Please enable JavaScript.
Coggle requires JavaScript to display documents.
Origins of a Replication Crisis - Coggle Diagram
Origins of a Replication Crisis
Incentive structure of academia
Papers required to get:
Job
Grants
Performance reviews
Prestige
Promotion
Led to:
Small sample sizes
Lots of papers
Unreliable estimates
Type 1 errors (false positives)
Frith (2019)
Wrote paper about struggles of writing reports/studies
"More speed, more haste, more stress, more waste"
Drive for novel findings
Journals publishing cutting-edge research articles
Novel research methodologies
Novel + broadly important data
Publication bias
Significant results have a better chance of being published, published earlier + published in journals with greater prestige compared to null/negative results
Can lead to psychologists only looking for positive findings (
p
<.05)
Questionable Research Practices (QRPs)
Fraud
Academic misconduct/Fraud = When a researcher makes an effort to falsify or misinterpret data
QRPs are not seen as fraud - 'grey area'
Example of academic misconduct/fraud: Diederik Stapel
130 papers, 24 book chapters
Developed studies + pretended he conducted them
Confessed to previous point
Criminally prosecuted
54 papers retracted so far
John et al (2012)
QRPs are a research norm
P-hacking
When someone unnecessarily influences data collection or statistical analyses performed to produce a statistically significant result
Can be done in numerous ways
Optional stopping
When a researcher repeatedly analyses data during the data collection process and stopping once they have found a significant finding
Dropping outliers after looking at data
When researcher analyses data and decides to remove certain data to achieve significant results
Cherry picking/ Selective reporting
When a researcher only reports certain data and omits others
Running multiple analyses
Dropping ppts after looking at data
Different measures for same variables
Ignoring certain DVs
HARKing
Hypothesising After the Results are Known
Researcher looks at data, then develops hypotheses
Paper written as a 'story'
Increases chances of Type 1 error (False positive)
May mask 'no effect' - wasteful resources when people try to repeat study/experiment