PSY2001 SEMESTER 2 - WEEK 9 Flashcards
(36 cards)
whats meta-science
using sciences to test science
what does replication do
- protects against false positives/sampling error
- controls for artifacts
- addresses researcher fraud
- test whether findings generalise to different populations
- test same hypothesis using different procedure
define direct replication
scientific attempt to recreate critical elements of og study
with same finding indicating finding accurate/reproducible
define conceptual replication
tests same hypothesis with diff procedures
same result indicate finding robust to alternative research designs, operational definition, sample
in open science collaboration (2015), how many different study actually replicate?
36%
name 2 findings that commonly do not replicate consistently
priming intelligence
use of spatial distance cues to prime peoples feelings of emotional closeness to their families
name 5 reasons for non replications
faking,
sloppy science,
moderators,
scientist error,
publication bias
give well-known examples of faking
Diederik Stapel- Dutch social psychologist: v well known, and influential. 2011, came out as Fraud, struck off and 50 paper retracted
name 9 circles of scientific hell (sloppy science), getting worse as go down
limbo
overselling
post-hoc storytelling
p-value fishing
creative outliers
plagiarism
non-publication
partial publication
define non-publication
deciding not to publish ie for bad reasons, because of bad collection…
define partial publication
only publishing sections of results which you like
what is outcome switching/p-value fishing?
changing outcomes of interest, depend on observed result
give example for p-hacking
taking decisions to maximise likelihood of statistically significant effect, rather than on objective, or scientific grounds
what are issues with small samples
lack statistical power= likely finds effect
due to chance
how commons sloppy science (John, 2012)
survey, 2,000 psychologists in US on their involvements in questionable research practices ie;
- failing to report all measure/condition
- deciding to collect more data after looking to see if results signif
- selectively reporting study
CONCLUSION: % of respondents who have engaged in questionable research, surprisingly high
what did Simmons (2011) find regarding flexibility in data collections being issues, regarding musical songs
listen to Beatles song, or Kalimba. then in a unrelated task, indicated birth date and fathers age, and used fathers age to control and found a predicted effect that those were year and a half younger after listening to beatles >Kalimba
whats moderators
variables that influence nature of effect (country/culture)
identification improves research and understanding of any second generation research
what is scientist error important to remember?
Doyen et al criticised Bargh research on priming effect, but actually Doyen hadn’t used correct methodology, etc
so is reminder of actual paper being correct, and criticising papers being incorrect!!!
define publication bias
findings statistically signif more likely to be published than those that are not
can be for good reasons: ambiguity over reason for null finding however,
whats file drawer problem
published studies represent 5% of findings occurring by chances alone?
define open science
process of making science and process of producing evidence and claims transparent and accessible to others
name 6 principles of open science
data
source
access
methodology
peer review
educational resources
define open methodology
documenting method, process how method developed/decided for
define pre registration
define RQ, methods, approach to analysis before observing research outcomes = prevent HARKing
