Glad it was useful 😉

Correct link (without typo) to the blog by Cox: tinyurl.com/28taeyet

This was right after I finished my dissertation, so a sudden decrease is more likely!
Correct link to the blog by Cox: tinyurl.com/28taeyet

7/7 In conclusion, although this is an elegant and theoretically interesting model, if the empirical premise of its explanation for polarization does not hold, I'm not sure what to make of it.

6/7 Finally, such a marked increase in connectivity cause by social media is not known in the (sociological) literature on personal networks, nor does it show up in other data sources (e.g., tinyurl.com/28taeyet)%3E Indeed, this would have been remarkable discovery, worthy of a paper in itself!

5/7 Most of the "transition" around 2008 seems driven by the fact that almost all data points before and after are based on different measurement instruments, respectively name generators ("with whom did you discuss...?") and asking for aggregated numbers ("how many close friends do you have?")

4/7 The data sources for some of the data points are rather unclear. For example, to my knowledge, core discussion networks were not measured in the GSS after 2010 but they are reported in the Figure. Also some other data points (e.g., for Norway and Germany) are attributed to the wrong data sources

3/7 The supposed "transition" in connectivity around 2008 is based on the data points of 4 countries, but then compared to polarization in only the US. If one would take only the US data points for connectivity too, the transition is much less clear

2/7 The selection of countries is strangely selective, focusing only on the US and North-West Europe

Following up on this: should we start new journals (which creates huge coordination problems) or try to wrangle our beloved top journals out of the hands of the big publishers? Are there any good examples/strategies for the latter?

Thanks for rubbing it in...

Sure, I don't think these are conflicting views. *If* the data allow for a causal interpretation, all the better. My point was rather that even if they don't, that doesn't mean we cannot learn anything about causality at all (which seems a common position in some circles, but I don't mean you)

How I've always understood the logic of observational research is that we test (and try to falsify) predictions about correlations that we should observe in the data, given our *causal* theory. We don't infer causality from correlations, but the other way round!

Daarnaast hoort een beoordelaar in goede collegiale verhoudingen ook niet "bang" te zijn voor een examencommissie (zeg ik als lid van de examencommissie), hoewel ik snap dat je geen zin hebt in gedoe.

M.i. hoort een 10 niet voor een platonische "perfecte scriptie" te zijn, maar voor een werkstuk dat geheel voldoet aan de beoordelingscriteria (wat best zeldzaam mag zijn). Als het voor een (menselijke) student theoretisch onmogelijk is aan de criteria te voldoen, dan kloppen die criteria dus niet.

Waarom geen 10 voor scripties?

Maar het probleem daarmee is weer dat de criteria per universiteit verschillen, net als de regelingen rond herkansingen etc (wat potentieel weer effect heeft op de cijfers). Het vergroot bovendien kunstmatig het verschil tussen mensen net onder en boven de streep. Ook afschaffen wmb dus.