Please enable JavaScript.
Coggle requires JavaScript to display documents.
Open Science - Coggle Diagram
Open Science
-
Remedies
Open Science
- Proactive approach
- Fix problems before studies run
- Aim to improve scientific practice through remedying some of ongoing issues
Remedy Incentive
- Recreate incentive structure to reward good practice rather than good results
- Is process rather than outcome that is important
- Currently a worldwide movement across hundreds of institutions
- More journals accept non-significant results if paper otherwise sound
- Teach about proper scientific practice
Remedy Malpractice
- Success defined as rigorous methods & honest, transparent conduct
- Reward whistle-blowers who report fraud/QRPs/guilty parties
Replication
- Reactive approach
- Address problems after studies run
- Re-test hypothesis on new sample of Ps to see if results replicable
- If result real then should be detectable most of time
Replications
- Don't accept first result
- Sig results may happen by chance alone
- Conduct multiple replications
- Accept consistent sig effects as the real results
Power
- Ability to detect effect when there is one to be found
- Estimate of expected effect size & determine sample size based on desired power of study
- Important in study design & replicability
- Fundamental element of Neymann-Pearson philosophy of significance testing
Alpha
- Type I error
- Probability of concluding is an effect where one does not occur
- Usually set at .05
Beta
- Type II error
- Probability of concluding is not an effect where one does occur
- Usually set at .2
- Are willing to do 20% of time in long run
Calculating Power
- 1 - beta
- Lower means are lacking sensitivity to detect effects that are interesting/meaningful
- If effect found is unlikely to be replicated
Relationship
- Direct relationship between power
- Statistical power
- Effect are interested in
- Alpha level
- Sample size
- If know three values can calculate fourth
G*Power
- Free software to carry out power analysis
Calculating Effect Sizes
Cohen's d
- d = (u1-u2)/stdev
- Small effect: d = .2
- Medium effect: d = .5
- Large effect: d = .8
-
Cohen's f
- Use effect size calculator
- Uses F-value & degrees of freedom
-
-
-