Today’s March for Science

Today’s national March for Science brought perhaps 200 protesters to rally and then parade through the downtown in my city, Walnut Creek in the East San Francisco Bay Area. I am not unsympathetic, but I got the impression that many of the speakers and marchers were insincere and naive.

These seemed to me to be grownup versions of students I had in the years when I was a Statistics professor — liberal arts and social science majors who resented having to take some STEM for the university’s General Education requirements. To suddenly defend the importance of science seems a bit disingenuous, a thinly-veiled excuse for attacking Pres. Trump. There is much on which the president might be criticized, but why not do it openly, instead of engaging in hypocrisy?

Worse, though, is their naivete, revealed in their treatment of science as hard and fast, unshakable, undeniable truth. A number of signs read, “Science Is Real Facts, Not Alternative Facts.” On the contrary, science currently is in a state that many consider a crisis, in which many “facts” are questionable. In the last couple of years, the confidence scientists themselves feel in their fields has been badly shaken, with a new term coming out — reproducibility.

The problem is simple: Much published research, even in the most prestigious journals, simply isn’t replicated when other researchers give it a try. When Scientist B tries to re-do the experiment conducted by Scientist A, it often turns out that B gets different results alternative facts, if you will.

The reasons for this unsettling state of affairs are complex. Yes, sometimes there is outright fraud, but more commonly the causes are more subtle: Hidden variations from one lab to the next; unconscious differences in assumptions from one researcher to the next; lack of understanding of statistical methodology; the tendency for authors and journals to report only “significant” findings; and so on. The old Keynes joke about economists, “You could lay all the economists of the world from end to end, and they wouldn’t reach a conclusion,” turns out to apply to science as well.

This is not news at all to the scientists themselves. It’s just that now people are openly talking about it, as the above link to Nature, one of the most prestigious journals in the world, shows so well. If you are the type who enjoys car crashes and such, plug “reproducibility crisis” into Google. You will be stunned to see how urgent the matter has become.

None of the above is meant to be a comment on the climate change controversy. I have not read enough of the research to have a point of view, a little obstacle that doesn’t stop many in the media and politics from speaking out. Nor is it a statement on Trump’s science policies. It’s too early for any fair person to have a strong opinion on the latter anyway. But for those marching today who took those college science classes for a Pass/Not Pass grade and put in minimal time studying, it’s not too late to learn some science and start using the Scientific Method, rather than blind hatred for the man half of Americans chose as our president.

Advertisements

21 thoughts on “Today’s March for Science

  1. Now this is something I am glad to hear somebody speaking out on.
    If only our media would apply the same analysis to economics, etc.

    The number one goal of every software developer should be that the code produces the same result, whether it be run one time, or 1 million times.

    Liked by 1 person

  2. Many thought leaders, from legislators to reporters, seem scientifically illiterate and innumerate – e.g., the current “reports” that a single diet drink multiplies the risk of stroke.

    Like

  3. Thank you for your measured response to the science is based on absolute facts controversy. Having been both a CS and Economics major, I have had a window into the hard facts and soft facts world of “science”. The issue that the average person is unaware of is the degree to which professionals have sold out to big money in the fact gathering business. On Wall Street they buy the research they need whether it’s on GM or global warming, and there are many willing to sell their souls for that extra cash. This is really the conversation that needs to be taking place, how is the research and researcher funded.

    Like

    • Fragomen, the largest immigration law firm in the U.S., is headed by founder Augustus Fragomen. He once said, of the AILA, “We commission studies to support our views.”

      Like

      • The priorities of American sci research have always been steered indirectly by the government. I won’t go into details, but academics know what I am saying here.

        ‘Disciplined Minds’ by Jeff Schmidt is book I highly recommend for anyone wanting to know how tenured faculty and sci research is steered to serve govt interests and opinion.

        With regards to the Trump administration, there are valid fears amongst the sci community that the govt will actively try to impede environmental and climate change related research. This concern is shared by the gender studies scholars because of the administration’s anti abortion leanings. Regardless of what is right and wrong, it is usually & morally expected that the govt not pressure and influence sci opinion.

        Trump administration has removed some open data related to EPA and Climate Change from WH websites. (Citation needed). Organizations like https://www.datarefuge.org are trying to archive this sci research before it is erased.

        Thus, the March for Science is completely justified in today’s America. About the involvement of social sciencey people and those of non sci bent , well you don’t need to know the theory of relativity to cheer for Einstein.

        Like

        • We are seeing one set of funding biases replaced by another.

          Some years ago, two sociologists made a research proposal on H-1B. From the reviewer comments, it was clear that this research was verboten.

          Like

  4. Hi guys,
    Interesting article reg. H1-b.
    http://timesofindia.indiatimes.com/business/india-business/us-accuses-top-indian-it-firms-of-violating-h-1b-visa-norms/articleshow/58327745.cms

    ———
    Quote from that article:
    “And those three companies are companies that have an average wage for H-1B visas between $60,000 and $65,000 (a year). By contrast, the median Silicon Valley software engineer’s wage is probably around $150,000”

    “So if the current system that awards visas randomly without regard for skill or wage is changed to a skills-based awarding, it would make it extremely difficult to use the visa to replace or undercut American workers”
    “Because you’re not bringing in workers at beneath the market wage. And so it’s a very elegant way of solving very systemic problems in the H-1B guest worker visa”
    ————

    Like you have said Norm, some more of the intels good infosys bad, but there is more focus on the pay level in those statements and from what I could guess, any move to raise the salary floor (notice the number on that officials mind $150k), that could I think stem abuse from both Intel’s and Infosyses.

    And we still have to see the fine details of what defines “highly skilled”, but the intention in that statement to avoid undercutting American workers is in the right direction.

    Like

  5. Great article, Norm. I’ve noticed the repro problem blossoming, too. I read an article a few years ago about “the hole in the scientific method.” Apparently, only the *desired* results get published. You can do a test 50 times, and 3 of those tests give a positive correlation, so you publish those 3 tests and discard the other 47. So there’s less than 10% chance the next lab is going to find matching results.

    In other words, *all* test results must be documented, otherwise you are discarding useful information — a “useless” result is just as important as a “useful” one.

    Like

  6. Science needs support from marchers like a fish needs a bicycle, to steal someone else’s anti-cliché.

    The “reproducibility crisis” is a crisis in publishing or academe, not science. Nobody doubts that one of the answers is right and the other is wrong, or there is some separating principle that would reconcile them.

    Few scientists are actually taught *about* science, that would be too “meta”, as we say in compsci. Although actually the reverse is true, in eighth grade science, and in high school science, the first thing students are taught is the method, before any specific result. But apparently a lot of people call in sick for those classes or took social promotion … so they could go out and march instead. Up Against The Wall Of Science, as Firesign Theater once said.

    And scientist or not, lots of people lie, if they can get away with it. That’s exactly one of the reasons that the scientific method includes the requirement for reproducibility. And too many in academe just love to hear themselves talk, whether they are any good at the science or not. There is of course the Sokal Affair. There is also Ludwig Wittgenstein, who condemned most of the practice of philosophy, as he saw it in his time, as word games. Too many peer-reviewed articles, much less any other kind, are inane gibberish, unreadable. You should see the ones rejected – and then again, sometimes good articles are rejected because after all, how good is the review process, really?

    When I was in school many, many moons ago the Great Panic was the upcoming mass starvation, the world could not possibly support FOUR BILLION PEOPLE, the riots and uprisings were about to begin, we would be out of oil, out of food, out of iron and copper and gold, by 1970. Or maybe 1975. Or maybe 1979. And now here we are in (checking my calendar) 2017, with 7.5 billion people, and our main problems are obesity and too much oil. Res ipsa loquitur.

    Like

  7. The whole thing is absurd. Science as most people understand it is hardly on the ropes these days; if anything it has garnered too much respect and could stand to be knocked down a few pegs. Real science is not used to promote political agendas; it is a pursuit of truth and knowledge. When people are allowed to define mutually contradictory truths and not be called on the simple fact that mutually exclusive propositions cannot both be true, that’s when real science is in grave danger; yet anyone who tried to start a rally in favor of the principle of non-contradiction would probably be stoned to death.

    Like

  8. Simple data principle…
    Raise the ‘noise floor’ to a sufficient level ( a ratio change greater than…) so that all data becomes buried in what use to be background noise.
    It takes a new interdisciplinary to science and its foundation in learning.
    Not only scientists and engineers but primarily now students must functionally edit what is science (logic) based data and what is humanist (emotional) data and of those what can be heard above the roaring noise of ‘junk’ data.
    When junk data is, for lack of a better example in data science, the new ‘0db’ , the data, methods and principles that drive change and innovation through analysis and application are going to have a hard time being heard.
    It may come down to a few individuals holding the candle close and keep it from guttering out completely in this new dark age to come.
    Thankfully, especially in America, boundless imagination and not agenda driven knowledge is the great engine of innovation and change.

    Like

  9. My feelings as well. Science only remains superior as long as it rejects disproven results. People with no scientific backgrounds whatsoever are pontificating on matters they have no knowledge of and stifling the ideas of those who disagree with them. I’m tired of statistically ignorant articles by authors who have no concept of the underlying assumptions and requirements of the statistical methods they use

    Sometimes I think economists think they’re entitled to use any statistical methods they wish without regard to whether the underlying assumptions are reasonable.

    Like

  10. > The problem is simple: Much published research, even in the most prestigious journals, simply isn’t replicated when other researchers give it a try. When Scientist B tries to re-do the experiment conducted by Scientist A, it often turns out that B gets different results — alternative facts, if you will.

    On this topic, there’s an interesting post on the difference between reproducibility and replication at http://jblevins.org/log/rep . What you are describing is replication where independent researchers go out and collect new data or re-do the same experiment on new samples. I can better understand how the new experiment may yield different results, especially if it involves complex manual steps. The researcher must be very careful to do the experiment exactly the same and understand the processes being studied well enough to control for any extraneous factors.

    Reproducibility, on the other hand, is independent researchers analyzing the same data. When the analysis is being done with computers, there’s a great deal that could be done to test reproducibility. From my experience, however, there is little pressure on the original researchers to follow these steps. As an example, http://econdataus.com/amjobs.htm describes the reproduction of a study that claims to find that, on average, each foreign-born STEM worker an with advanced degree from a US university creates 2.62 jobs for native workers. I was able to request the STATA program from the study’s author, convert it to R, and reproduce the numbers. That allowed me to analyze the calculations that produced the results and find a number of problems. The conversion to R was necessary because Stata is a relatively expensive package used by academics and did take some time and effort.

    Hence, if we really want to test our studies for reproducibility, we should require that the programs that produce those results be made available, preferably in a free language like R. Those programs should start with raw, publicly-available data and do all of the calculations to obtain the final results. That would allow other researchers to check the code to make sure it was written correctly and try alternate methods for processing the data. Unfortunately, there is little, if any, pressure on the original researchers to do this.

    Even once a study is done, there are often problems with the way in which its conclusions are interpreted and repeated. Two such examples are at http://econdataus.com/claim400k.htm and http://econdataus.com/claim2_4m.htm . It takes just a little fact-checking to find these problems but it seems that the great majority of people do not even check sources, much less do any fact-checking. It would seem to be a great achievement if artificial intelligence could eventually reach a point were it could crawl the web and do some of this simple fact-checking. Initially, it could just flag items for manual fact-checkers but eventually it could possibly even provide the explanations for its conclusions.

    Like

    • Hi Norm,
      From the background briefing, some interesting things I noticed:

      Senior WH Official:
      “we graduate about twice as many STEM students each year as find jobs in STEM fields.”

      “So we’re going to ask all four departments to tell us everything they think they can do. And, you know, it’ll be — some of it will be in the administrative bucket and some of it will be in the legislative bucket.

      And I think that there’s going to be a real serious legal review to find out what that is. I mean, preliminarily, we do think that we can make improvements to wages for H1B workers administratively. To what extent, we will see. And we do think that that we can also potentially increase some of the application fees as well, and that’s something that we’re going to be looking at too.”

      So, they are at the information gathering phase / reform possibilities gathering phase.

      I d say this would be the time for you to create a brilliant article that dissects H1-b and other related visas and propose possible solutions. Maybe put out a collection of all possible “good” reforms. The administration is looking for it and I d say this is a good time and as good as it gets.

      Like

      • I have written many articles and blog posts, no need for new ones. They are all on the Web, easily found.

        From both public and private information, it is clear to me that both the administration and the influential critics of H-1B are taking the Intels Good, Infosyses Bad view. My articles, which take the Intels Bad, Infosyses Bad view, are well-known to this crowd. Sorry, but this is not my day job. 🙂 All I can do is make my analyses available, and as I said, the “powers that be” are indeed aware of them; many of them read this blog and know me personally. They either do not find my analyses convincing or find it politically expedient to ignore them.

        Like

  11. What is lacking in most discussions and analyses is plain common sense. Someone need not be educated to have it. In fact, the more expensive an education an individual has been given, the less common sense that person seems to use.

    Like

  12. Great article, I concur. I starting reading one dense scientific article on the so-called global warming and my head almost exploded. The amount of CO2 coming from our oceans, wow. And then we had Climategate 1.0 and 2.0.

    I recall taking my science requirement and analyzing a book by The Club of Rome. Thirty acclaimed scholars said we would all be dead by now! No debate. But they forgot 2 important factors in their exponential growth argument: technology, and the role of price (market).

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s