I did Pearson's correlation coefficient between my independent variables and a dependent variable. I got a significant relationship between independent variables along with relationship between dependent and independent variables. So the problem is Iam unable to interpret the positive correlationship between these independent variables.
If you have multiple independent variables, run Multiple regression. It will give you the correlation value between each independent variable with dependent variable. so you can easily interpret the results.
I don't see how Pearson's correlation will give you a relation between variables that are independent. The coefficient will be 0 because correlation coefficient only detects linear dependencies between two variables. Like Shanti said, your best option is to run multiple regression because regression can accommodate dichotomous independent variables. A two-way ANOVA could also be employed since you have 2 independent variables and a single dependent variable.
Do you have multiple dependent and independent variables? And are the dependent variables normally distributed. If so you could run a Mulitivarite analysis of variance. This would account for the unexpected relatedness between the independent variables and the relatedness between the dependent variables. Good luck.
I have only one dependent variable and other categorical independent variables. So, when I did the Pearson Correlation Coefficient in SPSS, the matrix showed the significant value to the dependent variable and other independent variables horizontally. Suppose Age(independent variable) to distress score(dependent variable) I found significant value along with other independent variables like marital status, education and income.
Then the best analysis is performing a regression. You will need to dummy code your categorical variables. Make sure you begin with zero this will make your intercept interpretible. It will give you a partial coefficient for each independent variable. Other continuous variables will need to be mean centered. For example the intercept if you coded males as zero, high school as zero and mean centered income regressed into distress, could be interpreted as low education males with an average income experience report higher (or lower) distress than females with a higher education. A Regression model would also account for the overlap in variance to make it easier to interpret. In this model you could also explore moderations which may prove interesting. Good luck.
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation