tag:blogger.com,1999:blog-7890764972166411105.post7636525989319400670..comments2024-03-29T06:02:41.835+01:00Comments on Nick Brown's blog: Just another week in real-world science: de Venter et al. (2017)Nick Brownhttp://www.blogger.com/profile/00172030184497186082noreply@blogger.comBlogger8125tag:blogger.com,1999:blog-7890764972166411105.post-47019253792191374732018-10-28T11:52:10.622+01:002018-10-28T11:52:10.622+01:00The motives of the authors reminded me to the moti...The motives of the authors reminded me to the motives of Omar Al-Sheikhly in an e-mail of 15 June 2015 in which he refused to give us access to a specific set of raw research data of his 2013 publication in the journal 'Zoology in the Middle East' ('this would formally be a one-sided review' and 'the EiC already acted as independent and impartial referee').<br /> <br />Max Kasparek, the EiC in question, was subsequently contacted on 21 June 2015 for a formal request to get access to the raw research data. There was until today, 28 October 2018, no response from Max Kasparek.<br /><br />Taylor & Francis, the publisher of the journal, was contacted at the same moment. A response from TF was received on 16 June 2016. 'Like the majority of scientific journals, this one does not compel the author to provide the raw data of the research to anyone. We will not be responding to your request to provide you with this.'<br /><br />This response was received shortly after I had started to send daily reminders in which I was also referring to remarks in 'out-of-office auto-replies' which I had received from TF.<br /><br />TF also told me on 16 June 2016 in this e-mail: 'no-one in this organisation will respond further to your emails'. <br /><br />Backgrounds at https://osf.io/j69ue/Klaas van Dijkhttps://www.blogger.com/profile/05507228199903986218noreply@blogger.comtag:blogger.com,1999:blog-7890764972166411105.post-81519298138336627342018-10-25T14:22:01.865+02:002018-10-25T14:22:01.865+02:00I'm not sure what the ethical rules are about ...<i>I'm not sure what the ethical rules are about publishing e-mail conversations</i><br /><br />I feel that any e-mail conversation on substantive topics can and should be publicly available unless the e-mailer requests and obtains an agreement to keep the correspondence confidential.<br /><br />Quite simply, if you send me a letter, I can publish it.jrkrideauhttps://www.blogger.com/profile/04869979887929067657noreply@blogger.comtag:blogger.com,1999:blog-7890764972166411105.post-21282546181176917682018-10-24T05:01:03.665+02:002018-10-24T05:01:03.665+02:00I note that they say of the total abuse score '...I note that they say of the total abuse score 'we did not incorporate the TEC Total Score in the analysis because this summated score is not independent from the subscale scores.'<br />One wonders if it the correlation with total score would also be the thing that most people would look at to see if the effect was real, but it was non-significant, so left out.SpeedEvilhttps://www.blogger.com/profile/07752655707914548437noreply@blogger.comtag:blogger.com,1999:blog-7890764972166411105.post-1552253893582678262018-10-19T15:32:12.289+02:002018-10-19T15:32:12.289+02:00Thanks. I hadn't thought of that approach.Thanks. I hadn't thought of that approach.Nick Brownhttps://www.blogger.com/profile/07481236547943428014noreply@blogger.comtag:blogger.com,1999:blog-7890764972166411105.post-25209679167306499562018-10-19T10:06:21.718+02:002018-10-19T10:06:21.718+02:00That's true. But with the limited amount of in...That's true. But with the limited amount of information provided by the authors in this case, we can't tell if it was appropriate or not to include those correlated predictors. It also seems to me that using (only) a standardized coefficient with a beta above 1 and a p value of .03 as evidence for your effect suggests that you don't have anything better to offer, such as a substantial (delta-)R^2.<br /><br />Of course, the authors could have cleared most of this up by providing their data, or even just the *table of correlations*.Nick Brownhttps://www.blogger.com/profile/07481236547943428014noreply@blogger.comtag:blogger.com,1999:blog-7890764972166411105.post-22124929167636273812018-10-19T09:56:00.256+02:002018-10-19T09:56:00.256+02:00Copy/pasted from https://www.uantwerpen.be/images...Copy/pasted from https://www.uantwerpen.be/images/uantwerpen/container2123/files/Nationale%20code%20Belspo_en.pdf (page 8):<br /><br />"Researchers’ work is deemed to be verifiable when it allows colleagues to follow the progress of the research and to reproduce it, if need be."<br />"The information given should be verifiable. The results of the literature study, the hypotheses, the organisation of the research, the research and analysis methods, as well as the sources, are described in detail (in a research logbook, a laboratory diary or a progress report) so that other researchers can verify the accuracy of the process and reproduce it."<br />"When publications, especially review and summary articles, do not contain all the necessary data for verification, the data should nevertheless be available."<br /><br />It seems to me that there are grounds to argue that the apparent refusal to share these data are not in line with the above listed quotes from the “Code of Ethics for Scientific Research in Belgium” and it thus seems to me that there are grounds to argue to report this behaviour to Willem Lemmens, backgrounds at <br />https://www.uantwerpen.be/en/research-and-innovation/management/quality-assurance/scientific-integrity/ Klaas van Dijkhttps://www.blogger.com/profile/05507228199903986218noreply@blogger.comtag:blogger.com,1999:blog-7890764972166411105.post-54079627537304321302018-10-19T08:15:46.881+02:002018-10-19T08:15:46.881+02:00Ha Nick
Applying a correction for multiple testing...Ha Nick<br />Applying a correction for multiple testing and no effect is statistically significznt from zero. Also if you do not like NHST, no evidence here that justifies a strong conclusion.<br /><br />But Nick, you also provide strong statements and interpretations on high standardized beta coefficients. Yes, a standardized beta > 1 implies a correlation between predictors, but this may entail true correlations and not correspond to an artefact.Marcel van Assenhttps://www.blogger.com/profile/08459483740874064850noreply@blogger.comtag:blogger.com,1999:blog-7890764972166411105.post-83920001912320167932018-10-18T23:47:19.342+02:002018-10-18T23:47:19.342+02:00I don't think that you are going to get the da...I don't think that you are going to get the data.<br /><br />In fact, Dr. Van Den Eede sounded a bit defensive.<br /><br />I have never seen any reason why one would not share suitable sanitized data unless there is a clear contractual agreement with a data supplier not to do so.jrkrideauhttps://www.blogger.com/profile/04869979887929067657noreply@blogger.com