Dear Editor:

My name is Colin O’Haire.  I’m a math teacher at Dougherty Valley High School.

I was quoted in the Wildcat Tribune in regards to the perceived epidemic of a growing disease labeled “test flu.”  In that article, I gave only what was anecdotal evidence that I believed well-represented the trend happening at this school.  My argument was that there existed a clear pattern of students missing classes, disproportionately, on test and quiz days or days when a large assignment was due.

After some reflection, I decided that it would be prudent to provide some scientific evidence.

The Basic Idea

I decided to create an experiment with the following hypotheses:

Ho – There is no difference between the number of students who have “test flu” compared to the number of students who miss regular school days.

Ha – The proportion of students absent on test days is higher. Using P<.001 to show the result is statistically significant.

Mathematically, if P1= percentage of students with test flu and P2= percentage of students missing standard school days, the goal was to show that the probability that P1-P2=0 is very low, using a 2 sample z-test for difference in population proportions.

Experimental Design

I went through Infinite Campus (our attendance tracking software), and found the number of students absent in each of my four Precalculus classes on each test day.  There were a total of 125 absences and a total population of 1,870 students attending the 14 days with assessments.

Then, using a random generator that corresponded random numbers to days of the school year, I found the number of absent students on the days generated.  There were 46 absences for the same sample size of 1,870 students.

I only included absences that were listed as “sick,” “unexcused” or “cut”. I did not include students who were absent for school-related events.

The Results

When I put the data into my calculator to run the test, the return was a z-score of 6.18.  This z-score corresponds to a P value of  0.00000000134 percent.  The P-value is the probability of getting a sample proportion as high as or higher than the one obtained if the null hypothesis is true.  The P-value obtained shows that it is quite unlikely, and therefore, allows me to reject H0 in favor of Ha.  

The Interpretation

Statistics show that there is a difference in attendance between the two populations.  We cannot say definitely that the reason students are absent more frequently, to an extreme degree, is that there is an assessment of some sort. There could be other confounding variables in play, like test flu microbes in the water supply, or perhaps test flu has become self-aware and is targeting students specifically on test days.  

I hope, for all our sakes, the latter is not true.  

While we have the tools to vaccinate against test flu — tools in the form of reviews, ACCESS period, checks for understanding and our own moral codes — there may never be a cure.  If test flu has, in fact, become self-aware, we may not be able to curb its growth through conventional means.

Test flu does not discriminate.  Test flu does not care if teachers have to spend hours (over the course of rewriting 20 or so tests) making new forms of their assessments.  Test flu does not care if students have an unfair advantage over their peers by taking a test after everyone else has seen the content.  

Ending the epidemic falls on all of our shoulders.

Yours in solidarity and hope,

Colin O’Haire