There are a couple of notes that we would like to make that are relevant for the paper:
1. Due to a programming glitch, we did not record age in Study 2.
2. In Study 1, we included the SRP. We did not analyze these in depth, but the data are available in our Study 1 dataset.
3. In Study 2 we also asked participants to describe the relationship with each of the five persons. We did not analyze this further.
4. Our Study 2 was preregistered as an exercise as part of Bart van Heeren's master thesis (see [here][1]). We then preregistered the main effect (cold vs. warm), as we had not yet analyzed the data in full, and we had expected this main effect to replicate. Despite that it may appear not to, the main effect did replicate, if one takes into account distribution across conditions of the RCR (randomization failed; distribution of RCR across conditions varied).
5. We took a number of steps in coming up with our exploratory results. We first examined correlations between the average IOS and each individual items per condition (subtracting cold from warm, and looking at maximal differences between the conditions). We then came up with the approach to do a factor analysis that included the average IOS and each item per condition, and again looking at the maximum difference per condition. Our final approach however we thought to be more accurate, examining the interaction term per item. We therefore chose to retain this, and this effect replicated.
[1]: https://osf.io/5skyx/