Thread by @NicholasZaorsky

Publication productivity and academic rank in medicine.

Via @AcadMedJournal
ncbi.nlm.nih.gov/pubmed/32028299
From @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state

[tweetorial] on the impact of h-index and m-index on promotion/tenure.

#academicmedicine #MedEd #scholarship

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state Are metrics for promotion and tenure at academic institutions easy to understand?

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state Most would say the requirements are nebulous.

e.g., here are requirements from top tier institution, for non-tenure track and tenure-track faculty.

Historically, some said:
Asst prof = regional reputation
Assoc prof = national
Full prof = international

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state Sometimes, there are stories of young clinicians becoming a full professor.

(congrats, @montypal )

https://twitter.com/neerajaiims/status/1222412843040952321

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal On the other hand, there are more stories of promotion/tenure denial for unclear reasons.

nytimes.com/2020/01/02/us/…
Via @nytimes

Universities often lose gifted junior physicians and scientists.

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes In 2005, Jorge Hirsch suggested using "h-index", which quantifies publication quantity and quality.
@PNASNews
pnas.org/content/102/46…
How to calculate: line up one's publications from most cited to least cited. h-index = number of pubs with at least that many citations

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews H-index won't be falsely elevated by low impact publications, since they're never cited.

Also, if you're a middle author on high impact paper, h-index will still only rise by 1 from this paper.

Other caveats with h-index via @Publons : blog.publons.com/blog/5-things-…

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons H-index can be calculated with various programs/websites.
scholar.google.com easiest, IMO.

@AWHarzing

https://twitter.com/AWHarzing/status/1222094190881968128

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing The m-index is a newer term, it is meant to capture GROWTH of h-index over time.

Theoretically, those with highest potential for growth in h-index will have scientific success. Growth can be predicted.
nature.com/articles/48920…
@daniel_akuna @KordingLab

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab Distributions of h-index and m-index among ranks have been published for several fields.

e.g., data for #RadOnc from @DrEmmaHolliday @cd_fuller
via @ASTRO_org

ncbi.nlm.nih.gov/pubmed/27637135

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org There is association between h- and m-index; i.e., the ALL STARS of field (top right) have been continuous high-impact publication machines. The RISING STARS (above and left of trend line) may have relatively low h-index but high m-index, suggesting future success.

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org We performed a SR/MA of the literature for h-index and m-index across all academic ranks.

@EricLehrer did statistical wizardry: when the 95% CI wasn't reported, he digitized plots to back calculate from SDs or SEMs.

We included~17K physicians; ~14K in meta-analysis

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org Summary effect sizes for h-index among instructors:

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org Summary effect sizes for h-index among assistant professors.

Mean h-index 5.22 (95% CI: 4.21–6.23, n = 6,609)

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org Summary effect sizes for h-index among associate professors.

Mean h-index 11.22 (95% CI: 9.65–12.78, n = 3,508)

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org Summary effect sizes for h-index among full professors.

Mean h-index 20.77 (95% CI: 17.94–23.60, n = 3,626)

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org Summary effect sizes for h-index among chairs. (This is not mutually exclusive from full prof)

Mean h-index 22.08 (95% CI: 17.73–26.44, n = 816)

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org Mean m-indices were 0.53 (95% CI: 0.40–0.65, n = 1,653) for assistant professors, 0.72 (95% CI: 0.58–0.85, n = 883) for associate professors, 0.99 (95% CI: 0.75–1.22, n = 854) for full professors, and 1.16 (95% CI: 0.81–1.51, n = 195) for department chairs.

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org Here are the included medical specialties (based on @AAMCtoday) with h-index and m-index for each academic rank

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org @AAMCtoday There are some predictive powers of the h-index. For example, academic success of orthopedics departments correlates with h-index of chair and research director. @ORSsociety #orthoresearch
onlinelibrary.wiley.com/doi/full/10.10…

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org @AAMCtoday @ORSsociety On the other hand, even @NobelPrize laureates don't need high h-index

https://twitter.com/timgbenton/status/1047737125398089728

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org @AAMCtoday @ORSsociety @NobelPrize H-index, like any other measure, is a surrogate endpoint. Chasing higher h-index can lead to careerism

@VPrasadMDMPH

https://twitter.com/VPrasadMDMPH/status/1227360253651517441

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org @AAMCtoday @ORSsociety @NobelPrize @VPrasadMDMPH Creativity has no age, but productivity does.

Every time we pursue a research project, it is like buying another lottery ticket. Success can come at any time.
- @barabasi
ted.com/talks/albert_l…

@AcadMedJournal @EricLehrer @DrEmmaHolliday @PennStHershey @penn_state @montypal @nytimes @PNASNews @Publons @AWHarzing @daniel_akuna @KordingLab @cd_fuller @ASTRO_org @AAMCtoday @ORSsociety @NobelPrize @VPrasadMDMPH @barabasi Even Jorge Hirsch believes the h-index can fail spectacularly.
arxiv.org/ftp/arxiv/pape…
@biorxivpreprint

#MedEd #AcademicTwitter

Recently asked by @VPrasadMDMPH

https://twitter.com/VPrasadMDMPH/status/1292832685400846336

Here are the data from our analysis for number of publications and citations.