WEEK 11 - research next steps in journey

P values and NHST

p value tells us how compatible data is with the null hypothesis

NHST- produce a NULL hypothesis and ALTERNATIVE hypothesis, choose signifiance alpha level, calculate how likely the data is if the null is true, reject the null if the probablity is small

Statistical inference

drawing conclusions of a population based on sample data

approaches - NHST and estimation

beyond p values shift

effect size - how big the effect is

precision - how certain we are (confidence intervals)

context - whether effect matters in real research settings

estimation approach

effect size - quantifies magnitude of a effect, how large or meaningful it is

confidence levels (CI) - gives a range of plausible values for the population ffect, shows precision of estimate

effect sizes and power problems

underpowered studies reduce reliability of research findings

lack of stats power, small samples don’t detect real effects which leads to false negatives, inconsistent findings and poor replication

stats power

the probability of detecting a effect if it exists

depends on - sample size, effect size and alpha levels

should be conducted before data collection

FAIR principles

• Findable → clearly labelled and searchable with metadata

• Accessible → retrievable through standard systems

• Interoperable → compatible with other datasets and tools

• Reusable → well-described so others can reuse it

open science

transparency in research methods and data, sharing data for replication and reanalysis,improving reproducibility

ethical issues with sharing qualitative data, concerns about context loss in re-analysis, tension between positivist (objective) and qualitative (interpretive) approaches

publishing and peer review

academics often contribute to unpaid labour, often behind paywalls with author fees

peer review process- experts review manuscripts anonymously and evaluate the methodology, theory, ethics and clarity

outcomes- accept, minor revision, majory revision, reject

authorship rules

ICMJE says - authors must contribute to design or data collection/analysis, to writing or revising the paper, approve the final version and take responsibility for the work

publishing pressure and fraud risk

reliance on metrics like - citation counts, H-index, altmetrics

risks to practices/fraud, quanitity over quality and publish or perish culture