Category : Science and Education
Archive   : EGATESTS.ZIP
Filename : T-MANUAL.DOC

-Page 1-




EGATESTS Version 1.0

Users Manual


I. DESCRIPTION:

EGATESTS is an easy to use descriptive statistical program
designed to facilitate in the evaluation and assessment of classroom
tests. The purpose of this program is to give a high degree of
feedback on the performance of students on the examination and also
the effectiveness of the measuring instrument itself.


II. LIMATIONS:

There are two limitations on the kind of data that will be
accepted by the program: 1. All data must be integer data (UPPER
LIMIT 300); 2. The range of scores given must not exceed 88.


III. INPUT PARAMATERS

When the program is run the user will be prompted on various
options. These are: a) the number of points on the examination used
for calculating percentages b) the nature of the response data--
number of correct or incorrect responses. c) the manner of data
input--file or directly from the keyboard.

Note: To end input from the keyboard simply hit the t-key and press
ENTER.


IV. FILE FORMAT

Files need to to be simply a list of numbers in pure ASCII comma
deliminated format as follows:

3
4
6
7
8
13
10
9
7
t

The t at the end of the file indicates to the computer to terminate
input. In creating files, I simply use DOS EDLIN with each number
placed on a seperate line. See the sample file (SAMPLE.DAT) which has
been included for an example (120 pts., data is INcorrect responses).








-Page 2-




V. INTERPRETING AND PRINTING THE OUTPUT

A. Percentiles were calculated using the following formula:

PERCENTILE = 100 - [ (RANK - .5) / N x 100 ]

Note: The .5 subtracted from rank deals with the problem of
whether the subject being a member of the data set
scored better or worse than him or herself.

B. RELIALBILITY (KR21): Test reliability is estimated by the use
of a slight modification of Kudor and Richardsons KR21 reliability
statistic:
Points Mean x (Points - Mean)
KR21 = --------- x { 1 - ------------------------
Points - 1 Points x Variance

Please note that this statistic assumes that the data is dichotomous--
i.e. all answers scored as right or wrong (1, or 0). Violations of
this assumption (example: some questions worth 2 or more points) will
OVER-estimate the test reliability.

It is important to remember that reliability gives an indication of
how well the test maximized variance between students on one
homogeneous construct. Following is a subjective list for assessing
the reliability coefficient:

.60 to .70 Satisfactory
.70 to .80 Good
.80 to .90 Excellent
> .90 Professional test quality.

C. HISTOGRAM--The green connected lines on the historgram reflect
raw data frequency while the white bars represent smoothed frequencies.

D. To print your output I have included a copy of Charles
Petzold's EGAGRAF.COM screen print utility. This must be installed
before the program is run. To print the screen(s) simply hold down
the SHIFT-key and press the PRINT SCREEN key.


ACKNOWLEDGEMENTS:

Thanks to Dale Evers (Quincy, IL) and his BASWIND4 routines.
Microsoft Quick BASIC Version 4.0

Please send any comments or suggestions to:


Paul A. Wangerin
1821 W. Daisy Lane
Glendale, Wisconsin 53209

or

co Mark Wangerin via Exec-PC Mail