John Tukey said that the best thing about being a statistician is that you get to play in everyone’s backyard. It’s a great quote. While it more obviously reflects the tendency of applied statisticians (defined broadly) to be involved in lots of studies, it also captures something about statisticians that is often missed - that we are also scientists, delighted in the pursuit of new knowledge.
Love, love, love. But how does this information reach the eyes and ears of health professionals who bring junk research to the coalface of practice? If only.
When I began graduate school, few statisticians could get federal funding as a statisticians. Cattell slipped in as a personality psychologist constructing an inventory with fancy exploratory principal component analyses (PCA). John Tukey, author of the very readable Exploratory Data Analysis was the brilliant hero of a lot of us who struggled to find the best statistical solution possible for our psychological research. We were rather clueless. We mostly consulted statisticians when we had already collected and analyzed data and hoped they would say we had not done anything wrong.
In the early 2000s, NIMH methods grants were created. Tom Ten Have, Patrick Shrout, and Niall Bolger got some of the first highly competitive awards, all talented guys whom I had the honor of working with.
Wonderful piece Darren. One of the hardest things for us statisticians to do is to convince a researcher NOT to start a project. We don't win any immediate favors in doing this, but the effort is needed. It helps greatly if you offer as we do a daily walk-in (really zoom-in) clinic where researchers talk to us in the very earliest stages of their research idea formulation. We have in my view done a lot of good at these early stages, sometimes getting the researchers to find better projects.
Thanks Frank. I find these conversations very tricky. They are asked to do research to progress their careers, but aren't provided the support to do it correctly. And if if they are lucky enough to have a statistician/methodologist around, it's likely some grump like me suggesting that what they want to do is likely pointless! :P
Researchers' behaviour is a downstream problem. The solution is to target Universities, funders and things like ethics boards. The lack of statisticians/methodologists isn't down to luck, it's the result of deliberate choices of where to invest and who to collaborate with. Start getting those with gate keeping power to place more emphasis on statistical competence, then more posts will created be created for methodologists and their will be less positions for those principally motivated by career prestige.
You are so right. For a subset of cases I suggest that they link up with researchers at other institutions to form a collaboration network for sharing data so that at least the "futile sample size problem" is overcome.
Love, love, love. But how does this information reach the eyes and ears of health professionals who bring junk research to the coalface of practice? If only.
Excellent post.
"try telling that to a university press office. Tell that to a scientist talking into a journalist's microphone about their most recent publication".
That hits the nail on the head.
You always know how to cheer me up.
When I began graduate school, few statisticians could get federal funding as a statisticians. Cattell slipped in as a personality psychologist constructing an inventory with fancy exploratory principal component analyses (PCA). John Tukey, author of the very readable Exploratory Data Analysis was the brilliant hero of a lot of us who struggled to find the best statistical solution possible for our psychological research. We were rather clueless. We mostly consulted statisticians when we had already collected and analyzed data and hoped they would say we had not done anything wrong.
In the early 2000s, NIMH methods grants were created. Tom Ten Have, Patrick Shrout, and Niall Bolger got some of the first highly competitive awards, all talented guys whom I had the honor of working with.
Wonderful piece Darren. One of the hardest things for us statisticians to do is to convince a researcher NOT to start a project. We don't win any immediate favors in doing this, but the effort is needed. It helps greatly if you offer as we do a daily walk-in (really zoom-in) clinic where researchers talk to us in the very earliest stages of their research idea formulation. We have in my view done a lot of good at these early stages, sometimes getting the researchers to find better projects.
Thanks Frank. I find these conversations very tricky. They are asked to do research to progress their careers, but aren't provided the support to do it correctly. And if if they are lucky enough to have a statistician/methodologist around, it's likely some grump like me suggesting that what they want to do is likely pointless! :P
Researchers' behaviour is a downstream problem. The solution is to target Universities, funders and things like ethics boards. The lack of statisticians/methodologists isn't down to luck, it's the result of deliberate choices of where to invest and who to collaborate with. Start getting those with gate keeping power to place more emphasis on statistical competence, then more posts will created be created for methodologists and their will be less positions for those principally motivated by career prestige.
You are so right. For a subset of cases I suggest that they link up with researchers at other institutions to form a collaboration network for sharing data so that at least the "futile sample size problem" is overcome.