Nope, theoretically true but not the least bit feasible in reality. My sample could at best represent my 160k people city, where 80% voted for our local Mussolini, not my 200m people country, which has wide regional differences (and didn't elect the local Mussolini, even though my city and some entire states would have)... imagine India, with almost 1.5b people, and those interviews can't possibly have the necessary geographical and social scope in countries that size with such an insignificantly small sample... what's their methodology? Random pop-ups on social media? You already excluded a huge chunk of the population that doesn't use social media at all, and depending on the social media, you just made an age cut as well, your sample is no longer random but only representing a certain demographic. Was it people standing in a big city interviewing passersby? Time and location will taint the results and, again, the geographical scope is insignificant. Random phone calls? Well-informed people have their numbers on "do-not-disturb" lists (a public service), don't answer random questions from strangers on the phone, etc, etc, so your sample is not random enough anymore because a demographic group won't be represented in it, etc, etc.