Why Aren't My Correlation Results Significant? I am using SPSS to do a Spearman test on data from a Likert scale style survey. I correlated two questions, one of which 81.9% chose "1" and 10.2% chose "2", and less than 5% choose 3,4,or 5, with a question that had a similar trend in the results: 59.3% choose "1" and 16.4% chose "2", 16.8% chose "3", and less than 5% chose 4 or 5 (I was unable to insert the charts of these results). The correlation coefficient was .006, which surprised me because it is so insignificant. I was expecting at least some correlation since although the results weren't exactly the same, they still showed the same general trend. Do you think I am doing something wrong in my analysis or does this seem like an accurate outcome? If so, how should I interpret it? 
 A: You're confusing similarity of marginal distribution with correlation. They're (almost) entirely unrelated things. 


*

*For the two variables with very similar distributions to be positively correlated, the 1's or 5's (etc) would need to tend to occur together in the same individual. There's no indication from the univariate distributions whether this tends to happen. It might happen or it might not. 

*The variables could be perfectly correlated with very different distributions and on the other hand with the same distribution they might have almost any correlation. 
(a) Consider if on Q1 all the answers were 1 or 2 and on Q2 all the answers were 4 or 5, but the 1's and 4's occurred together as did the 2's and 5's. Very different distributions, but perfectly correlated
(b) Now consider for both questions you have 25% in each of 1,2,4,5 for both questions, but now 1 always pairs with 5 and 2 always goes with 4; identical distributions but the correlation is -1.

Here's a simple example with somewhat similar characteristics to your problem. Consider two variables (X and Y) that each take the values 1 or 2 with probabilities 0.6 and 0.4 (i.e. they have the same distribution, a 60-40 split).
Here are three bivariate tables (joint distributions) that each fit with this information:
       Table I          Table II         Table III
          X                X                 X
       1    2           1     2            1     2
    1 0.6  0.0       1 0.36  0.24       1 0.2   0.4
  Y                Y                  Y
    2 0.0  0.4       2 0.24  0.16       2 0.4   0.0  

The first table has correlation* 1, the second table has correlation 0 and the third table has correlation of about -0.655, but all three tables have exactly the same marginal distribution on the two variables (60% 1's, 40% 2's).
* Here it makes no difference whether we look at Spearman or Pearson correlation.
