In order to answer this question, we add dummy variables (Hardy, 1993) which encode the use of a particular type of material and then resample the database 500 times. We then average the simulation results and compare these to the baseline mean and pass rate. This procedure is continued by dropping out and adding in different combinations of
The results of the lower-bound materials and in-house teacher training characteristics.
'value-added' simulation is shown in Table 2j.