In addition to the more flexible aradina scheme, there is also provided a means of scoring muGiple forms of the same examination. The use of multide forms makes cheating more difficult in crowded examination rooms. Multiple for& may he developed by scrambling the question order and/or scrambling the order of choices for any given question. The program can he used to process such multiple form examinations in an automatic manner with no manual intervention hy the instructor. A duplicate summary of each student's performance is produced by means of the program. One summary is given to the student while the other is kept by the instructor for reference. This record, as shown in Figure 4, contains the student's resnonse to each ouestion. the correct resnonse to each &stion,'individual question scores, and an o v k r a ~total ~ for the student. Findlv a statistical summarv of each form of the examination is printed for the instructor. Since the original scanner sheets are never returned to the students, any questions concerning grading can be quickly and easily resolved with erasures not beina susuect. The program can hevsed in conjunction with others or as a "stand alone". At the Universitv of Maryland, it is used to provide input to our standard gradehook program, GRADE. For a numher of years the faculty of the Chemistry Department a t the University of Maryland, College Park, have k e d this program for all lower-division coursesand for some high enrollment upper-division courses. The development of this central record-keeping facility has been an aid in solving some of the problems associated with teaching and managing large multisectional courses. Instructors are relieved of the tedious computational chores associated with high enrollment courses. Students have available a c o m ~ l e t erecord of their test-item status; errors, omissions and deficiencies are easily identified and corrected. From the departmental viewpoint, a single permanent record, in a standardized form, can he maintained. Such a record has proven useful in adjudicating grading disputes after a course is finished. In addition, since archive records are maintained in machine-readable form, the archives representing a numher of semesters and courses can he used to answer cluestions concernina the effectiveness of hensive-test recording and analysis program. ~ h r o u g ha mechanism of control cards it is capable of recording, editing, recoding, weighting, and scaling test scores in five item groups; examinations, quizzes, a final examination, lahoratory evaluations, and a suhjective evaluation. The lahoratory section and entire class information can he analyzed in a variety of ways. The information is then presented in aformat suitable for faculty/assistant use and also in a form suitable for viewing by students. Information can he represented as histograms and in tabular form. Students can he "dropped from the statistical analvses and students can also he flamed -- for "monitoring" purposes. In addition, an archive summary of each student's ~erformancecan he woduced in a form suitable for semester tb semester analysis- For security reasons the gradehook program is designed as a card-oriented system. Yet, it is possible to use GRADE in an interactive mode with the appropriate information stored in adequately protected files. The GRADE program has been run on UNIVAC 1100 series machines hut should he transportable to any machine with an ANSI FORTRAN compiler. The program will deal with an arbitrary numher of students-it expands to fit the class. The maximum numher of scores that can he recorded is thirteen quizzes, twelve labs, six exams, one final and one evaluation. The program runs in 12,000 thirty-six hit words olus the mace reauired to store student records. The records
............................ E X I I H I N I T I O N TVO CHEY 2 1 3 dAR3YLTTER
: ( 14.0/ SCORE 222222222
LUKE
FORH:
1
ITEL(
SCORE PERCENT
1 2
.o
.O .O
3
1.0
4 5 6
2.0
50.0 100.0 20.0 100.0
.o 1.0 5.0
7 8
2.0 40.0 3.0 100.0 SCORE PERCENT
IlEH
COHNENT:
LETTER GRADE.
Figure 4. A typical
RESPONSES A 0 C D
32.0) = 43.7% SECTION 114 ANSWER KEY
E
l
B
C
+
+
+
A
B
C
D
E
D
E
+
+
+
+ A
S
D
C
E
0
Student Examination
Summary
prepared
by program
OPTSCAN.
GRADE, CN-24, are available in machine independent form. Both programs are written in RATMAC, which is a free-form structured FORTRAN with a MACRO capability. Programs written in RATMAC are translated into standard FORTRAN by a pre-processor that is itself availahle in machine-independent ANSI FORTRAN. Copies of the OPTSCAN and GRADE programs together with documentation are available uDon reauest from the Universitv Proeram Lihrarv. Pavment oE $100 kill he required for the GRADE-OPTSCAN program oackaee. This charee will he used to cover the cost of the , R I I ~conipurer rim?. l'he rnlw rnamltt1c t a p , I I ~ I I I I X ~maouals. ntII he a 11inerrwk mmnvlic tux. in AS('11 code, LHE('L XII. RLKSIZ-3600,1600 BPI, with three files. Inquiries should he addressed and checks made payable to: Program Library, Computer Science Center, University of Maryland, College Park. Marvland 20742. The L I Wd t l w prooram requiresa ihrer scnnnrr S I K ~1 5 t h ~ Y ~ t i o m I( ' i n n ~ u r S;;\.srtm> t~ 'l'r:~n.optic x.iutner. Thv .oftware of the scanner must produce coded file or tape for communication with the machine using the OPTSCAN program. The OPTSCAN program has heen run on UNIVAC 1100 series machines but should he transportable to any machine with an ANSI FORTRAN compiler. The distributed program will grade 84 questions and three different examination forms. It is designed to expand to accommodate the size of the class being graded. The basic OPTSCAN runs in 26,000 thirtv-six bit words on the UNIVAC. A discussion of the advantages of writing scientific software in a language such as RATMAC is hevond the scoue of this oaoer (9-12). ~~~
a
Teaching. The support and encouragement of Dr. Shoenherg, Dean of Undergraduate Studies, is gratefully acknowledged. Appreciation is expressed to Professor E. C. Lingafelter for drawing attention to the need for a multiple choice grading program capahle of dealing with more than one response for each question. The computer time for this project was supported through the facilities of the Computer Science Center of the University of Maryland.
KEMGAM: A Chemical Adventure Richard T. Carman 7 Roda Drive Mastic. NY 11950
This program is a unique way of introducing students to reactions of some common laboratorv substances. The orogram can he utilized by students of general science, first-year chemistrv. and colleee chemistrv. In the program rhe atutlrn~chrr,ar; six wl,~titncts irtm a list I ~ E I P V C I I . S ht. ntudt i ~ l wI Iioms~~ the nlnount d e n c h suhVolume 58
Number 9
September 1981
695
stance. These substances are to be taken hy the student on an imaginary trip. As slhe nroeresses throueh the imaeinarv . " . lauds. s h e is confronted with situations that involve imaginary dangers. The student is to escane the danger hv huruine one substance or by mixing twq of thkm. In either case, s h e must specify the amounts to he used. If slhe chooses a solution that is ineffective, then slhe remains in that situation and tries again. In each situation the optimum solution earns the most points. Ineffective solutions, wrong answers and careless errors will cause points to he lost. The program is written in standard BASIC and has approximately 400 statements. The program was originally written for the DEC system PDP-20 where it requires about 20 blocks of memory (2984 8 hit words). The program listing comes with an explanation of how to expand it by adding more lands and situations, if desired. The oroeram as written would also function on the TRS-80(16K). To obtain a program listing, along with a line hy line commentarv and an exnlanation on expanding the program, send your name and address along witha checkor money order for $10 Cmade out to Richard Carman) to R. Carman a t the address ahove.
Non-Linear Least-Squares Optimization of Parameters for First-Order Kinetics on Programmable Calculators Dornenico De Marco, Athos Bellorno, and Alessandro De Robertis Instilute of Analytical Chemistry University of Messina Messina, Italy In first-order or pseudo-first-order kinetics the value of an experimental variable y that is linearly related to concentration is given by y = y-
+ Aye-k'
where y- = limy,
Ay = y o - y - ,
y o = limy
I-m
i -0
and h is the rate constant, and t is time. We describe here a non-linear least-squares program for optimization of k , Ay, and y, on jy, t J experimental data. Two equivalent listings of the program are available from the authors. One is for the Hewlett Packard HP 9825 desk computer equipped with 9821112 Matrix ROM. The other is for the Texas Instruments TI59 Programmable Calculator equipped with 1014976-1 Master Library Module 1. The least-squares sections are identical for hoth machines, but there are differences in data input and in calculation of starting values. On the basis of exoerimental values. the proerams selfdefine y , Ay and k siarting values and kini&ze"the sum of the squares of the differences between calculated and experimental values of y by Newton's method ( 1 3 ) .At the end of each iteration the user must decide whether to stop or continue the fitting procedure. When convergence is reached (h, y - , and Ay the same in successive iterations), the user must transfer program control to a second section where standard deviations in h, y - , and Ay, as well as other measures of goodness of fit are calculated and output. Program listings and directions for use are available from the authors at the address given ahove. 696
Journal of Chemical Education
Commentary on Countercurrent Distribution Program Distribution Curves (Chromatograms) for Many Stages George Vogel Boston College Chestnut Hill, MA 02167 David Holdsworth's note on countercurrent distrihution (14) in Bits and Pieces, 2, invites several comments. First, no
separate expression (and no corresponding, much longer program) is required when the volumes of moving and stationary phases are different. Instead, one simply replaces K , the distribution coefficient, by K', the distrihution ratio KVJV,. (It is clear that, e.g., doubling the volume of the mobile nhase must have the same auantitative effect as doubling K:) Mv second comment is that, althoueh the ex~ressioneiven for calculating the amounts in each fkction isberfect~icorrect. in wactice it cannot be used when the numher of stages (n) kxceeds 69, because 70! already overflows virtually any calculator. Even though Holdsworth's second expression avoids explicit calculatkn of factorials, for n > 69 overflow will occur euentuallv (unless the calculation is stopped .~ nre. niaturel! or #~111lti{~li~~~t1;o1i I.< alternattd with d ~ ~ i . ~ i
