I’m not looking for this email to be read on the show… but honestly, I sort of wish I was because your #890 was such a perfect start to my Easter Sunday. In comparison to the Moth podcast which Dr. D mentioned, I would easily select TWIV over Moth these days… actually TWIV would always be my first choice these days, except when I am looking to fall asleep… then I would never want to listen to TWIV, lest I nod off and miss a word. Another highlight from my Easter Sunday listen to #890 yesterday was Dr. Barker’s anecdote regarding her use of her virus crystal ball… so pleased you shared my appreciation, given the title you selected for the episode.
#890 was also nice because you included research out of my alma mater, Washington University in St. Louis… WUSTL is where I early on learned the issue that prompted this email: overlapping Confidence Intervals DO NOT necessarily mean non-significance. It is certainly true that CI for two samples that don’t overlap must be significantly different from each other. So one can appropriately say “They are obviously different because the CI do not overlap”. But one risks making a serious Type II error saying something like “The CI look like they overlap, so they are not significant.”
As was thankfully beat into my head at WUSTL (as I worked to analyse my first-ever PI data from a rct that has gone on to be cited 100s of times) the issue at hand is not the TWO samples’ variances… the real issue is the ONE variance within the sampled difference between the two means. A google search returned a number of hits regarding this confusion … for example, the link below… not the best explanation I saw, but I thought you might appreciate checking a PubMed hit first:
I think Dr Condit has made this mistake a few times. I hurt for you and the team when this happens, because there must be many other listeners beside me who were corrected on this assumption in the past! I remember early in the pandemic the team was corrected via email regarding the tails of BoxPlots (i.e., they are not related at all to CI inference). Way more important for me, I also remember you publicly being admonished for being rude to Dr. Despommier… your 100% apologetic response was so extremely admirable, you will always have my deepest respect for doing that within a podcast. So this CI thing… no comparison in my mind… just something to discuss with the team?
I hope you can appreciate why this matters to me, because you personally seem the finest model I know for young researchers … your banter as you carefully struggle to get the nuance within a pre-print… your thrill of discovery within dense Nature results… I doubt there are many better in this world than you at these sorts of things. And in my opinion, this world will be a better place if you keep doing what you are doing for decades, or at least until everyone forgets about Joe Rogan. 🙂
I recently watched TWiV 879, the episode where you discuss how Omicron preferentially utilizes receptor-mediated endocytosis as a route of entry over TMPRSS2-dependent membrane fusion. I found this very surprising because I assumed the opposite would be true. TMPRSS2-dependent membrane fusion is a faster route of entry for the virus and it permits cell-cell fusion, which provides the added benefit of entering a new cell without exposure to the extracellular space. Given that Omicron was the most transmissible variant seen yet, why would it evolve to use the slower mechanism of entry?
Integrated Biomedical Sciences (IBMS) Graduate Program
University of Notre Dame
Dear Twivers et al.,
Thank you as always for your continuing commitment to educating us mere mortals (and yourselves) on all things virological, immunological, epidemiological, epistemological, ontological, meteorological and whimsically illogical. The abstract below, which you have not hitherto discussed on TWiV, clearly indicates that TWiV saves lives.
All the best,
(J. Clinical Ambivalence, 2021, VI, 297-301)
Game-Changing Immunogenicity of Pandemic-Era Podcasts
A blinded, placebo-controlled trial
Bea Leevmié, Ph.D., Wi Che Karful, Ph.D., Anne Otayshian, Ph.D., et al.
The Covid-19 pandemic appears to have been paralleled by a misinformation pandemic on the subjects of the SARS-CoV-2 virus, viruses in general, vaccines, equine dewormers, and the technical feasibility of manufacturing microchips small enough to be injected using a normal clinical syringe.
Our study attempts to determine the efficacy of varying levels of podcast exposure upon prevalence of false SARS-CoV-2 information, with a secondary target of determining the effect of podcasts upon survival rates at a population level.
We randomly assigned a cohort of 3,252 internet users to one of two groups. The first (1,897 total, designated “Group 1” in accordance with our institution’s standard practice) listened to two, occasionally three, podcasts per week. The remaining 1,355 (“Group 2”), acting as the control arm, were not exposed to any podcasts, being limited only to broadcast news. Both groups were similarly diverse according to age, sex (if any), ethnicity, and grumpiness level on the Safir-Sesame-Oscar test. Researchers and subjects were blinded as to which subjects were assigned to which group.
In previous work on podcast information content, Balalaika, Viola, et. al. (J. Bov. Scat., xivii, 467-479) showed that TWiV was the most consistently accurate, if grumpy, source of virological data for general consumption . Accordingly, we used TWiV and occasional epitopes of TWiM as our test material for the study.
We evenly divided Group 1 into α (TWiV) and β (placebo) sub-groups, maintaining the same diversity as the overall study population. The α group were provided with two epitopes of TWiV weekly—plus, when available, one of TWiM—on a USB flash drive. The β group received an assortment of podcasts on popular topics such as business, barbershop harmony, or Beyoncé.
Neither researchers nor participants were told which podcasts they were hearing. Titles and credits were removed by an independent laboratory. The blinded recordings were assigned randomly coded filenames and supplied to researchers and participants on coded flash drives enclosed in plain, brown-paper wrappers. (Note: We decided after week 6 of the trial that it was not possible to blind Group 2 to the fact that the flash drives they received were, in fact, blank. Remarkably, however, in post-study interviews a significant number, 19.72%, reported not having noticed this.) The accounting firm of Dewey, Cheatham, and Howe secured all codes in a locked manilla envelope, which remained unopened until completion of the study.
All participants were followed for a period of 12 months, from 1 June 2020 to 31 May 2021. During the study period 93 participants dropped out and 36 died of non-COVID related causes, including 4 sky-diving incidents. Participants reported their health status weekly, including vaccination status after the US FDA issued Emergency Use Authorizations for Pfizer/BioNtech and Moderna vaccines in late 2020.
All participants completed pre- and post-study questionnaires to gauge their knowledge of viruses in general and SARS-CoV-2/Covid-19 specifically. Participants also reported their final vaccination status.
Of study participants who were exposed to TWiV, 961 (97.0%) were fully vaccinated by the end of the study. No COVID deaths were reported, 27 (2.7%) were admitted to hospital for COVID-19 treatment, including 2 ICU admissions, and 223 (22.5%) reported testing positive for SARS-CoV-2 in PCR tests at some point during the study. Questionnaire scores in this group indicated a 93% increase in accurate pandemic-related information and an absence of fake information by the end of the study period.
Results in the placebo group showed a significantly (r2=.97, Figure 3) lower vaccination rate and higher COVID morbidity than either of the other groups. 417 (46.0%) were hospitalized, of whom 163 (18.0%) were admitted to ICU and 82 (9.1%) died.
Among Group 2, exposed only to broadcast news, 2179 (67%) were vaccinated, 391 (12%) were hospitalized, with 122 (3.8%) ICU admissions and 66 (2.0%) deaths, roughly in line with national averages.
An important finding was that participants in the placebo and broadcast-only groups scored significantly lower (-28.7% in aggregate) on the post-study questionnaire than on the first (Figure 1). The β group achieved a remarkable and unexpected negative score on the second test (Figure 2). We are unsure how this is even possible, but we hypothesize that some placebo USB drives may have been accidentally contaminated with Joe Rogan podcasts. Further study is needed. Meanwhile, we are conducting a follow-up analysis of the scores in collaboration with the mathematics department.
CONCLUSION AND RECOMMENDATION
TWiV listeners were significantly less likely to experience severe Covid disease than either the placebo or control groups, and exhibited a 100% survival rate. In short, TWiV saves lives. Given the clear evidence of immunogenicity and lowered mortality imparted by TWiV, we recommend that the CDC immediately recommend broad TWiV listening for the general population.
A follow-up study is needed to determine the best age groups, but we suggest that, in the meantime, ages 14+ should be encouraged to listen at least twice weekly.
I’m listening to #890 and just heard Rich’s (I think) little rant about “encodes for.” When are you all going to address “data point?”
Seriously, though, thanks for all you do.
First: Vincent’s comments that science is politically neutral, is simply false.
Look at Peter Hotez’s accounts of why Stalin pushed Lysenkoism, or your wonderful #773 guest about who pays (the well-off) vs who benefits (all, but especially the less-well-off) from public health, and the political potential of subjugating people who won’t toe the party line are clear. (Yes, I believe it’s not just potential in 2022.)
Idealists looking to make a better world for all need to use a system built upon the notion of enlightened self-interest and increase the “enlightened” part
That’s your job; thank you for your excellence at it, so public interest might more often win out over the tendencies toward oligarchic power
Second, by visiting TheAtlantic from a browser that stores no cookies, I can confirm TheAtlantic story Amy liked was NOT paywalled for non-subscribers…at least, not for a first-time visitor. And while the circulation isn’t up with the AARP Bulletin, I didn’t see other mass-media publications that would consider a story of that heft.
The link is to a talk on the current state of time measurement with the improving accuracy of atomic clocks and its reliance on advances in ultra low temperature physics. The technology involved in reaching within a pico Kelvin of absolute zero – and therefore the limit set by the uncertainty principle – is impressive. This gives a clock accuracy of 1 second in the age of the universe.
The implications in many fields inc. gravitational wave detection and quantum computing are intriguing.
You may want to skip the first half-hour which is fairly basic.