ang confidence level (standard iyan) at +/- 3% nationwide ang margin of error ng SWS survey, at +/- 6% sa bawat regional groups.

Ang resulta ng survey ng SWS ay nationwide 71% ang satisfied kay Duterte. Ngayon papaano natin ito interpret na kasama ang margin of error? Ganito po: Kung ang survey ay uulitin ng 100 na beses, ang percentage na nagsabi na satisfied sila kay Duterte ay mag range between +/- 3% ng 71% (or 68 to 74%) 95% of the time. And that there will be 5% times na pwedeng mas mababa sa 68 o mas mataas sa 74.

Sa regionally naman +/- 6% ang margin of error. Bakit mas mataas kaysa sa nationwide? Dahil po mas mababa ang SAMPLE SIZE sa bawat region. Lower sample size = higher margin of error.

So let's interpret the regional result of Duterte satisfaction ratin considering the margin of error.

NCR = 73%
Kung ang survey ay uulitin ng 100 na beses, ang percentage na nagsabi sa NCR na satisfied sila kay Duterte ay mag range between +/- 6% from 73% (or 67 to 79%) 95% of the time. And that there will be 5% times na pwedeng mas mababa sa 67 o mas mataas sa 79.



Bal Luzon = 64%
Kung ang survey ay uulitin ng 100 na beses, ang percentage na nagsabi sa Bal. Luzon na satisfied sila kay Duterte ay mag range between +/- 6% from 64% (or 58 to 70%) 95% of the time. And that there will be 5% times na pwedeng mas mababa sa 58 o mas mataas sa 70.

Visayas = 72%
Kung ang survey ay uulitin ng 100 na beses, ang percentage na nagsabi sa Visayas na satisfied sila kay Duterte ay mag range between +/- 6% from 72% (or 66 to 78%) 95% of the time. And that there will be 5% times na pwedeng mas mababa sa 66 o mas mataas sa 78.

Mindanao = 84%
Kung ang survey ay uulitin ng 100 na beses, ang percentage na nagsabi sa Visayas na satisfied sila kay Duterte ay mag range between +/- 6% from 84% (or 78 to 90%) 95% of the time. And that there will be 5% times na pwedeng mas mababa sa 78 o mas mataas sa 90.

HINDI matutukoy ng margin of error kung ilang percentage ang dapat itaas para maging "statistically significant." Echosera itong si Heydarian.

Ito naman ang ibig sabihin ng term na "statistical significance" na sinasabi ni Heydarian: ang resulta ng test or experimento ay hindi bunga ng chance kundi pwedeng i-attribute sa cause na tinutukoy. Hindi magin of error in itself ang test ng statistical significance kundi p-value. Wala namang p-values ang SWS survey.

Watch the Interview with Richard Heydarian:



Source: Sass Rogando Sasot

International Political Analyst Lectures Richard Heydarian on "Margin of Error" Issue (Video) International Political Analyst Lectures Richard Heydarian on "Margin of Error" Issue (Video) Reviewed by Phil Newsome on January 03, 2018 Rating: 5
Powered by Blogger.