Figure 1.

Example data for one person, generated by a three-level empty model, with randomly varying day means and randomly varying affect scores within days. The top part shows that relative to the day means (solid lines) there is no visual indication of large carry-over from moment to moment. While inertia is often hard to detect by looking at the data, lack of any visible carry-over does indicate that the inertia, if present, is relatively small, whereas visible carry-over in the data would imply a large inertia. In the bottom part of the figure, where the affect scores are only evaluated against the estimated trait level in the two-level AR(1) model, it appears like there is much carry-over, because entire days are characterized by above-average or below-average scores. Thus, it is no surprise that this model results in a significant estimated beep-level inertia.