Organic Experiment Shoul Thomas Dee

The ideal way to approach a question like this is through the means of a random experiment (students randomly selected). In this manner, the study would be subject to as few confounding variables as possible. Secondly, it would be optimal to control for unobservable characteristics such as ability bias and student perception in the study. In turn, this allows us to reduce the error term which could lead to a clearer possible causation as opposed to just correlation. Thirdly, the sample size should be as large as possible while being accompanied by an increase in the scope of comparison from two to four classes. Like in any other experiment, increasing the sample size yields better results that can be applied more universally. Fourthly, the experiment

should take into consideration a student's past test scores in relationship with the gender of a students' previous teachers. This, thereby reduces the odds of giving credit where it is not due. For example, it would be incorrect to ascribe student test score gains to a teacher's gender if the student was performing well in previous years. Lastly, it is vital to check in with students at later years to determine if test scores were consistent with earlier findings. In this way results would be based off of a few years as opposed to just one year to give us a better picture of the overall effect. In following the list above, we should get a better representation of a teacher's gender true impact on the test scores of students.
Thomas Dee relies on a natural experiment to conduct his research.

Unlike a random experiment, a natural experiment is mostly observing what has occurred. What's more, a natural experiment gives researchers limited control over a study. Like a random experiment, variables are separated and compared to determine the effect of an independent variable on a dependent variable. At any rate, Dr. Dee selected to go with a natural experiment on the grounds that a random experiment was not a viable option. In particular, Dr. Dee's natural experiment focused on the results of a 1988 National Education Longitudinal Survey of 8th grade teachers. Specifically Dr. Dee examined the effect that a teacher's gender had on student test scores. Dr. Dee takes this a step further by considering the size of a classroom, the race of a teacher, years of experience a teacher has, and whether a teacher was certified in their subject of teaching. It is important to remember that Dr.Dee uses the results of one year and does not take into consideration following years. Beyond this, Dr. Dee also examines student perception about the four core subjects. In short, Thomas Dee examined and manipulated what had occurred to come to his

conclusion.
The biggest factor that separates a natural experiment from a random experiment is the lack of control over the independent variable. In this case, the impact that a teacher's gender has on student outcomes is being observed rather than experimented on. As such, a test score increase or decrease may be attributed to a teacher's gender when in fact, it may be the doing of another variable such as the open mindedness of peers. If Dr. Dee ran a random experiment the chances of this occurring could be reduced by controlling for certain variables. Another important difference between the ideal and natural experiment concerns the sample size. Dr. Dee draws on over 21,000 students for his study which is a large sample size by most standards. However, this sample size is small in comparison to the U.S student population. It may be that what works for a conservative student, may not work for a liberal student. Beyond all this, the experiment looks at the results from a specific year, one data point. This is of particular importance for two reasons. First, there is the issue of future effects on a student's education down the line. It may be that the educational gains produced as a result of a teacher's gender may diminish within a year. Second, there is the argument that past results do not guarantee future results. Equally important and more convincing is Dr.Dee's survey on subject perception by a student. Dr. Dee finds that a student's perception of a class coincidentally parallels their performance in that class. All things considered, it is hard to label differing teacher genders' as the cause of the discrepancies between boys and girls on test scores given the case of perception.

Citation
"The Why Chromosome - Education Next." RSS. Web. 5 Nov. 2014.