Evaluating a new ESL curriculum through quasi-experimental design in Hutchinson schools

¶ … Hutchinson school district wants to investigate the effectiveness of a new curriculum designed at teaching English as a second language to elementary students. The design for the study will be a quasi-experimental design (Creswell, 2012). The independent variable will be the implementation of the new curriculum. The dependent variable will be the test score on the Stafford English Language Proficiency Test (Stanford ELP). The school district can implement an effective quasi-- experimental design designed by doing the following:

First, it is important to select two separate classes at the same grade level that include an equivalent number of students that would potentially benefit from this intervention. It would be best to utilize classes at different schools to avoid any potential breaches of the security and internal validity of the quasi-experimental design. Teachers in each condition will go about the normal business of teaching their classes without knowing that their data will be used in a quasi-experiment. One class will serve as a control group (students being taught as they are traditionally taught) and the other class will serve as an experimental group (students receiving the new curriculum). Teachers in each of the classes will not be informed that there will be a comparison between the two classes. Thus, utilizing two different schools will also limit any complications that might occur if students in the same school from different classes became aware that they were involved in different programs (demand characteristics from both teachers and students).

Secondly, both classes will take the Stanford ELP at the very beginning of the school year to establish a baseline score. At the end of the school year the Stanford ELP will be re-administered the both classes (there are parallel versions of the test; Pearson Assessments, 2015). This will allow for comparison between the classes regarding the relative the treatment in both the control and treatment condition (within subjects design; Creswell, 2012).

The null and alternative hypotheses for the overall result are as follows:

H0: Control Group time 2 Stanford ELP = Treatment Group time 2 Stanford ELP

H1: Control Group time 2 Stanford ELP < Treatment Group time 2 Stanford ELP

Where the time two Stanford ELP represents the change score from time one to time two.

There are five dimensions of language measured by the Stanford ELP and making comparisons at all five levels will provide more complete information. In addition, student grades in individual classes as well as their overall GPA could be utilized as dependent measures for more descriptive analyses. Letter grades can be operationalized on a continual ratio scale of 0.0 through 4.0 in the same manner that overall GPA is often presented. Other information that could be useful for analysis could be gather satisfaction data from both teachers and parents regarding how the children progressed in each program. This would require the implementation of an existing questionnaire or the development of a self-report questionnaire to serve this purpose.