Clear controls mean that it is replicable and it was also replicated in a way by the second procedure which suggests reliability. Also, Loftus has done many other experiments which prove its reliability.
Estimates of speed and saying yes/no to broken glass provides quantitative data so there is no interpretation meaning the data is more objective.
The students watched videos and this means that they were not under the same strain, emotions and pressure that actual witnesses are under so they experiments lack validity.
Only students were used and so this means that the results are not generalisable to the whole population. Therefore they do not have a representative sample.
All the students were studying psychology and there may have been demand characteristics because they might have worked out the point of the experiment and answered accordingly. They might have worked out that the questions were leading.