As Rui noted, to get a trend you need to focus on percent gray vs age since hair color is not an ordinal/rank variable. Then use a measure of association designed for rank variables of which there are many: Spearman?s r, Kendall?s tau, gamma, tau-c, Somers-d. All of them are available in package DescTools.
David C
From: vod vos [mailto:vodvos at zoho.com]
Sent: Tuesday, January 10, 2017 6:03 AM
To: David L Carlson
Cc: Rui Barradas; r-help
Subject: Re: [R] how to proof the trend of two columns of data?
as far as I know, ?kruskal.test will show us the differences between three or more groups. But it could show the trend.
---- On ???, 09 ?? 2017 10:12:14 -0800 David L Carlson <dcarlson at tamu.edu<mailto:dcarlson at tamu.edu>> wrote ----
The list does not assist with homework problems. If this is not a class assignment, you should be more specific about what you have tried and provide a reproducible example (a sample of the real data or some made-up data that has the same columns and data types). In the meantime you could also try the following R command:
?kruskal.test
-------------------------------------
David L Carlson
Department of Anthropology
Texas A&M University
College Station, TX 77840-4352
-----Original Message-----
From: R-help [mailto:r-help-bounces at r-project.org<mailto:r-help-bounces at r-project.org>] On Behalf Of Rui Barradas
Sent: Monday, January 9, 2017 11:28 AM
To: vod vos; r-help
Subject: Re: [R] how to proof the trend of two columns of data?
Hello,
Inline.
Em 09-01-2017 14:55, vod vos escreveu:
Hello everyone,
If there are two columns, one is age (numeric, cut to several groups), the other is hair color type(factor: yellow, black, white).
If the age column is not normal distributed
If you use ?lm, it's the residuals that should be normally distributed,
not age.
You can also use ?glm with a binomial link, in which case you should
recode type as white/not white.
Hope this helps,
Rui Barradas
, which statistic method should use to prove the trend relationship
between them, for example, the older has more probability of white hair
type? Are there any existed R package to figure out this situation?