Fig 1.
Predicting relationship stability based on negative interaction (on the left) and positive interactions (on the right). Results indicate that relationship stability could only be predicted by the emotion data from the positive interaction.
Fig 2.
As we were unable to predict the breakup with model 1 based on the negative interaction (left bar), removing information such as mean, variance and coupling had no meaningful effect on the predictive power.
Fig 3.
The left bar indicates the prediction for relationship stability for the positive interactions (Model 1).
In comparison to this, the other bars indicate that mean centering leads to hardly any change, whereas removing the variance and the coupling (i.e., time ordering) have strong effects.
Table 1.
Point-biserial correlations examining whether the emotional variance of someone’s time series correlates with whether they break up.
Fig 4.
Depicts bootstrapped histograms for the average log of the magnitude of the CPSD over six sub-bands for the positive interaction.
Each sub-band represents a range of fluctuation rates. For example, 0-5cpm represents fluctuation rates between zero and five cycles in a minute. We present two histograms in each subplot, one for the group of couples who broke up (grey) and one for those that stayed together (blue). Note that the couples who stayed together (consistently the histogram towards the right hand side of each subplot) have higher log-magnitude-CPSD than those that broke up (consistently towards the left hand side of each subplot). Given that CPSD measures a type of coherence in fluctuation, this suggests that couples who stayed together exhibited higher coherence in their fluctuations than those that broke up, over all sub-bands.