Some background – for almost my entire career I’ve studied Dark-eyed Junco song, specifically the structure and function of the two types they sing, long-range (broadcast) song and short-range (soft) song. (2/n)pic.twitter.com/jRyEtsSgKU
U tweetove putem weba ili aplikacija drugih proizvođača možete dodati podatke o lokaciji, kao što su grad ili točna lokacija. Povijest lokacija tweetova uvijek možete izbrisati. Saznajte više
Some background – for almost my entire career I’ve studied Dark-eyed Junco song, specifically the structure and function of the two types they sing, long-range (broadcast) song and short-range (soft) song. (2/n)pic.twitter.com/jRyEtsSgKU
The short-range song is substantially more complex than long-range song and males sing it while courting females. Males respond most aggressively to short-range song playback, presumably because it’s interpreted as a rival male courting their mate! (3/n)
Male response was easy to measure, but I also wanted to know if females responded differently to short-range song, a prediction of the hypothesis that it was a courtship signal. We tried to test that question multiple ways, but everything failed. Science is hard! (4/n)pic.twitter.com/nK4NBdHqVx
In other species, female hormone levels change when they are stimulated by male courtship, so I teamed with two other graduate students to test whether female juncos that heard short-range song had different hormone levels than females that heard long-range song. (5/n)
I conducted the playbacks, and we collected blood samples after 45 minutes of song stimulus and multiple days of song. One collaborator measured estradiol levels in the blood, and the other measured corticosterone. (5/n)pic.twitter.com/Pp2TwI6HWQ
Each collaborator was blind to song treatment during the hormone assays, so it was up to me to pair the hormone measures from their spreadsheets to the appropriate treatment groups and analyze the data. (6/n)pic.twitter.com/KMqwfXC6vs
The estradiol data were a bust – most of the females had levels that were so low that they weren’t detected by our assay. Detectability is a common problem with estradiol EIAs, so we were disappointed, but not surprised. (7/n)
The corticosterone levels on the other hand showed something interesting. Females that heard male song of either type had LOWER corticosterone levels relative to controls! It was almost as though hearing song had a “soothing” effect (to anthropomorphize). (8/n)pic.twitter.com/1OK7mS4PyW
Although not what we hoped for, this was a COOL RESULT, and we started to think about where we could publish it. We weren’t aware of any other data that showed such a relationship, so it seemed to be novel. (9/n)
Then I noticed a HUGE error. I had accidentally copied the blinded corticosterone data from my collaborator’s file into a spreadsheet with the individual rows sorted a different way. The copying error had effectively randomized the hormone values across treatments. (10/n)
I re-ran the statistics with the correct data…no treatment differences. The significant effect was totally due to random chance. We were floored. That promising experiment became a two-sentence quip in my dissertation rather than an entire chapter. No publication. (11/n)pic.twitter.com/qspWBZO9T4
From the experience I learned to always triple check my data. When you get an exciting result, take a breath and make sure it’s real before you move ahead. Mistakes will happen, and our reward system tends to favor quantity over quality, which can be a recipe for disaster. (12/n)
Open Science policies are certainly a step in the right direction. The current situation has made that all the more evident. It also highlights the importance of replication and publishing negative results, two things that are critical but incredibly undervalued. (13/n)
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.