Spearman correlation fails in ranking strings

Hi all,

I would like to compute the correlation between two rankings. Namely, I have two ranked variables, say:

table GDP

ln_gdp             country1
6.62273632394984	'BURUNDI'
6.96602418710611	'MALAWI'
7.00124562206948	'DR CONGO'
7.15539630189673	'MOZAMBIQUE'
7.36518012602101	'CHAD'

and a money2 table:

money              country2
0.0871934820000000	'MALAWI'
0.123754360000000	'DR CONGO'
0.193845570000000	'CHAD'
0.230036880000000	'BURUNDI'
0.264654610000000	'MOZAMBIQUE'

Now simply I would like to compare the two variables ranked country1 and country2 as:

corr(money2.country2, GDP.country1'type','Spearman');

but the following error occurs:

Check for missing argument or incorrect argument data type in call to function 'imag'.

Error in corr (line 192)
complexdata = ~(isreal(x) || all(imag(x(:))==0));

what am I doing wrong?

P.s. the rownames are the names of the country.

Thank you

Answer 1

[Jeff Miller]

The first two inputs to the corr command have to be the numerical values that are to be correlated, so you will want something like

corr(money2.money,GDP.ln_gdp,...)

However, you have to make sure that the rows of the two tables correspond (e.g., first row of each table has BURUNDI, second row has CHAD, etc). Maybe you can do that with sortrows()