Open FlorianSeitz opened 4 years ago
Hier noch einige Projekte mit Daten aber bayesianischer Schätzung:
https://osf.io/5semf/ (Pachur & Schulte-Mecklenbeck) https://osf.io/ngc45/ (Kellen, Steiner, Davis-Stober, & Pappas); preprint: https://psyarxiv.com/qvcbk/
Maybe this could be something?
https://osf.io/6euqj/ (Glöckner & Pachur; paper) In my opinion, they used softmax/Luce's choice rule and didn't use Bayesian estimation. However, as far as I can see, in the paper they only report median estimates across participants (?)
@JanaJarecki : Is there any possibility in the cpt to say that alpha and beta need to have the same value (this is what Glöckner and Pachur did in their study, so only one common exponent for gains and losses)?
Absolutely there is, hopefully intuitive: fix = list(alpha = "beta")
Maybe you need to update to the latest version before (pull from repository and re-compile the package)
How are things going?
@JanaJarecki I'm facing (again) a problem with loading the cogscimodels package. It has something to do with the utils-checks.R file. When I execute devtools::load_all(), R gives me an error saying: Error in parse(text = lines, n = -1, srcfile = srcfile) : C:/Users/Sylvia/Documents/cogscimodels/R/utils-checks.R:22:1: unexpected input 21: 22: <<
It seems that R cannot read the "<<<<<<< HEAD" in the utils-checks file. Do you have any idea what could be the problem?
entschuldige mein Fehler, ich habe gerade einen Bugfix gepuscht, sollte jetzt funktionieren.
Die >>>> sind Zeichen eines nicht sauber gelösten Merge-Conflicts.
@JanaJarecki First results are here:
Repetition 1 Paper: c(alpha = 0.74, beta=0.74, gammap = 0.61, gamman = 0.89, lambda = 1.27, tau = 16.66) R_res: c(alpha = 0.77, beta=0.77, gammap = 0.58, gamman = 0.88, lambda = 1.23, tau = 17.08)
Repetition 2 Paper: c(alpha = 0.76, beta=0.76, gammap = 0.58, gamman = 0.89, lambda = 1.19, tau = 16.66) R_res: c(alpha = 0.77, beta=0.77, gammap = 0.56, gamman = 0.90, lambda = 1.18, tau = 13.33)
Die Resultate sind sicher sehr nahe am Original, die Abweichungen könnten von den Unterschieden im Fitten stammen, was meinst du?
Hey Jana, we could use the data from Rieskamp (2008) Study 2. It is a bit smaller than the other data set we have (ca. 30 subjects, 180 trials per subject). However, only aggregate parameter estimates across subjects are reported (p. 1455), so I guess the data set is still too large then. In any case, here's the link to the paper and the osf repo. https://www.neuronetwork.unibas.ch/brainweek09/documents/2008_Rieskamp_JEPLMC.pdf https://osf.io/eypgb/ (repo is from a follow-up paper, but it includes the original data of Rieskamp, 2008)
I have not yet found a paper that reports actual individual parameter estimates unfortunately.
Hi @JanaJarecki
Ich habe bei osf mittels "prospect theory" NOT bayes nach Daten gesucht, aber nichts gefunden. Kann aber auch sein, dass ich irgendwie komisch gesucht habe, weil mir trotz des NOT bayes Resultate mit bayesianischer Parameterschätzung angezeigt wurden.
Wir hatten letzte Woche kurz über die osf-Suche gesprochen. Könntest du mir nochmals kurz schreiben, wie du auf osf nach Daten suchst? Einfach, damit ich sicher nichts verpasse.
Liebe Grüsse Florian