Current issues in research Flashcards
The definition of bias
Any trend or deviation from the truth in data collection, data analysis, interpretation and publication which can cause false conclusions (Gardenier & Resnik, 2002).
Bias can occur in plannning , anaylsis and interpretation
Bias in planning= flawed study plan, and selection bias what are the two types
there can be randomization issues, must ensure a clear timeline, blinding, single or double can effect outcome. and objective neasures.
Selection Bias:inadequate selection criteria avoid introducing bias through unclear selection or overly strict criteria for selection
Bias in planning 2
Channelling Bias:
* Assignment of Participants: Ensure participants are assigned to study groups using objective and rigorous criteria.
Review Bias:
* Methodological Influence: Beaware of how research methodology choices can introduce
bias.
* Literature Review Bias:Ensure balanced reporting of literature,avoiding selective citation in the introduction and review sections.
What is recall bias
participants are asked to remember past events.
o Objective Data Sources: Always prioritize objective data sources over subjective ones to reduce memory-based inaccuracies.
o Validated Subjective Measures: If subjective measures are necessary, ensure they are validated to improve reliability and minimize recall errors.
What is performance bias
Duffering care or encouragement between groups
Exclusion Bias
Exclusion Criteria: Clearly define exclusion criteria upfront to avoid arbitrary exclusion
of participants.
o Withdrawals: Track and analyse differences in participant withdrawals between groups during the study.
o Intention to Treat/ Per Protocol Analyses
Publication Bias/ the file drawer effect
occurs when the outcome of the study sways the decision to publish,
only showing strong results of findings
Types of research bias
P hacking:
inclusion/exclusion of data:
HARKing: hypothezizing after the results
selective reporting by trying multiple staistical analyses to achieve significant results
selectively including or excluding outliers of variables to influence results
formulating a hypothesis after the data is correlated, undermine the integrity of a hypothesis driven thesis
What is the null hypothesis?
Represents the default position or statement that there is no effect or no
difference.
o It is what researchers aim to test against, assuming no relationship exists between variables
statistical tests either fail to reject or reject the null hypothesis
Alternative hypothesis h1 or ha
represents a statement that contradicts the null hypothesis, if rejected the alternative hypothesis becomes plausible,
Null hypothesis significance testing :
- arbitary cut off (p ≤ o.o5):
2.
- The reliance on a strict threshold for statistical significance (commonly p ≤ 0.05)
is often arbitrary.
o This “yes or no” approach oversimplifies complex data, potentially leading to misinterpretation of results.
what the null hypothesis doesnt tell us
- Magnitude of the Effect: It tells us whether an effect exists, but not the size or practical importance of that effect. Knowing if a result is statistically significant doesn’t indicate how large or meaningful the effect is.
- Precision of the Estimate: NHST does not provide information on the confidence or accuracy of the effect size estimate. We need confidence intervals or other measures to determine how precise our estimate is.
what is the strength of association between vairables
magnitude statistics help quantifyb the strength between independant and dependent variables.
interpreting the magnitude of effect
effects sides allows a more meaningful interpretation, by showing real world evidence
What is the impact of manipulations
interprets how much an expirement or interpretation impacts the outcome
