You are on page 1of 25

Ap Stats Culminating Project Rough Draft

Hunter Lacamu
AP Stats
Mr. Kiker
How Smart Do You Think You Are (How Smart Are You Really?)
Students Assigned Tests based on random subjects will estimate how well they will do then will
take a test to see if there is an association between how well they think they will do and how
well they actually do. This data will be used to determine if students tend to do differently than
expected on exams.

Do you think that how well you think you will do on a test will affect the actual score?
From personal experience I know that if I feel bad about a particular test I am about to take I
usually do not do so well. The same applies for most people, if someone is not comfortable with
the subject of algebra that will probably show on a test that utilizes some form of algebra. Different tests make people feel differently about how they do, for example someone might feel better
about how they will do on a chemistry test rather than a US history test.
This project may be of use to those who worry about tests that they prefer not to take, it
would be useful to know that maybe if they set higher standards they will actually perform better.
Multiple choice tests are more likely to be this way, since when it comes to open answer tests
people either know the answer or they will get it wrong, In multiple choice tests there is always a
chance of getting the answer right and a persons mental state concerning the test may be a
factor in how well they do.
Using data from students of Austin High I tested the hypothesis that what a person thinks
about how well they will do will affect their test score by administering tests to students and having them estimate their score before they take the test. Signatures will be gathered from all the
people that I test and will use the same wording for each individual when giving instructions. My
tests will consist of 10 multiple choice questions of one random subject test of 4 available subjects. These tests will allow me to run a chi-squared test as long as the students selected and
the test administered to them are random, and as long as I eliminate as much bias as possible.
A chi-squared goodness of fit test works for the test above, because I will be testing multiple subjects and I am looking to see if how a person thinks they will do affects how they will do.
In other words, I am looking to see if the expected values are relatively the same as the observed. If the students were asked to study for these tests, I would have had to worry about the

lurking variable of study time, but in since the test is being administered on site this will not be a
problem.
Once my data was collected I used my independence test to determine a p-value that
will then allow me to decide whether I can reject the null hypothesis that the two variables are
independent. If this is the case I will have evidence to suggest what type of association there
may be between the two variables and my results can be applied to personal experience as well
as what I expected to occur from testing my subjects. Things like undercoverage may occur in
my experiment because there may be a test subject that some consider important that I may not
include. The objective of this experiment does not have so much to do with which test a subject
takes, but how well they did versus how well they thought they would do. I personally think even
if I gave everyone the same test I would still be able to reject the null hypothesis, but in order to
have a completely accurate experiment I made sure to include different tests. This also made it
possible to run a chi-squared test with multiple categories to really determine if the two variables
are independent or not.
My p-value came back rather large (.63) and because of that I was unable to reject the
null that the distribution of observed values was the same as that of expected for these tests.
My data did not give me evidence to suggest anything, but it did allow me to see that perhaps
how a person expects to do on a test is about how they will do since the observed and expected
values were fairly close and the p-value was so high.
If I were to do this experiment again, I would definitely focus more on the difficulty of the
tests, and by that I mean I would make sure that the tests were about of equal difficulty regardless of subject matter covered. I feel like my math test was too easy for those that took it because the majority of people that took the math test got a perfect score. In the future it would
also be necessary to decide my wording beforehand, including how I reveal which subject
someone will be taking.

Works Cited

"Teachers' Expectations Can Influence How Students Perform." NPR. NPR, n.d. Web. 27 May
2015.

Appendices

You might also like