Monday, September 23, 2013

ACS Salary Survey: unemployment down slightly, median salaries up slightly

The full article by Sophie Rovner about the 2013 ACS Salary Survey is out, and there are a variety of interesting things to comment on. The short version is that unemployment is down slightly, from 2012's 4.2% of respondents to 3.5% in 2013. Median salaries for all chemists rose between 2012 and 2013, going up 2.2%. 

Inflation: I think the most interesting graph of the article is below:

It basically shows that, since 2003, salaries for all chemists (academic, industrial, governmental, at all degree levels) have failed to keep pace with inflation (as measured in 2003 constant dollars). While you could argue that this is due to the broader economy, it's not a pretty story, especially combined with the relatively high rates of unemployment.

Unemployment: We've covered the basic unemployment numbers from the ACS Salary Survey already. Here's a problematic comparison between the US population of college graduates and ACS members that I see in the web version of the article:

I'm not convinced it's an apples-to-apples comparison when you're comparing a professional society of over 60% Ph.D.s to a college-educated labor-force with less than 10% Ph.D.s (it's probably more like 5%, actually.) 

U6-like lowest since 2008: I've long titled the sum of the percentage of unemployed chemists, part-time chemists and postdocs in the ACS Salary Survey as the "U6-like number." In other words, it's a broader measure of unemployment in chemistry ("U6" refers to BLS' broadest measurement of unemployment in the U.S.) That number for the 2013 Salary Survey was 8.9%, which is lower than it has been since 2008, when it was 7.5%. That's nice to see, even it is quite high. 

Weird geography: I don't think Andre the Chemist or I could have predicted this one: 
Unemployment rates among ACS members also varied across the U.S. The percentage of out-of-work members who were looking for a job as of March 2013 was lowest in the East North Central region,which consists of Illinois, Indiana, Ohio, Michigan, and Wisconsin, and highest in the Pacific region. New England, which at 6.9% had the highest unemployment rate last year, saw that number drop to 4.2% this year. The unemployment rate in the Pacific region eased from 6.5% to 4.9% during that same period.
The Upper Midwest with the lowest chemist unemployment? The Pacific region with the highest? Huh? 

The Eka-silicon caveat: Ol' E-s isn't going to like this one: 
The survey was sent to a random sample of 25,000 ACS members under the age of 70. The sample excluded student, emeritus, and retired members, as well as members living outside the U.S. The survey recipients returned 7,078 complete responses, for a response rate of 28%.
I'm pretty sure a 28% response rate is the lowest we've ever seen. 

The human side of chemist unemployment: David Harwell makes some sad points about long-term unemployment amongst chemists:
Still, Harwell cautions that the unemployment numbers might be artificially lowered by a small number of out-of-work chemists who have given up on searching for a job and thus are no longer counted in unemployment statistics. “We know that some people are dropping out,” he says. “We’re seeing people leaving chemistry.” 
The situation is toughest for what he terms the “very long-term unemployed,” who find it difficult to reenter the workforce for myriad reasons. “Their connections go cold; it’s hard for them to keep up their skills,” Harwell says. “They can go back to school, but that costs money, and if they have kids or a mortgage that is also eating through their savings, they just have to do something to survive.” 
And that doesn’t mean they are “making a targeted move from one sector of professional employment to another,” says Harwell. Instead, some of these chemists are now working at retail chains or in restaurants to make ends meet.
We've still got a ways to go. Best wishes to all of us.  


  1. So, the loss of chemistry wages pre-dates the Great Recession, which only seemed to accelerate the trend. I wonder if chemistry employments and wages will ever get back to what it was.

  2. "Median salaries for all chemists rose" also known as "gainfully employed were more likely to participate".

  3. Emil "Säure" BummerSeptember 23, 2013 at 4:10 PM

    our greatest ally in driving up chemistry salaries is chemophobia among the general public. rather than pushing kids into science classes we should make them believe that chemistry is something to be feared and loathed. we've got to thin the herd. there will always be chemistry jobs, but there will never be an unlimited supply of them

    As for the issue of imported students, the ACS should invest in propaganda to be distributed abroad that depicts the life of an immigrant chemist as miserable, hazardous

    1. "As for the issue of imported students, the ACS should invest in propaganda to be distributed abroad that depicts the life of an immigrant chemist as miserable, hazardous"

      Well, I guess they could pay a small premium to Chinese chemistry blog authors to make sure they post on a more regular basis.

  4. "rather than pushing kids into science classes we should make them believe that chemistry is something to be feared and loathed"

    This does more than thin the herd. This decreases the number of jobs available for chemists.

    You cannot market products with novel chemical actives directly to the general public. People want stuff with fewer chemicals or "all natural". Very few people understand that natural actives can have more problems than those designed to be safe for humans.

  5. yup, that would be an understatement!

    OK, so the good news, given the survey data the unemployment rate is 3.5 % (95% CI 3.1 - 4.0).

    Pretty reasonable. Now, for a sensitivity analysis of the nonresponse bias...

    If all 17,922 non-respondents are employed the UR is 3.7 % (95% CI 3.5 - 4.0).
    If all 17,922 non-respondents are UNemployed the UR is 72.7 % (95% CI 72.1 - 73.2).

    This kind of thing is only important if you suspect that non-response might be influenced by the outcome of interest. Good thing we don't have to worry about that. ;)

    There's really no issue at all in using the 248 unemployed chemists who are still ACS members and who responded, to draw inferences about all the xx thousands of unemployed chemists. Perfectly reasonable.

  6. One other point, the UR from 2013 is 3.5 % (95% CI 3.1 - 4.0) and from the 2012 survey it's 4.2% (95% CI 3.7 to 4.7) assuming 7000 respondents

    (not clear here

    As the 95% CIs overlap, this means there is NO evidence to suggest there has been any improvement from 2012 to 2013, as much as we all wish it were so.

    Why can't the world's largest scientific organization do the most basic of statistics??!!

    ACS, are you listening? It's embarrassing!

    1. I have to say, thank you for putting some error bars on this analysis. I personally suck at statistical analysis, so I'm glad someone can elucidate.

    2. They would most certainly use standard error (CI = 67%) to prove any point when possible. Mendellin the biologist would see to that.

    3. You are being harsh and unfair. The quality of the pictures clearly indicates that we are using the most modern and advanced statistical tools.

    4. I guess that those error bars must be so super tiny to be covered by the point. My most humble apologies.

  7. "As the 95% CIs overlap, this means there is NO evidence to suggest there has been any improvement from 2012 to 2013, as much as we all wish it were so."

    Only if you confine yourself to the world of classical statistics, and ignore Bayesian thinking. Given outside data for the broader economy indicating a real but weak recovery, you would expect to, as a baseline, see a similar weak recovery for chemists. That is exactly what we see. I've always thought it odd that people are so insistent on setting their null as "no change" rather than "change according to prior observed trend" or "change of the subset equal to change of the observed change for the more robust full set".

    I have seen no data as to what the employment rate is for pool cleaners, or how it changed last year. However, my null hypothesis would not be that it didn't change, but that it changed in the same proportion as overall unemployment did for the general population.

  8. Yes, you are totally correct Chad. Perhaps you can suggest some informative priors for post-meltdown chemical employment??

    Pretty sure the non-response rate and differential non-response by employment status problems dwarf any classical/Bayesian issues... and moreover, the main point is that the entire survey and 'analysis' by the ACS meets none of the most basic standards for survey work or responsible reporting of data. Excel doesn't cut it!

  9. Whatever statistical analysis you use, these salary surveys never seem to pass the smell test. $72K median salaries for BS chemists? How come all the BS level chemistry jobs seem to be in the $10-$15 /hr range? The ACS survey sounds about as reliable as getting height and weight information from dating websites. I suppose you have to be somewhat delusional to keep paying ACS dues. The median ACS member is 7 feet tall, drives a Bentley, and lives in Malibu… renew today!

    Students: If you want to know what your degree will be worth, look here and weep.

    1. The ACS salary survey is a poll of ACS members, who are not average chemists and who seem to have higher salaries than most. For example, new PhDs normally enter as something equivalent to's "Chemist IV" at the companies I have worked at, starting at the bottom/middle of this distribution and working their way up.

      That's a full $10k less than what ACS is reporting as the typical salary for PhDs with 5-14 years since their BS, which is pretty much the same group of people. I think you are better off looking at's distributions for Chemist IV (younger PhD) and Chemist V (after your first major promotion, small group leader) than the ACS survey.


looks like Blogger doesn't work with anonymous comments from Chrome browsers at the moment - works in Microsoft Edge, or from Chrome with a Blogger account - sorry! CJ 3/21/20