I propose that any researcher or scientist who unwittingly gets into a quagmire with the Geier’s should be referred to as being ‘quag-geier-ed’. Its a handy way of referring to people who’ve (possibly accidently) stumbled into a great big pile of shit.
associations between neurodevelopmental disorders (NDs) and exposure to mercury (Hg) from Thimerosal-containing vaccines (TCVs) by examining the automated Vaccine Safety Datalink (VSD)
One would imagine that any serious researcher who valued her career would be reluctant to associate with the Geier’s who have odd ideas about what is valid research but maybe Professor Young simply didn’t know.
Anyway, I forwarded the paper itself onto Epi Wonk, a blogger who:
…has a Ph.D. in epidemiology from an Ivy League university. Before that I got a bachelor’s degree from a different Ivy League college, a master’s degree in developmental psychology, and a master’s degree in medical sociology from another Ivy League University. I worked for more than 30 years as an epidemiology professor in medical academia and schools of public health, and in the senior biomedical research service at the Centers for Disease for Disease Control and Prevention (CDC). During my career I have been the editor of two epidemiology journals and one more general biomedical journal. I am now retired.
So very, very bad was the quality of this paper that Epi Wonk took three (and possibly a couple more in the future) posts to tackle the numerous issues with it. I plan to recap them here but here is Epi’s take on the paper itself.
So in part I, Epi found the following:
dubious “imputing” or imputation lies at the bottom of the author’s little trick…..
Imputastion is simply – using known data to ‘guess’ at unknown data. Epi gives an example:
…let’s say a researcher has a file of data on children and 8% are missing data values on parent’s household income, 4% are missing data values on gestational age at birth, and 1% are missing data values on birth weight. She decides to use an imputation procedure to impute values for parental income, gestational age, and birthweight where they were missing. Perfectly fine, legitimate, and scientifically valid under most circumstances.
However, when we are dealing with something like autism….
She examines the data and sees that in certain cohorts in her study population the distribution of autism isn’t quite what she would like. So she “imputes” autism cases into the data set. Except that she’s not imputing a value on a variable for an existing study participant. She’s adding imaginary autism cases into the analysis. This isn’t imputation — it’s cooking the data.
Epi was very disturbed about this to the point that xe said:
This is just not done. It’s not valid. It’s not ethical. Adding imaginary cases into a data set borders on scientiific fraud.
Later on in the comment thread that developed, commenter Andrea asked:
Does this mean that Young, Geier and Geier added 45 and 80 cases that were not in the original data sets, that they MADE UP those 125 cases just to add imaginary data points to make the stats results look more like what they wanted?!
Thats exactly what happened.
In Part II of xes detailed look at this paper, Epi concluded that when it come to controls and particularly controlling for any confounding variables:
….there’s no attempt to control for, or adjust for, the confounding effect of birth cohort. Just one look at Figure 1 (or a basic knowledge about trends in autism) tells you the regression coefficients (slopes) are being driven by increases in autism risk over time. Given the increase in frequency of autism (and other neurodevelopmental disabilities) during time time period, you could do an ecological regression analysis of almost any factor that varied over time and you would find an an association with autism. I would bet that you could enter number of sushi bars per capita into an ecological regression and you’d find an association with autism rates.
This is not a good or valid paper. It seems, based on the expert analysis of a professor of epidemiology that this paper is fundementally flawed.