SCORE: An APL Program for the Scoring of Tests
Vector 5:4, page 86
Transcribed from the printed issue; not yet reviewed. Read it in the PDF of the issue, page 86
Introduction¶
Score is a general purpose program for the scoring of tests. The program scores data according to a key that is customised by the user, and options exist for keyed, reversed, and weighted scoring of data.
Detailed instructions for the program (a printout of these is included as an Appendix) can be accessed anytime by the user. Two methods of input are allowed:
- Interactive input, where the user is queried for up to 512 (including blanks) elements of data.
- Batch input, which allows for a virtually unlimited amount of data.
Output includes the scoring key(s), subject identification, input vector, total number of elements entered, and scores for all the factors (i.e. each scoring scale).
Score has three advantages:
- the program is flexible, with uncomplicated, customised keys that allow scoring of an unlimited number of items and factors.
- data can be scored in a variety of ways by changing the keys and rescoring.
- large sets of scoring keys for different tests (e.g. batteries) can be compactly stored and easily retrieved for future use.
Score in APL¶
SCORE is a dyadic function with both arguments containing segmented character strings. The right argument KEY contains the scoring key. This argument requires a major segment for each scale and optional minor segments for special scoring features of each scale. The optional left argument DATA contains subject identification and item responses for each subject.
If DATA contains no elements, SCORE reads data interactively and the user is prompted to enter subject identification and item responses. In this mode the major segments of the KEY argument (i.e. each scoring key) as well as each scale in KEY and DATA (or, if interactive, each subject entry) are separated by a carriage return. Thus, for input, each subject's data and each scoring key for that data is entered on a new line. Each minor segment within the scoring key and within the data set is separated by '/'.
Two main functions are used for scoring: SCORE and RECODE. SCORE is the primary function that calls RECODE if recoding of data is done. In addition, two one-line utility functions, sSINDEX and sSSHAPE, are also used. For all functions, the variable DATA contains all subjects' data, and SCORE loops through DATA for each subject. Within SCORE, the variable D contains data for one subject and F is a counter indicating which factor is being calculated. Similarly Nf is the number of factors contained in KEY, S is the subject loop counter, and Ns contains the number of subjects stored in DATA.
DATA SCORE KEY;INT;D;Ns;NF;F;S;T;V;VALS;⎕ELX
--------------------------------------------
[1] ⍝ Key is segmented string (e.g., 'ONE/1 2/1/0'). Each scale
[2] ⍝ is separated by ⎕TCNL & "/" separates fields in a scale.
[3] ⍝ Fields: 1=scale name 2=items 3=multiplier 4=add.
[4] ⍝ DATA is the optional left argument of a character vector
[5] ⍝ with Subject ID, "/", and the item values. Each on newline.
[6] ⍝ If DATA does not exist, the program asks for Subj & data.
[7] 0 0 ⍴1 ⎕POKE 96
[8] 'Scoring key:' ⋄ NF←sSSHAPE ⎕TCNL,⎕←KEY
[9] →LOOP×⍳INT←0=⎕NC 'DATA'
[10] Ns←sSSHAPE DATA←⎕TCNL,DATA ⋄ S←0
[11] NXTs:→0×⍳Ns←S+1 ⋄ D←1↓DATA sSINDEX S
[12] '----------> Subject ID: ',¯1↓(T←D⍳'/')↑D ⋄ 'Data :',D←T↓D
[13] →nUM
[14] LOOP:⍞←'----------> Subject ID: ' ⋄ →0×⍳0=⍴⍞ ⋄ ⍞←'Data : '
[15] →0×⍳0=⍴D←⍞
[16] nUM:→OK×⍳' '∊D ⋄ D←((2×⍴D)⍴ 1 0)\D
[17] oK:D←⍎D ⋄ (⍕⍴D),' numbers entered.' ⋄ F←0
[18] NXTf:→PAU×⍳Nf←F+1 ⋄ V←'/',(1↓(⎕TCNL,KEY)SSINDEX F),'/1/0'
[19] VALS←D RECODE 1↓V SSINDEX 2
[20] (1↓V SSINDEX 1),'=',⍕+/(⍎1↓V SSINDEX 3)×VALS+
(⍎1↓V SSINDEX 4)
[21] →NXTf
[22] PAU:→LOOP×⍳INT ⋄ →NXTs
∇
Lines 1-6 are comments that describe essential features of the program.
The ⎕POKE on line 7 discards the character prompt utilized in the interactive mode.
Line 8 displays the scoring key(s) and determines the number of factors (i.e. the number of major segments) in KEY. The results of this line are then stored in Nf, the loop factor counter.
Line 9 skips to LOOP (i.e. the interactive mode) if DATA does not exist. The program then reads data interactively (lines 14-16).
Line 10 catenates a carriage return to DATA, and calls sSSHAPE to determine the shape (i.e. the number of subjects) in DATA. It stores the result in Ns and initialises the subject counter S.
sSSHAPE, the function used to count the number of arguments in the segmented strings, returns the number of carriage returns present in the string. A carriage return is thus added to KEY (and also to DATA) so that the first element is included in the count.
Similarly sSINDEX returns selected delimited segments from a longer string, the delimiter for the segmented string is the first character (i.e. carriage return).
R←SS sSINDEX I
--------------
[1] ⍝Function for the indexing of segmented strings
[2] R←((+\SS=1↑SS)∊I)/SS
∇
Line 11 of SCORE increments the subject counter and exits when all the data has been processed. Otherwise the next subject is selected and the data stored in D.
Line 16 checks if there are blanks in D. If not, spaces are automatically inserted between the single digit integers; this facilitates entry of single digit data.
Line 18 goes to PAU when all of the factors have been calculated. This line also catenates the default multiplier, 1, and the default adder, 0, to the segmented string, and stores the result in V. These default options eliminate the need for user input of values when items are added with equal weight.
Line 19 passes the 'recode' subsegment to the RECODE function. If this subsegment is not empty, the data is recoded:
R←D RECODE C;T;NR;N;V;M;I
-------------------------
[1] ⍝ Recode data D_from S e.g 1:1=4 3=7$3:1=4 2=9
[2] ⍝ IF D→1 2 2 ¯4 then R→4 2 9 ¯4 [2nd value not recoded]
[3] →rEC×⍳':'∊C ⋄ R←(×T)×D[|T←⍎C] ⋄ →0
[4] rEC:Nr←sSSHAPE C←'$',C ⋄ N←0 ⋄ R←⍬
[5] L1:→0×⍳Nr←N+1 ⋄ V←1↓C sSINDEX N
[6] V[(V∊':=')/⍳⍴V]←' ' ⋄ I←1↑V←⍎V
[7] M←((0.5ׯ1+⍴V),2)⍴1↓|V
[8] R←R,(×I)×1↑((M[;1]∊D[|I])/M[;2]),100
[9] →L1
∇
Line 20 of SCORE computes the scale score by multiplying and then adding the possibly recoded values.
Line 21 returns to score the next factor. Finally if the program is being run interactively, line 22 goes to LOOP to query the user for additional data. Otherwise the program returns to nXT for the next subject.
Jerry M Brennan
University of Hawaii at Manoa
Department of Sociology, Porteus Hall
2424 Maile Way, Honolulu
Hawaii 96822
Appendix: Screen Hardcopy of Program Notes¶
To PRINT these NOTES, toggle CTRL-PRNTSC on or off.
The APL workspace name is TESTSCOR. Load this workspace with one of the following:
To SEE the KEYS & DATA in the workspace TESTSCOR, type:
SCORING KEY:¶
- To CREATE the scoring key for the test type )EDIT and some name - probably the name of the test.
For example:
- ENTER the scoring key for each factor on one line.
For example:
FACTOR NAME/ITEMS ON FACTOR/MULTIPLIER/AMOUNT TO ADD
FACTOR NAME/ITEMS/MULTIPLIER/AMOUNT TO ADD
...etc
WHEN YOU ARE FINISHED CREATING OR MODIFYING THIS KEY, LEAVE THE EDITOR WITH
Ctrl-E.
To SAVE the scoring key for use on another day, save the workspace with:
)SAVE
KEY: EXAMPLE 1: (Note that each factor begins on a new line)¶
This test has three factors labeled ONE, TWO, THREE, and FOUR.
The first factor, ONE, contains Items one and two. The multiplier for Item one is 1 and for Item two the multiplier is 3. Minus 2 is added to all Items. The second factor, TWO, contains three Items and all are multiplied by 1 and have 0 added. The third factor, THREE, contains Items four and five. Item four is added and Item five is subtracted.
Finally, factor FOUR has three Items: Two, five, and six. Items 2 and 5 are added and 6 is subtracted.
If Item 2 has a value of 1, it is scored 10; if it has a value of 2, it is scored 20. All other values of Item two are scored 0.
Similarly, if Item 5 has a value of 1, it is scored 30 otherwise, it is scored 0.
If Item 6 has a value of 1, it is scored 5. A value of 2 on Item 6 is scored seven. Otherwise, this item is scored 0. Note: Since Item 6 has a minus, these values are subtracted.
NOTE: ............':' delimits item numbers from their scoring weights. In the above example, a score of 1 equals 10 for Item 2. '¤' delimits items and their weights from each other. Also, the high minus is used to subtract items. This minus is Alt-2 on the keyboard and may look like a small two on your screen. For all computations, addition is done first, then multiplication and, finally, the sign is changed.
KEY: EXAMPLE 2:¶
EXAMPLE 2 has two factors: The first has two items and the second has three items. By default, nothing will be added to the items and there is no multiplier.
KEY: EXAMPLE 3:¶
This last example illustrates a scoring situation in which some items are REVERSED. The scoring for one REVERSED factor might be as follows:
Item 1 is punched 1, 2 ,3 and is to be scored 1, 2, 3.
Item 2 is punched 1, 2, 3 BUT is to be scored 3, 2, 1.
This might be keyed as follows:
DATA ENTRY:¶
- CALL the program with the following command:
SCORE SHORTMEMORY [The name of the program is always SCORE,
you insert the name of your key--in this case
SHORTMEMORY.]
The program will query for a "Subject ID". This identification can be any length and any characters--including blanks. After Subject ID is entered, the program queries for the items. Data consisting of all single digit numbers may be entered as a string (i.e., without blanks between each single digit). Data consisting of double (or more) digits must be entered with blanks separating each number. To QUIT data entry, strike RETURN when queried for "Subject ID".
Note: .............. Continue typing at the end of each line. The program will wrap automatically. The limit is 512 characters, including blanks.
- Another way to ENTER DATA is to use the editor as you did for the scoring keys. However, this time a variable will be created that contains the data.
For example:
Enter the Subjects' ID followed by a "/". Next, enter the data. (See above for information about data entry). Begin each new subject on a new line. For example, the data for two subjects might look like this:
When data entry is complete, quit with CRTL-E and save with )SAVE.
TEST SCORING:¶
- To RUN the scoring program (using the key and data created in the last two examples):
Many keys and subject data can be saved in this workspace. In fact, you can save over 500 separate variables.
- To ERASE a key or a data file, do the following: