Why the A/AN prediction effect may be hard to replicate: a rebuttal to Delong, Urbach, and Kutas (2017)

Click to increase image sizeClick to decrease image size

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Aine Ito http://orcid.org/0000-0003-4408-8801

Notes

1. Even if we had failed to replicate a previously published noun effect, the null-hypothesis significance testing (NHST) and frequentist statistics dictate that failure to observe “real” effects (Type II error) and observation of “false” effects (Type I error; Gelman, 2015) are part and parcel of the scientific endeavour. Had we failed to observe a noun effect (which we did not), such a case would not diagnostically indicate there were technical problems with our experiment or data, but rather that, in NHST terms, we restricted our acceptable error to a probability less than 0.05. This uncertainty in observation or sampling error is at the heart of scientific inference and we adamantly maintain that Type I and II error and sampling error be considered in the conclusions we all draw from the effects we observe.

Additional information

Funding

Andrea E. Martin was supported by a Future Research Leaders grant from the Economic and Social Research Council of the UK [grant number ES/K009095/1].