brothers

Performance Analysis

1. Based on the data analysis from this test, the class scored poorly. The mean score

of the test was a score of fifty-one. A mean score from seventy-five to eighty-five would

have been an optimal score. Because they scored lower on the mean than seventy-five,

either the students didn’t know the material, the test was poorly written, or the mean was

affected by outliers. The mode of the test was sixty-one. This is my most frequent score,

but it only focuses on one group of students. The median score of the test was fifty-six. A

reliable test should have the mean, median, and mode within five points of each other.

My scores were fairly close to each other. The median and the mode were five points

apart and the mean and the median were five points apart, but if you compare the mode to

the median it is ten points apart. There are probably some reliability issues with my test. I

should probably go back over the test and review each item to figure out which items the

students did poorly on and which ones the students did well on. The highest score on the

test out of all fifteen students was a seventy. The majority of the class failed the test. The

frequency distribution graph was bimodal with a high peak and a low peak. There wasn’t

a very good range of scores, so it did not make a bell curve. The graph was negatively

skewed and the skew score was 2.27. This means that there were more scores below the

mean. My standard deviation was 2.28. The standard deviation is the group score around

the mean. A good score is 7-13. My standard deviation is low and this means that too

many students scored close to the same level whether that level is good or bad. In this

case, the students all scored on a low level. My mean score for test objective one was

about forty-four and my mean score for test objective two was fifty. Neither of these

mean scores was very high, so this indicates that the objectives were not covered

effectively in class and the students were probably unprepared for the test. In order to

help my students do better next time on the test, I should probably allow more time to

cover the material and use different teaching methods than I used to prepare for this test. I

could probably have a review with the students the day before to make sure they know

the objectives before administering the test. Using this data from the analysis of the test

scores, I can analyze what is wrong with the test items or determine a weakness in my

teaching of a subject covered under the objective. This information will help solve the

mystery of why the students did so poorly on the test. I should analyze any changes I

made in teaching practices, because this probably affected the students’ ability to learn

the material effectively. I can go back and look at each question and find out what items

students struggled on. Based on the measures of central tendency, the skew, the

frequency distribution, the standard deviation, and the mean scores of the test objectives,

there is definitely good reason to go back and alter the test to make it more reliable for

future students.

Item Analysis and Reliability

2. There was a mixture of results in my item analysis. According to item difficulty,

items two, four, seven, ten, fourteen, and eighteen were a little too easy for the students.

Multiple-choice items tend to have the best reliability when their item difficulty is about

.69. According to the scores for item difficulty, items three, five, six, nine, eleven,

twelve, thirteen, fifteen, twenty, twenty-one, twenty-two, twenty-three, twenty-two, and

twenty-three were too difficult for the students. In response to the scores on item

difficulty, I should go back and work on making the easier questions a little more

challenging and making the more difficult questions less difficult. Some of the scores fell

around .69, so I would keep those. As far as item discrimination, some questions proved

to be good items based on the discrimination between the upper and lower students. Item

discrimination is meant to show the comparison of students who were prepared for the

test minus the students who were less prepared and got the right answer on the item. An

item discrimination of .4 or higher usually demonstrates a good item based on

discrimination. All of the test items, except one, two, nine, sixteen, eighteen, nineteen,

and twenty proved to be good items based on item discrimination. In order to fix the bad

items from the test based on item discrimination, I should consider the readability, the

distractors, and check to see if the item matches the objective being tested. I find that

based on item difficulty and item discrimination, most of my test items should be revised

and some items discarded. Two items that should be thrown out or revised are items five

and nine. Using distractor analysis to analyze the questions, we can understand what

might need to be changed to make the item better. On question five, options B and C

should be discarded because the less-prepared students scored better than the better-

prepared students. Option A is a good distractor because both the less-prepared students

and the better-prepared students chose it equally. Option D was the correct answer and it

is reliable because the less-prepared students scored lower than the better-prepared

students. On question nine, options A, C, and D should be discarded or revised because

the less-prepared students scored higher than the better-prepared students. Split-half

reliability determines if the test was reliable as a whole. This test scored a .6 which is a

good score to prove reliability. Some potential threats to the reliability of the test were the

testing environment, a change in teaching practices, and if the items didn’t measure the

objectives. I could improve these by changing the testing environment, altering my

teaching methods to a more reliable method, and revising or discarding test items that

didn’t measure the objectives. The results from this analysis can be used to make the test

more reliable in the future. The mean scores of the essays were 2.53 and 1.8. I ensured

the consistency of the essay scores by using a rubric. The essays were graded randomly to

ensure reliability as well. The cover sheet with the student’s name was covered so the

essays were graded anonymously.