040639 UK Exact Tests not only for Experimental Economics (MA) (2013W)
- Anmeldung von Fr 06.09.2013 09:00 bis Fr 20.09.2013 14:00
- Anmeldung von Mi 25.09.2013 09:00 bis Do 26.09.2013 17:00
- Abmeldung bis Mo 14.10.2013 23:59
Termine (iCal) - nächster Termin ist mit N markiert
Ziele, Inhalte und Methode der Lehrveranstaltung
Art der Leistungskontrolle und erlaubte Hilfsmittel
data sets, and analyzing data sets. Each of these three parts will be separately graded and
counts equally towards the final grade.
Prerequisites: knowledge of statistics at an undergraduate level.
Mindestanforderungen und Beurteilungsmaßstab
testing hypotheses and running regressions that are exact. One goal of this course is to teach
students how to use R in order to analyze data sets. Laptops will be used in class to
demonstrate methods. Students will learn how to analyze data sets and how to read and
understand empirical papers.Who is this course for? Anyone who is curious and
who is genuinely interested in uncovering what is hidden in the data and who is interested
in making mathematically sound claims. Of course many applications cannot be dealt (yet)
with an exact method as often there is too much going on. However this course will
demonstrate that there are lots of relevant areas where one can make exact statements,
including running linear regressions.
most statistical methods typically make claims about data sets that are not in accordance
with the underlying theory and methodology. This is because they make claims about
significance that are based on assuming that the data is infinitely large (they are based on
asymptotic theory). Remember that typically we do not think that the data is normally
distributed, but that is approximately and we will talk about why this sort of approximation
is not what one needs.
Basics: Concepts: Null and alternative hypothesis, type I and II error, level, size, power, pvalue,
confidence interval. Data types: Single sample, independent samples, matched pairs,
Discussion of the usefulness of normality for large data sets.
Binomial test with and without assuming identically distributed data, sign test, confidence
interval for the median and other quantiles, confidence interval of distributions,
permutation tests including Wilcoxon Mann Whitney and spearman rank correlation test.
Test for the mean of a single sample, for comparing means given two independent samples
and for matched pairs, for the variance of a single sample and for comparing variances and
for analyzing covariance, for investigating tendencies described by a stochastic inequality,
and for running linear and ordinal regressions.
Motulsky, Harvey (1995) "Intuitive Biostatistics," Oxford: Oxford Univ. Press.
- a bit vague but precise
Lehmann and Romano (2005) Lehmann, E. L. and Romano, J. P. (2005), Testing Statistical
Hypotheses. New York: Springer.
- very precise but too mathematical for the applied
Schlag (2013): Exact Hypothesis Testing without Assumptions - New and Old Results
not only for Experimental Game Theory
and then there are original papers that some lectures will be based on:
Gossner and Schlag (2013): Finite-sample exact tests for linear regressions with bounded
dependent variables, Journal of Econometrics 177, 75-84.
Hoeffding, W. (1956), "On the distribution of the number of successes in independent
trials." The Annals of Mathematical Statistics, 27, 713-721.
Massart, P. (1990), "The tight constant in the Dvoretzky-Kiefer-Wolfowitz inequality," The
Annals of Probability 18, 1269-1283.