What question did this study set out to answer?

To evaluate the empirical familywise error rates of commonly used parametric and nonparametric statistical methods across major fMRI software packages.

June 28, 2016Proceedings of the National Academy of SciencesOpen Access

Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates

View Full Paper

Ask AI

Bookmark

Authors

AEAnders EklundLinköping University TNThomas E. NicholsUniversity of Oxford HKHans KnutssonLund University

Discussion

Member takes

Implication

Empirical evaluation reveals inflated false-positive rates in common fMRI software packages, highlighting the need for validated statistical inference in neuroimaging.

Key Points

To evaluate the empirical familywise error rates of commonly used parametric and nonparametric statistical methods across major fMRI software packages.
Analyzed resting-state fMRI datasets across 3 million random task group analyses.
Evaluated parametric tools across SPM, FSL, and AFNI alongside a nonparametric permutation method at a nominal 5% familywise error rate.
Parametric statistical methods were conservative for voxelwise inference but invalid for clusterwise inference, producing markedly inflated false-positive rates beyond the nominal 5% threshold.
Spatial autocorrelation functions in real fMRI data did not follow the assumed Gaussian shape, serving as the primary cause of invalid clusterwise inferences.
Nonparametric permutation testing successfully achieved nominal familywise error rates for both voxelwise and clusterwise analyses.

Cite This Study

Eklund et al. (2016) studied this question.

synapsesocial.com/papers/69d6866fbcbb69330b88b5aa https://doi.org/10.1073/pnas.1602413113

View Full Paper

Ask AI

Bookmark

Implication

Empirical evaluation reveals inflated false-positive rates in common fMRI software packages, highlighting the need for validated statistical inference in neuroimaging.

Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates

Discussion

Member takes

Implication

Key Points

Cite This Study

Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider