Personally, it was a highly anticipated symposium at CfA because I was fascinated about the female computers’ (or astronomers’) contributions that occurred here about a century ago even though at that time women were not considered as scientists but mere assistants for tedious jobs.
I learned more history particularly about Ms. Henrietta Leavitt who first speculated the period-luminosity relation from Cepheid stars. Her work is a real painstaking task that cannot be compared to finding a needle in a haystack. It’s like finding some needles from a same manufacturer from countless haystacks, which may or may not have a needle from the specific manufacturer. The worst part is, needles are needles. Not many needles have tags like your clothing for an identification.
However, I was disappointed because of two reasons. First is a minor disappointment but very valuable. The author (George Johnson) of the book, Miss Leavitt’s star – I haven’t read, actually I didn’t know it exists until today – answered my question that he does not think Ms Leavitt’s was exposed to statistical research. Finding a relationship between period and luminosity is closely related a simple regression analysis and I thought she knew about statistics to associate her discovery to now so called, the Leavitt’s law. This disappointment actually lead me to question when the statistical analysis kicked in in astronomy, particularly finding relationships in any studies related to the standard candle, to find out the correct estimate of the Hubble constant.
The second reason of my disappointment is very poorly executed statistics. Obviously, it’s not Ms. Leavitt who imposed such strange trend and statistical malpractices (or carelessness) in regression analysis among astronomers. Whenever speakers bring out scatter plots with regression lines and data points with error bars, I keep murmuring silently, “Oh, my God, how come they blindly did that?” There were statistical issues to be addressed prior to stating that their results support a certain hypothesis instead of putting a straight line and claiming that – “see, how good the slope is” – the Hubble constant is # plus minus $. A high leverage point on the right in addition to less than a dozen points clumped in the left corner, without various diagnostics tools in regression analysis, one does not claim that the straight line is a good fit nor can say that the analysis backs up the hypothesis. Perhaps these statistical diagnoses only advocate their concluding estimates and their uncertainty, and so are omitted. However, my feeling upon looking plots tells me that a simple bootstrap could prove that their estimates are not accurate as they think. Until you try, you don’t know, though. I may email those speakers politely if I can have data points they used for their scatter plots. Unfortunately, I know no one is willing to give me their data points for such unjust cause since even good causes, I had experiences of indifference (I myself might do the same if I were in their positions, no complaints!!!).
Regardless of these disappointments from the statistical instinct, it was a scientifically very interesting symposium and like to thank who made great efforts to put things together. It helped me to resolve some of my crave to know about Ms. Leavitt and to satisfy one of my old wishes that her work to be recognized under her name. If there is one, I wish I could have attended the symposium to commemorate the centennial of student-t, this year. It’s always good to know the history better to move forward.
Asides, during G. Johnson’s talk, he showed pictures of apt. building, which I see everyday, that Ms. Leavitt made her residence until her death and of Mr. Auburn cemetery, very beautiful calming place, where she was buried. I wished she had lived longer to see a glimpse of her great contribution to astronomical science.