I voted too -- I'm not satisfied with them, but see I know we got some self--selection bias here so I think we should be careful with the figures. Anyway, keep it up -- your blog is perhaps one of the few candles in the dark we have in Mauritius!
Thanks. True, small samples can be afflicted with all kinds of biases. But didn't we see quite a few surveys with large samples that had predicted a Clinton presidency?
Here too we've seen much larger polls get it quite wrong.
I am definitely not going to put any poll, big or small, on a pedestal. I rather look at its information content (its shape essentially) and we know that small samples can capture most of the information of a much larger one if the answers have enough variability.
Hopefully Collendavelloo will move into the opposition soon so we can check its accuracy.
I voted too -- I'm not satisfied with them, but see I know we got some self--selection bias here so I think we should be careful with the figures. Anyway, keep it up -- your blog is perhaps one of the few candles in the dark we have in Mauritius!
ReplyDeleteThanks. True, small samples can be afflicted with all kinds of biases. But didn't we see quite a few surveys with large samples that had predicted a Clinton presidency?
ReplyDeleteHere too we've seen much larger polls get it quite wrong.
I am definitely not going to put any poll, big or small, on a pedestal. I rather look at its information content (its shape essentially) and we know that small samples can capture most of the information of a much larger one if the answers have enough variability.
Hopefully Collendavelloo will move into the opposition soon so we can check its accuracy.