Grant: Undergraduate Student Research Awards (USRA)
University of Toronto
$6,000.00
Estimating Publication Bias and False Discovery Risk in Clinical Trials
EF 2024-25
Résumé
The social and medical sciences are facing a crisis of confidence. Due to pressures to publish and biases against
inconclusive findings, researchers are motivated to report statistically significant results. This leads to
publication bias; that is, selective publishing of statistically significant results. However, selective publishing
of significant results makes statistical significance uninformative and makes it impossible to distinguish true
and false findings. It also inflates effect size estimates in original studies and meta-analyses. In medicine, an
influential article by Ioannidis suggested that published results report more false than true results1 In
psychology, an attempt to replicate original findings found only 37% of significant results in original studies
produced a significant result again in an unbiased replication attempt. Since 2012, I have developed statistical
tools to examine publication bias and to estimate the false discovery risk of published research findings. This
research project will use the z-curve method to compare publication bias and false discovery rates in Cochrane
meta-analyses and corresponding original articles.
Programme
Undergraduate Student Research Awards (USRA)
Bénéficiaire
Emplacement
Ontario, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
CIHR Funding Decisions Database
Tel que publié
Id
506760
Name
Tweneboah Natasha
Region
Ontario
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
Estimating Publication Bias and False Discovery Risk in Clinical Trials
Programtype2
Miscellaneous Programs
Orgnameinp2
University of Toronto
Cihramount2
$6,000
Cihrequipment2
$0
Programname2
Undergraduate Student Research Awards (USRA)
Keyworddelim
Clinical Trials; Cochrane; Publication Bias; Research Findings; Statistically Significant
Abstract
The social and medical sciences are facing a crisis of confidence. Due to pressures to publish and biases against
inconclusive findings, researchers are motivated to report statistically significant results. This leads to
publication bias; that is, selective publishing of statistically significant results. However, selective publishing
of significant results makes statistical significance uninformative and makes it impossible to distinguish true
and false findings. It also inflates effect size estimates in original studies and meta-analyses. In medicine, an
influential article by Ioannidis suggested that published results report more false than true results1 In
psychology, an attempt to replicate original findings found only 37% of significant results in original studies
produced a significant result again in an unbiased replication attempt. Since 2012, I have developed statistical
tools to examine publication bias and to estimate the false discovery risk of published research findings. This
research project will use the z-curve method to compare publication bias and false discovery rates in Cochrane
meta-analyses and corresponding original articles.
Competitiondate
202405
Approvedterm2
0 yr 6 mths
Pinamesdelim
Tweneboah, Natasha
Approvedterm
0 an 6 mois
Cihrcontribution2
~ $6,000~ $0~