Selective Reporting Thread: Only Post Selective Reporting (File Drawer Effect) Comments Here

This thread was created because selective reporting has come up too often as a topic and as there are strong feelings on both sides, this generally degrades threads where it is discussed. So this topic is now banished from the rest of the forum.

In no particular order, people are invited to post their arguments about selective reporting in parapsychology here, and only here.

Skeptics are invited to post here.
From a theoretical perspective, the risk of bias from selective reporting in the parapsychology literature is high, as there are not measures in place to reduce the risk of this bias. And we see that this high risk of bias is expressed by finding numerous examples where selective reporting is present. This is a list of some specific examples of different types of selective reporting. Please note that this is merely a tiny, illustrative sample, not a complete listing.

Selective reporting refers to between-study selection (often referred to as publication bias) and within-study selection (which includes selective outcome reporting and selective reporting of sub-group analyses).

"Publication bias" doesn't just involve a failure to publish a study. It also refers to a failure to include relevant studies and outcomes in meta-analyses, among other things.
A survey of student projects, including ganzfeld experiments, some of which were published and some not. The results in the unpublished studies are different from those in the published studies.
Ganzfeld meta-analysis which is missing studies and whose results are different from the included studies. (comments section)
Failure to report the main outcome (difference in mood between intentional tea and a control group - reported difference is on a small subset only) of the study in the report. Subsequent questioning of the author revealed that this outcome was not significant.

"Robertson, T. J. and Roy, A. E. (2004) Results of the application of the Robertson-Roy Protocol to a series of experiments with mediums and participants. JSPR 68.1"
Failure to report the main outcome (reading accuracy between recipients and non-recipients of mediumship readings under blind conditions). The authors have not provided these results subsequently, and my requests for this information from the authors have been unanswered. (under "Studies 101-105: Novice studies")
The reported results on Julliard students from Honorton's autoganzfeld studies was a post-hoc selection from two experiments rather than the original, full experimental results. The results of the original planned experiments are not reported.
Multiple measured outcomes in mediumship research, only some of which are reported. Of the 21 different outcomes measured in the 3 studies, all three were reported in the Robertson-Roy study (but only for 1 of the 8 different experiments which were carried out (experiment 5)), 3 out of 12 were reported in the Beischel study, and 1 out of 3 outcomes for study 1 and parts of 2 out of 4 outcomes for study 2. To be fair, Beischel states that their item-by-item scoring will be published later. So we are missing the results from 7 of the 8 experiments in Robertson-Roy (including Experiment 1 which was the only one which followed the Robertson-Roy blinded protocol), and 4 of the 7 outcomes from Kelly.