Table 4. Correlation.
Decile | Correlation | n | Null low | Null high |
1 | 0.156366 | 6703469 | −0.000960 | 0.000976 |
2 | 0.156163 | 7750717 | −0.000905 | 0.000815 |
3 | 0.156499 | 8253044 | −0.000812 | 0.000840 |
4 | 0.156321 | 8552987 | −0.000766 | 0.000832 |
5 | 0.157428 | 8882969 | −0.000750 | 0.000809 |
6 | 0.158828 | 8847712 | −0.000815 | 0.000793 |
7 | 0.161030 | 9041679 | −0.000794 | 0.000709 |
8 | 0.163359 | 9407243 | −0.000736 | 0.000755 |
9 | 0.169258 | 9356494 | −0.000788 | 0.000762 |
10 | 0.190253 | 9122388 | −0.000761 | 0.000785 |
The estimated correlation between a user's validated turnout and the validated turnout of her friends by decile of user friend interactions. See Figure 5 for a visualization of this relationship. To compare the observed values to what is possible due to chance, we keep the network topology fixed and then randomly permute the voting behavior of friends. We repeat this procedure 1,000 times and measure the correlation. The simulated correlation values generate a theoretical null distribution for the correlation which we would expect due to chance. The Null low and Null high columns display the 95% confidence interval of this null distribution. Note that the observed correlations exist well outside the null distributions.