Is social psychology self-correcting?
Social psychology has not self-corrected political bias. No hiring or journal fixes exist; “science is self-correcting” assumes machinery never built.
Read Article →Category
scientific validity
14 posts
Hurricanes as imagined, construed, and filtered
Klotzbach and Landsea find no significant rise in Cat 4-5 storms after better data. Media will hype scare papers and bury this--filtered climate reality.
Read Article →Nimble skepticism (Updated)
If new evidence kills the warming slowdown, skeptics should drop the identity. Climate is not a team; surface records are estimates, not meter readings.
Read Article →The climate science consensus is 78 – 84 percent (Updated)
Qualified surveys of climate scientists put attribution consensus at 78-84%, not 97%. Cook-style abstract ratings are not surveys and should not set it.
Read Article →The “hottest year on record” fallacy
On a temperature plateau, any year has a decent shot at “hottest on record.” That slogan does not refute a pause; some scientists botched the probability.
Read Article →p-values banned
Basic and Applied Social Psychology banned p-values and confidence intervals. Descriptive stats outrank inferential ones; a small p is not an effect.
Read Article →SPSP; data
SPSP’s integrity symposium is a draw. Student samples weaken claims; invalid methods make data cheap. Field and workplace samples would travel farther.
Read Article →Methods
To learn why skeptics think as they do, read their own accounts--not a blog survey that ties them to moon-hoax items almost none of them endorse.
Read Article →Structure of Malpractice 1: The Frog Jump
The frog jump: correlate a lopsided agree/disagree scale, then report the empty opposite pole. Linear r on dichotomous items can invert the truth.
Read Article →Significance
Statistical significance is not binary. The 0.05 cutoff is a ten-finger convention for Type-1 control--not proof a hypothesis is true in nature.
Read Article →Comment on Verheggen et al.; climate consensus research
Verheggen et al. invited non-climate authors and cannot say who answered. Consensus research is often junk; valid surveys still show a real majority.
Read Article →No results
Cook raters googled authors and mocked skeptics. Fraud plus an invalid method means there is no 97%--no number at all. Results require methods.
Read Article →Data is often overrated
Invalid studies need no new data to be vacated. Logic can refute Napier & Jost and Lewandowsky; more numbers cannot rescue a validity collapse.
Read Article →Reality is what it is
Unflattering findings about conservatives are not automatically invalid. Reality is whatever it is; Eidelman et al. looks methodologically clean.
Read Article →