Hierarchical Linear Models were used to analyze data from one Texas school and identify effective and ineffective mathematics teachers using their students’ scores on two consecutive years of the state test (TAKS) over a three-year period. A model was developed which attempted to control for student grade level, as well as whether a class was an honors course. Special attention was paid to requiring statistically significant results. Results were minimal and may lack validity. The barriers to getting better results include missing data, the small sample size of students for an individual teacher, the non-random assignment of teachers to courses, and the extent of variability in the data. Most of these are beyond the control of educators. A better way of measuring student growth could reduce variability and improve the prospects of using a data driven approach to evaluate teachers. / text