2 thoughts on “Monday Message Board

  1. I’m intrigued by the constant media referrals to the Liberal choice for the upcoming Wentworth bye-election as a :’former ambassador to Israel’. Even David Shama himself said in an ABC interview
    that he was, or had been a diplomat, no specific reference to where. Is this media specificity done with
    one eye on the large Jewish population of the eastern suburbs electorate? It seems oddly promotional.
    So I’m wondering if the same specificity will apply to favoured independent candidate, Dr Kerrie Phelps
    as a promoter of alternative therapies. She’s written a book about their practice. So far it is her former
    AMA presidency that the media seems worth mentioning.

  2. Generations?!
    Nyt articles has fancy graphics and my takeaways;
    1) “Events at age 18 are about three times as powerful as those at age 40, according to the model”. So depending on events at 18 a whole age group are shoehorned into one type of generational group. X, Z, oughts.
    AGE is relevant. COHORT  may be relevant. We seem to be conflating ‘PERIOD’ with experience and or events. I solidly agree with the effect size of 3x bigger at 18 compared to 40, yet i would be an outlier as I left school asap as I wanted freedom. My “generation” was totally irrelevant. At 35, working in the engineering consulting community I finally took an interest in why some people had attitudes I did not understand. So my effect size at 35 was 3x my 18.
    2) “The model works best for white voters.” One day we will overcome this bias too, which again conflates period with lived experience.

    I am keen to have this generational blame game statistically revealed. Or at least a better huristic known so I can ignore many statements from media, polies and my mum.

    I’d appreciate your take (anyone) on the model below for its ability to transfer from health to politic/ policy and socioeconomics and to events such as brexit… but “from which it is statistically impossible to estimate unique estimates for the three effects”.


    “”The presence of perfectly collinear predictors (age, period and cohort) in a regression model will produce a singular non-identifiable design matrix, from which it is statistically impossible to estimate unique estimates for the three effects. (5)””. Is this a model assumptions or design fault or data ?

    It the goes on to dicuss solutions to the identification problem. Maybe these types of analysis and huristics are only useful in very specific situations. Maybe one day it will be like genetics and epigenetics. A cloud of a billion genes and ways we express them. I’m not going to hold my breath tho. We may be discussing this forever. Thanks.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s