topic Re: t-value for the difference between two splopes in Statistical Procedures

t-value for the difference between two splopes

Juan_Manriquez — Thu, 30 May 2019 13:16:38 GMT

How I can run the equation to determine the difference between two slopes in SAS?

t = (b1 - b2) / SQRT(s^2b1 + s^2b2), df = n1 + n2 -4

b = slope ; s = standard error ; df = degree of freedom and n = sample size

Thanks in advance,

Juan Manriquez

Re: t-value for the difference between two splopes

PaigeMiller — Thu, 30 May 2019 13:20:34 GMT

This is equivalent to testing if the two slopes are equal. To do that, in a regression sense, you fit an interaction, and if the interaction is not statistically significant, then the two slopes are not statistically different.

For example, if the class variable ID contains values 1 and 2, and this is used to produce two slopes

proc glm data=have;
    class id;
    model y = id x id*x;
run;
quit;

Re: t-value for the difference between two splopes

Juan_Manriquez — Mon, 03 Jun 2019 14:03:51 GMT

Thanks Paige, I will check it
Juan