psych final quiz 4/23

QRPs

research practices that produces bias results through p-hacking, HARKing, and selective reporting

p-hacking

reanalyzing data repeatedly until statistically significant results are found

HARKing

present exploratory findings as if they were predicted before

selective reporting

only reporting significant results

how do QRPs affect research findings

inflate type I errors/false positives

type I error

false positive, incorrectly rejecting null

importance of reporting effect sizes

show importance of result not just significance

confidence intervals

ranges that esimate where true value likely falls

power analysis

justify sample size and improves reliability

reducing QRPs

preregistration, transparency, proper reporting

preregistration

registering hypothesis and analysis plans before collecting data

importance of replication

ensures findings are reliable, no false positives

statistical power

probability of detecting real effect

low power study

findings are unreliable and may miss real effects

sample size and power

larger samples increase power

main purpose of preregistration

prevent p-hacking and HARKing

p value

probability of the data assuming null is true

reject null

p<0.05

limitation to p-value

do not indicate effect size or importance

should be reported alongside p-values

effect sizes and confidence intervals

example of p value limitations

marshmallow test didn’t replicate, weaker effects than originally thought

effect size

strength of relationship

factor analysis

identify underlying variables in data

t test

compare averages of two groups

ANOVA

compare averages for 3+ groups

WEIRD

western, educated, industrialized, rich, democratic

limitations to WEIRD

not representative to whole population, bias for money/personal interest

MTurk

online plateform to easily recruit participants

advantages of MTurk

large sample size, low cost, fast

limitations of MTurk

not representative sample, bias for participating

original marshmallow test claims

ability to delay gratification preducts future success

marshmallow replication

not strong effect, findings did not replicate

correlated to personality simularity

life satisfaction in couples

positivity bias

people imagine future events more positively than past events