The biggest weakness in my study was a response bias. I only had 86 people people respond and out of those 86 surveys only 40 were usable. 46 people didn’t put their gender and grade level, rendering their surveys useless. Out of the 40 usable surveys, I received 19 male responses and 21 female surveys. I probably didn’t get the other 34 surveys back because they students were in Ombudsman, had a strict teacher that didn’t pass out the surveys, or the student didn’t feel like handing them out. I did, however, get all the teacher surveys back.
Because the size of usable surveys were so low, I had to change my study up a little. The original question was “Is there a correlation between reading and a person’s gender”. I asked additional questions on my survey that asks about the type of reading the sample partakes in. Using that information, I changed my question to “Is there a correlation between what a person reads and their gender?”. This change made sense because there was more data to work with and more variability.
Despite the improvements, I still had some issues. There was some funky stuff on my Chi-Squared test for the expected value for males who read blogs. The value was 4.696, which is less than 5. I ended up rounding that value up to 5, but the “Blogs” comparison might be a little off. The rounding change is minor, but it might make a small difference.
My results cannot be extrapolated anywhere else. I had some issues with my data and calculations that may of effected my results.
I think the population of interest should be widened. I had a limited age group in the high school and the teacher’s ages were too spread apart to make any effect when you compare them to the student’s ages. I would need to survey more adults in intervals of 8 years. My new sample size would be ages 10-17, 18-25, 26-33, and so on. This gives a lot more variability with different age groups. Speaking of changes, I would make my survey more clear. As I stated above, people didn’t fill out the gender or grade tabs. Next time I will ask that in a multiple choice format so the sample knows it isn’t optional and so it’s easier to see. This will help eliminate the response bias.