Karteikarten
7. nonparametric models
Karten in diesem Stapel (14)
- 1when to use nonparametric testswhen you can't correct the problems in your data, especially if you have a small sample (and thus cannot rely on the central limit theorem). they're assumption-free tests.
- 2are nonparametric tests less likely to detect a genuine effect than a parametric one?only when the assumptions for the parametric tests ARE met, otherwise no
- 3tied ranks + how to solve itranks given to scores that occur more than once in a data set. give them all their average rank
- 4when there are multiple p-values given, when to use the exact method?when your sample size is small
- 5under what name can you find the z-score in the output?standardized test statistic
- 6which value is a small effect size and which one indicates a huge effect size?small: 0.3
huge: >0.5 - 7idea behind the Mann–Whitney testrank the scores from lowest (1) to highest (n). the sum rank per group tells you something about the effect: if the sum rank per group is about the same then there is no difference
- 8idea behind the wilcoxon signed-rank testranking the differences between scores in the conditions. per condition, add up the ranks to get the sum of the positive ranks (T+) and one of the negative ranks. T(+) is the test statistic, which you can use to calculate the z-score.
- 92 models that can compare more than 2 conditions + dependent or independentkruskal-wallis test (independent) and friedman's ANOVA (dependent)
- 10idea behind Kruskal-Wallis testrank scores from lowest to highest, ignoring the group to which the score belongs. add up per group. sum of ranks per group is denoted with Ri, i = group.
- 11what does the Kruskal-Wallis test tell you?whether groups come from different populations, but not which groups differ.
- 12what does adj. sig. tell you in comparison to just sig.?adj. sig. is the significance value adjusted by the Bonferroni correction, so you should look at that one
- 13what is a 'homogeneous subset'?a cluster of multiple very similar groups
- 14idea behind Friedman's ANOVAyou rank the data for each person. then add up the ranks for each condition.
Informationen
- Beschreibung
- ssr exam 2 Karteikarten am 7. nonparametric models, erstellt von Sabine - am 17/10/2020.
- Erstellt von
- Sabine -
- Aufrufe
- 1
- Bewertung