“Exclusive: Embattled dean accused of plagiarism in NSF report” (yup, it’s the torment executioners)

Posted on by

The story is at Retraction Watch:

Erick Jones, the dean of engineering at the University of Nevada in Reno, appears to have engaged in extensive plagiarism in the final report he submitted to the National Science Foundation for a grant, Retraction Watch has learned.

The $28,238 grant partially supported a three-day workshop that Jones and his wife, Felicia Jefferson, held for 21 students in Washington, DC, in April 2022 titled “Broadening Participation in Engineering through Improved Financial Literacy.” Jefferson received a separate award for $21,757.

Jones submitted his final report to the agency in May 2023. Retraction Watch obtained a copy of that report through a public records request to Jones’s previous employer, the University of Texas at Arlington, and identified three published sources of extended passages he used without citation or quotation marks. . . .

Lots more details at the link.

Those torment executioners keep on tormenting us.

In all seriousness, between the University of Nevada salary and the National Science Foundation grants, this guy’s been taking a lot of public funds to produce some really bad work. Seems like a real failure of oversight at UNR and NSF to let this go on like this.

Good work by Retraction Watch to follow up on this story.

P.S. I forgot to include the quotations from UNR luminaries:

“In Erick Jones, our University has a dynamic leader who understands how to seize moments of opportunity in order to further an agenda of excellence,” University President Brian Sandoval said.

“What is exciting about having Jones as our new dean for the College of Engineering is how he clearly understands the current landscape for what it means to be a Carnegie R1 ‘Very High Research’ institution,” Provost Jeff Thompson said. “He very clearly understands how we can amplify every aspect of our College of Engineering, so that we can continue to build transcendent programs for engineering education and research.”

Also this:

Jones was on a three-year rotating detail at National Science Foundation where he was a Program Director in the Engineering Directorate for Engineering Research Centers Program.

Shameful that he would work for NSF and then pay that back by taking their money and submitting a plagiarized report. But, hey, I guess that’s what University President Brian Sandoval would call “understanding how to seize moments of opportunity in order to further an agenda of excellence.”

What could be more excellent than taking government funds and using it to publish plagiarized reports and crayon drawings?

It sounds like it’s fine with UNR if their dean of engineering does this. I wonder what would happen to any UNR students who did this sort of thing? I guess they wouldn’t get paid $372,127 for it, but maybe the university could at least give them a discount on their tuition?

P.P.S. That all said, let’s not forget that there are much worse cases of corruption out there. The UNR case just particularly bothers me, partly because it’s close to what I do—except that when my colleagues get NSF funds, we don’t use them to produce plagiarized reports—and partly because the problems are so obvious: as discussed in our earlier post, you can look at the papers this dean of engineering had published and see that they are incoherent and have no content, even before getting into the plagiarism. It’s hard to believe that his hiring was a mere lack of oversight; you’d have to work really hard to not see the problems in his publications. But, yeah, there’s lots of much worse stuff going on that we read about in the newspaper every day.

24 thoughts on ““Exclusive: Embattled dean accused of plagiarism in NSF report” (yup, it’s the torment executioners)”

  1. What I find truly amazing (from the examples in the Retraction Watch story) is that some of the passages were changed slightly by Jones. I think it is actually more work to make those slight changes than to use it verbatim and cite the source. It seems like they went out of their way to avoid citing the source to try to hide the plagiarism by modifying a word here or there. I think Andrew blogged about this recently – what motivates the avoidance of citing the source? Would Jones have not received the grant money if he had just appropriately cited? Does it really enhance his reputation (even if not caught) to pretend that the words were his? I just don’t get it.

    Reply ↓

  2. So is this guy going to keep his job? What about his wife and son?

    On the one hand, this is ridiculous, but on the other hand hardly anybody seems to be paying attention, Reno is a backwater, and those in power probably just want to ignore it and hope it goes away.

    Reply ↓

    • I almost paid the $53 to see the article – almost. From the description, it sounds like cognitive testing was done on 2 people that were readmitted to the hospital, one after an amputation and the other for cognitive reasons (not entirely clear from the abstract?). I’m fine with publishing case studies of interest. But to suggest any conclusions based on these 2 case studies is more than absurd. They do note that readmission due to pain from an amputation is different than cognitive impairment for other reasons – the basis for them suggesting that readmission might be avoided for the latter group by tailoring policies to them. But I’m curious how they tie their 2 case studies to such a conclusion. If anybody can find a free version of the paper (or send me the $53), I’d appreciate that.

      Reply ↓

        • Not for the first time.

          https://www.sciencedirect.com/science/article/pii/S0042698905005821

          Experiment 1 states, “ANOVA revealed a significant main effects of Trial Type [F_1,1 = 241.91] and Test [F_2,2 = 3.97] plus a Trial Type x Test interaction [F_1,2 = 8.67], suggesting that responding to the training and testing stimuli differed from test to test. ANOVA also found a significant Bird x Test interaction [F_1,2=8.55]”

          It’s not too surprising that the last three F values are not, in fact, significant.

        • Greg
          Can you expand your discussion a bit? I know absolutely nothing about this subject but it looks like the experiments take quite a bit of work to set up. Given that, analyzing 2 pigeons doesn’t seem too bad to me – but the “statistical analysis” seems unnecessary and misleading. As you say, it is n=2 and the ANOVA results seem silly. Why not just report what these 2 birds revealed in the experiments. Isn’t this really just an exploratory study and no analysis is really necessary (or appropriate)? I gather you are familiar with this type of research, so can you say a bit more about this particular paper?

        • Dale,

          I was really just responding to Shravan’s comment about a t-distribution with one degree of freedom. It reminded me of this paper. I don’t object to n=2 studies (I’ve run psychophysics experiments with n=1, using myself as the subject. I think that makes sense in some contexts.) I was being unkind when I added the comment “It’s not too surprising…”, but I have bad associations with this paper and the topic it studies. The bad association with the paper is that 3 out of the 4 experiments refer to “significant” F values that are not actually significant. This kind of repeated error makes me skeptical of the paper’s conclusions. (I also think NIMH, which funded the research, deserves better.)

          I have bad associations with the topic of the paper, object based attention, because sloppy reporting and bad designs are typical for this field. The topic itself is potentially interesting as it investigates how people (or in this case, pigeons) process visual objects; which is a fundamental aspect of visual perception. There’s a classic finding (which this paper builds on) that detection of a visual stimulus is faster when it appears on the same object as a previous stimulus rather than on a different object (Figure 1 in the paper describes their stimulus approach). My primary research involves modeling visual perception, and I thought my model could explain some object based attention findings, so I started looking into the literature.

          To my disappointment, the basic effect is tiny (a 10-15 milliseconds reaction time advantage). To my dismay, the studies reporting this effect tend to use sample sizes of 10-20 participants. A quick power calculation shows that this is off by an order of magnitude. Standard deviations for response times to identification are around 100 ms, and correlations between conditions are around .7 to .9. For a dependent t-test to produce a significant result with power of 0.8, one needs n=473 to n=159, depending on which correlation you think is appropriate. Even with the most favorable values, a study with n=20 participants has power of only 0.16. Even this low number is an upper limit because most studies were looking for additional outcomes in combination with the basic effect.

          Shockingly, _lots_ of studies with these properties only report significant effects, even though basic calculations say that their studies should generally produce non-significant results (just due to sampling error). There must be a massive “file-drawer” for these non-significant studies. A colleague and I analyzed all available papers and concluded that about half of the studies in this field seem “too good to be true”.

          https://link.springer.com/epdf/10.3758/s13414-022-02459-6?sharing_token=4E4f9dMrNTZOw9c-xf1x-ZAH0g46feNdnc402WrhzyqRqMR_Q9V5_udTos9sralkZHvMHEg-85qa_UtYbpTEmoGcc5d25jnw7xQ47w7FBtFtvW_8CbopQBT_if9jxClfzBBQ9lHvQpXEZmX7emelhgB2sM6aeDsvoZ_-qjVUh1I=

          So, I gave up trying to model this effect; I don’t trust the published results and I don’t want my model to be chasing noise.

        • Greg
          This makes me wonder whether there is any point to using hypothesis tests in this type of research at all. I’m not sure what to make of a p value with n=2, and if the power is so low with n=20, then it seems like the p value is really a distraction from what can be learned from the experiments. Then, I guess the question is whether doing experiments with so few subjects is worth doing at all. From the descriptions of the experiments, it should like it may be infeasible to use a sufficiently sized sample. But are exploratory analyses (such as this one, or others you refer to) still worthwhile?

          We’ve seen similar issues with some of the social psych research, where the noise renders the statistical analysis largely meaningless. While I’m not a fan of the social psych research, I am reluctant to dismiss entire fields of research that I have no expertise in. Similarly with this line of visual research – are you advocating dismissing all of it, eliminating the NHST approach, or something else?

        • Dale,

          The numbers might be different for pigeons, so maybe a small sample of pigeons could produce some useful data. I suspect it would not be worth the time/money/effort to gather such data from hundreds of pigeons. On the other hand, it would not be difficult to get a large sample of (typically undergraduate) human participants; researchers were just unaware that they needed to do so (because they never did a power calculation). I’ve heard from some researchers that they “gave up” on object-based attention studies because they found it to be an unreliable measure. That suggests that those researchers were actually running their experiments correctly because the measure _should be_ unreliable (in terms of statistical significance) with the sample sizes they were using.

          I think scientists in this field could do good work within the NHST framework; they just need to take it seriously and gather large enough samples to produce good answers to their questions. What they are currently doing is largely chasing noise (and this is evident in the existing literature, where different labs/papers draw opposite conclusions rather than converge on a common answer). Hardly any statistical analysis is going to work well when the signal is small relative to the noise.

          I actually put together a grant proposal to NSF to replicate many of these studies with appropriate sample sizes (I think the tasks and stimuli are clever and the experiments could address the questions of interest if they just had large enough samples). The feedback from the review panel was … not great. They suggested some revisions and encouraged a resubmission, but each time I did that the reviews got worse. I don’t know how to get award winning researchers to recognize that their careers have largely involved chasing noise.

          I think most of the field of visual perception is trustworthy. However, there seem to be specialized subfields that are pretty similar to social psych research.

        • Thanks for linking to your article Greg. As someone who works in a tangentially related area, this is super interesting stuff that I was not aware of.

        • I’m not sure what to make of a p value with n=2, and if the power is so low with n=20, then it seems like the p value is really a distraction from what can be learned from the experiments.

          What if n = 10 million? Then you are guaranteed to find a significant difference, since the null hypothesis used is known to be false to begin with. Instead they could figure out whatever their hypothesis actually entails, and try to find deviations from those predictions.

          Or how about use what they think is going on to perform an impressive feat. Eg, train some pigeons/crows/whatever to do something useful. Like find my golf ball and go squawk by it, then I give it food if the ball is found.

          You can vary the color, reflectivity, etc of the golf balls. Find out which is best for various terrain, weather, etc. Point is that any person could easily tell whether you have figured out something about birds via their performance at the task.

    • In all fairness, I’m not sure this particular article illustrates anything relevant to the misdeeds of Jones, et al. It is in a section of that journal called Clinical Comment/Case Study and is labeled as a Discussion. The conclusion seems to be a bit of a reach:

      “Conclusions: Not all readmissions are avoidable; however, if readmissions are related to cognitive impairment, implementing strategies tailored to this population may reduce readmission rates.”

      The one case study (presumably the unavoidable one) was someone readmitted after an amputation who did not do well on cognitive measures possibly “related to his pain level” while the other was readmitted due to cognitive impairment. I suppose the two case studies are illustrative of different circumstances which they then discuss, leading to their stated conclusions.

      I’m not sure this article really bothers me. And it is 6 pages long (not 2 paragraphs).

      Reply ↓

      • you got all that from the abstract? seems like a LOT of reading between the lines.

        if you read the whole thing (i didn’t), that might be a different story.

        Reply ↓

        • As I said, I don’t have access to the whole article and that might provide different information. But I think your skepticism here is unfortunate and dangerous. Once you are on a vendetta against someone, everything starts looking like bad conduct – the abstract provides some detail, but most of my comment came from looking at the table of contents for that issue of the journal. It is not unusual for case studies to be reported as N=1 or N=2 stories, and the article is clearly labeled as “discussion” not as a research article.

          I am appalled by the actions of Jones et al, but that doesn’t mean that everything that has been done by any of them (Jefferson in this case), and in conjunction with others (there are others listed as coauthors and you will find that they have done some serious research), is flawed and terrible. Your attitude comes close to qualifying as a methodological terrorist, actually more like a reputation terrorist. I think your characterization of this particular piece as “incompetence and corruption” is off base and dangerous. If you, in fact, have access to the full article and evidence of the incompetence that it generates, then please provide it so that we (I) can have an informed opinion. If you don’t, then I think it is wise to not label everything done with the names of any of these people attached as incompetent or corrupt.

Leave a Reply

Your email address will not be published. Required fields are marked *