What's normal anyway? Residual plots are more telling than significance tests when checking ANOVA assumptions

Chow (1996) offers a reconceptualization of statistical significance that is reasoned and comprehensive. Despite a somewhat rough presentation, his arguments are compelling and deserve to be taken seriously by the scientific community. It is argued that his characterization of literal replication, types of research, effect size, and experimental control are in need of revision.

Download Full-text

Nonparametric rank-based statistics and significance tests for fuzzy data

Fuzzy Sets and Systems ◽

10.1016/j.fss.2005.01.008 ◽

2005 ◽

Vol 153 (1) ◽

pp. 1-28 ◽

Cited By ~ 60

Author(s):

Thierry Denœux ◽

Marie-Hélène Masson ◽

Pierre-Alexandre Hébert

Keyword(s):

Fuzzy Data ◽

Significance Tests

Download Full-text

Origin, evolution, and tectonic setting of the eastern part of the Mexican Volcanic Belt and comparison with the Central American Volcanic Arc from conventional multielement normalized and new multidimensional discrimination diagrams and discordancy and significance tests

TURKISH JOURNAL OF EARTH SCIENCES ◽

10.3906/yer-1412-31 ◽

2015 ◽

Vol 24 ◽

pp. 111-164 ◽

Cited By ~ 14

Author(s):

Surendra P. VERMA

Keyword(s):

Tectonic Setting ◽

Volcanic Belt ◽

Central American ◽

Significance Tests ◽

Volcanic Arc ◽

Mexican Volcanic Belt ◽

Central American Volcanic Arc

Download Full-text

Was This in Your Statistics Textbook? IV. Frequency Data

Experimental Agriculture ◽

10.1017/s0014479700016392 ◽

1989 ◽

Vol 25 (1) ◽

pp. 11-25

Author(s):

D. J. Finney

Keyword(s):

Statistical Analysis ◽

Contingency Tables ◽

The Other ◽

Frequency Data ◽

Significance Tests ◽

Test Statistic ◽

Jumping To Conclusions ◽

Proper Role ◽

Measure Of Association

SUMMARYObservations that are frequencies rather than measurements often call for special types of statistical analysis. This paper comments on circumstances in which methods for one type of data can sensibly be used for the other. A section on two-way contingency tables emphasizes the proper role of χ2 a test statistic but not a measure of association; it mentions the distinction between one-tail and two-tail significance tests and reminds the reader of dangers. Multiway tables bring new complications, and the problems of interactions when additional classificatory factors are explicit or hidden are discussed at some length. A brief outline attempts to show how probit, logit, and similar techniques are related to the analysis of contingency tables. Finally, three unusual examples are described as illustrations of the care that is needed to avoid jumping to conclusions on how frequency data should be analysed.

Download Full-text