View Full Version : Statistical Significance
01-04-2009, 06:22 AM
I am completing a statistics project having been given data for 30 years.
I have calculated the pairwise correlation coefficients between Infant Mortality Rate and 3 variables, I now need to test the statistical significance of each correlation coefficient. How do I do this?
01-04-2009, 12:08 PM
Are you using the Pearson correlation?
Given the number of cases, you can work out the degrees of freedom (df=n-2). Then critical Pearson 'r' value for signficance at a particular alpha level (typically 0.05) can be found using a table of critical values such as this one:
However if you are using a stats program like SPSS or SAS it should tell you if the correlation is signficant automatically. Also remember to make sure you are using the correct correlation type, e.g. it is inappropriate to use Pearson if your data doesn't meet parametric assumptions.
Hope this helps
01-04-2009, 02:42 PM
Since he has 30 years, the parametric are not necessary (willimam means normal distribution of the data). Why don't you use multiple regression to put all the other three variables in a model and test their significant contribution to the mortality rate. Also, try some demographic statistics to see if there is something better for that, e.g. an actuarial model.
Powered by vBulletin™ Version 4.1.3 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.