First, the RCT is a much stronger study. I'm not sure why you're picking a fight with a correlational paper when there is a causal manipulation that I linked first.

Because I don't have access to that one. It could be amazing, it could be shit, I don't know.

1B isn't the graph of productivity; 1C is.

Good thing I was talking about output then, and not productivity!

You can't just look at a graph, either--you need statistics.

Thanks, I do actually remember a couple of things from when I got my doctorate. And my gripe is with the statistical analysis, or the

One of those things you learn in statistics is that you can't ignore trends when analysing data. You also can't just control for external factor just by doing some ordinary least square regression.

Columns 5 and 6 show that both WFH effect estimates on Productivity are negative, but only the estimate with seasonal time trend is significantly different from zero. We prefer that specification, since both the plot and the linear time trend coefficient indicate that a linear trend is not as appropriate.

Authors are assuming productivity (which is based basically on output / time worked) is seasonal, when the time worked is barely patterned, and output is very clearly not seasonal in the data shown. If it was seasonal, you would expect a roughly similar increase in at the same time each year, say February. What we see is the exact reverse.

There may be periodicity, but seasonality is not at all shown, and there is too little data shown to make any claims about periodicity in output. Say they're on an 18 month development cycle, and you're controlling for a winter-dip, you'd be skewing your data hugely for no reason.

In other words, the data wasn't significant (see emphasis) until they "corrected" for a factor that they haven't shown exists, and is, in my opinion, completely counterfactual.

And even if we assume the authors are correct, they only show the whole system was less efficient in the first 5 months of WFH, which is hardly a surprise to anyone. They don't show the cause lies with the workers, even though they constantly use wording that seems to imply that it does. I promise you it wasn't the coders who were calling for constant meetings.

So to sum up:

This paper completely assumes a demonstrably incorrect time factor, and uses it to create significance where there was none before. And even if you ignore that, they place blame on workers instead of on the system.