Accuracy and Error Edu645

Main Title: Accuracy and Error

Stephanie Grant

Ashford University

Learning and Assessment for the 21^st Century

Prof. Darrell Rice

December 13, 2012

 

Abstract

What is accuracy and error?  Accuracy is the closeness of agreement between a measured value and the true value, the quality or state of being correct or precise the ability to perform a task with precision.  Whereas an error; is a mistake the state or condition of being wrong in conduct or judgment.  An error is also, a variable in a statistical and / or mathematical mode, which is created when the model does not fully represent the actual relationship between the independent variable(s) and the dependent variable.  As a result of this incomplete relationship, the error term is the amount at which the equation may differ during empirical analysis. 

In this journal I will discuss what the aurthors (Kubiszyn & Borich 2010) mean by “all tests and scores are imperfect and subject to error.

 

 

 

 

 

 

 

 

 

 

 

 

 

The authors of your textbook state that “all tests and scores are imperfect and are subject to error” (Kubiszyn & Borich 2010), what do they mean by this?  According to (Kubiszyn & Borich 2010) test and the scores they yield are fallible.  They come with varying degrees of “goodness,” but no test or score is completely valid or reliable.  In other words, all tests and scores are imperfect and are subject to error.  If most or all test scores were perfectly reliable, we could put a great deal of confidence in a person’s score from a single test.  The notion of error in testing is very similar.  No test measures perfectly, and many tests fail to measure as well as we would like them to, that is, tests make “mistakes.”  They are always associated with some degree of error.  If you think about it if a computer can make an error on a person’s bill so can we as educators when administering a test, which lets you know no machine or person is accurate or perfect we all make mistakes.  If most or all test scores were perfectly reliable, we could put a great deal of confidence in a person’s score from a single test. And if test scores are perfectly reliable a student will always get exactly the same score.   Let’s review the table in chapter 18 which will give you a demonstration of hypothetical true scores and error scores.

Table 18.1 The Relationship among Obtained Scores, Hypothetical True Scores, and Hypothetical Error Scores for a Ninth-Grade Math Test

Student	Obtained Score	True Score ~ a	Error Score ~ a
Donna	91	88	+3
Jack	72	79	-7
Phyllis	68	70	-2
Gary	85	80	+5
Marsha	90	86	+4
Milton	75	78	-3

Hypothetical Values

Why it is important to recognize that there is no “perfect test”?  Knowing this, what can you do to ensure that you obtain the most accurate assessment results as possible?  It is important to recognize this because all tests are subject to various sources of error that impair the reliability of their scores and, consequently their accuracy.  If we understand these sources of error in test scores we should be able to minimize them in constructing tests and thereby improve test score reliability.  All test scores are fallible; they contain a margin of error there are many factors that increase error within a test such as: Trick questions, reading level that is too high, ambiguous questions, items that are too difficult, and poorly written items, to the extent that test items and tests are poorly constructed, test score reliability and accuracy will be impaired. 

What I would do to obtain the most accurate assessment results as possible is Assembling the test: based on chapter 11 (Kubiszyn & Borich 2010):

1.     Have written measurable instructional objectives.

2.     Prepare a test blueprint, specifying the number of items for each content and process area.

 

3.     Written test items that match your instructional objectives.

After I complete these activities I would package the test and reproduce the test.  With packaging the test I would follow the several guidelines they have listed which are:

1.     Grouping together all items of similar format.

2.     Arrange test items from easy to hard.

3.     Space the items for easy reading.

4.     Keep items and options on the same page.

5.     Position illustrations near descriptions.

6.     Check your answer key

7.     Determine how students record answers

8.     Provide space for name and date.

9.     Check test directions

10.  Proofread the test.

Before I reproduce my test I would also use the checklist which is listed in chapter 11 figure 11.1:

Test assembly checklist

Put a check in the blank to the right of each statement after you’ve checked to see that it applies to your test:

 

Put a check in the blank	Yes	No
1.      Are items of similar format grouped together?
2.      Are items arranged from easy-to-hard levels of difficulty?
3.      Are items properly spaced?
4.      Are items and options on the same page?
5.      Are diagrams, maps, and supporting material above designated items and on the same page with items?
6.      Are answers random?
7.      Will an answer sheet be used?
8.      Are blanks for name and date included?
9.      Have the directions been checked for clarity?
10. Has the test been proofread for errors?
11. Do items avoid racial and gender bias?

 

By utilizing all of these tools I have listed for packaging my test, I realize my test can still be subjected to flaws and errors.  Why?  For so many reasons I could have been tired that day and overlooked something, I could have been sick or had personal issues.  It doesn’t mean because a teacher test is flawed and have errors they are not qualified to teach we must realize that human error does occur just like computer errors.