Putanumonit: What statistical power means, and why I'm terrified about psychology

post by Jacobian · 2017-06-21T18:29:48.906Z · score: 11 (11 votes) · LW · GW · Legacy · 8 comments

This is a link post for https://putanumonit.com/2017/06/20/the-battle-for-psychology/

8 comments

Comments sorted by top scores.

comment by gwern · 2017-06-21T22:39:12.414Z · score: 4 (4 votes) · LW · GW

I'm not totally sure that is how measurement error works, and the 1ug/dl choice is rather arbitrary (looking at your previous post, isn't it already uncorrected for measurement error?). In any case, I think bringing in additional complications like measurement error (itself just one of several things that could be reducing power in your standard social psychology experiment) makes the core points about statistical power harder to understand.

comment by cousin_it · 2017-06-21T19:29:37.706Z · score: 3 (3 votes) · LW · GW

Nice read! For a more light-hearted take on this, see Shalizi's 2010 post The Neutral Model of Inquiry :-)

comment by Jacobian · 2017-06-21T20:11:15.547Z · score: 1 (1 votes) · LW · GW

That's a great link, and one quote, in particular, seems very relevant especially if lightly adapted:

It's true that none of these findings will last forever, but this constant overturning of old ideas by new discoveries is just part of what makes this such a dynamic time in the field of [psychology]. Many scholars will even tell you that their favorite part of being a [psychologist] is the frequency with which a new sacrifice over-turns everything they thought they knew about [deep human values] from a [5 point scale]!

comment by MrMind · 2017-06-26T14:07:37.908Z · score: 0 (0 votes) · LW · GW

Is it correct to calculate the standard deviation error in the cortisol measurement including the first term? In the original paper (CC&Y) it's not clear how they analyzed the cortisol results: did they averaged the cortisol level at time 1 and the level at time 2? Or did they just averaged the difference of the samples?
Because in that second case, you wouldn't need to account for the variation per person, you just need the average variation (your 2.5) and the inter-assay CV (which you put around 2.8, definitely not something you can left unspecified).

comment by Regex · 2017-06-25T00:05:17.595Z · score: 0 (0 votes) · LW · GW

Before even reading it I was confused.

Epistemic status for the first part of this post:

[image of thinking woman in front of math]

Epistemic status for the second part:

[Image of greek? philosopher preaching]

Admittedly I should probably know who the second image is of, but I have no idea what they're trying to say with either of these.

As we say in the Bayesian conspiracy: even if you’re not interested in base rates, base rates are interested in you.

No. Stop. This is just awkward to read.

comment by gjm · 2017-06-25T03:10:03.540Z · score: 2 (2 votes) · LW · GW

Well, the second picture is a 400-year-old painting, depicting Ezekiel's vision of the valley of dry bones. (So presumably the guy preaching is Ezekiel, obeying God's instructions to tell the bones to join together and form bodies.)

This still leaves it rather unclear what the two pictures mean as epistemic statuses, but I take the first to mean "here I am just explaining what the mathematics says, so there's not much scope for doubt" and the second to mean "here I am preaching religion, and you may if you wish dismiss me as a fanatic". This seems consistent with what he says at the start of the second part: "Part 1 was a careful explanation of a subject I know very well. Part 2 is a controversial rant about a subject I don’t."

comment by Jacobian · 2017-06-25T06:37:01.836Z · score: 2 (2 votes) · LW · GW

I've been hanging too long with the Ribbonfarm crowd where the default epistemic status is "clown on fire jumping out of a helicopter".

Part I is just math, part II is a prophecy of doom, and the pictures mean that I didn't feel like coming up with actual epistemic probabilities for both.

comment by scarcegreengrass · 2017-06-22T21:35:29.724Z · score: 0 (0 votes) · LW · GW

That's some good snark about the six-pointed cross.