New Data for Placement Procedures ANDREW H. MACPHAIL and LAURENCE S . FOSTER Brown University, Providence, Rhode Island
I
N 1939 we described the technic used at Brown University as an aid in the proper placement of students in beginning chemistry courses,' and pointed out that better placement resulted from the use of multiple correlations involving the scores on various aptitude aiteria. In that report, the Iowa Chemistry Training Test, the Mathematical Section of the Sones-Harry High-School Achievement Test; and the Brown University Psychological Examination were found to give the highest multiple correlations with the performance of students in the general college chemistry course. Recently, Brown has made some changes in the content of the battery of tests given to entering students just prior to the opening of college. The purpose of this note is to present some correlations, based on data from the new tests. In Table 1 are presented the three-variable multiple correlations, R's, of the 1940 study. The following conclusions have been drawn: The standing of the students in their high-school graduating classes, when properly weighted and combined with scores on the Iowa Chemistry Training Test, gives the highest predictive combiiation, hut the scores on the Iowa Chemistry Training Test, combined with either the scores on the Cooperative General Mathematics Test or the Q scores on the Psychological Examination of the American Council on Education, are almost as good. The L scores on the American Council Psychological Examination have no significant predictive value for general college chemis-
*.
TABLE I R's TO PBWIETGBNBULC B B . U S ~ Y GRADBS, 1940 STUDY
R 0.57 0.52 0.52 0.41
-
Variables (N , Gmeral chemistry grades High-Sehool Rank* Iowa Chemistry Traininz! Tesrt General chemistry gradu Iowa Chemistry Training ~estt* C&perative Gener.4 Mathematics Tutf General chemistry graden Iowa Chemirtrr Training Teott Q &ore (A.C.E.)I General chemistry grades Iawe Chemistry Training Test? L score (A.c.E.)~~
-
70)
* High-rehool rank on a rigma reale.
t Published by The Bureau of Educational Ruearch and Ssvice, Extension Division. The State Universitv of Iowa. Iowa Citv. .. Iowa.. Series C.T.-I. Revised B. f Published by Coap-tiye Test Service. 15 Amsterdam Avenue, N e w vnrt a t v .. .- r . .. , . $American Council on Educntioh Psychological Examination, published by the American Council on Education. 744 Jackson Place, Wsshington, D. C. il same as 9.
Obtained grades have been correlated with predicted grades with satisfactory results, as indicated by the fact that one regression equation used for predicting represented an R value of 0.57 and the r for predicted and obtained grades for 59 students was 0.60. In anMncPmn AND Fosmn. J. Casaa. Eouc., 16,270 (1939).
other case involving 80 students, the equation used for predicting represented an R value of 0.52 and the correlation between predicted and obtained grades was 0.66. In addition to the three-variable data reported the following regression equation (R = 0.61) was computed for four variables:
+
General Chemistry Grade = 0.16 Iowa Score 0.33 High-School Rankon a sigma scale 0.27 Co6pperative General Mathematics Scare f 23
+
For the same 80 students just cited, grades predicted by means of this formula correlated 0.76 with obtained grades. In the 1939 article we presented two examples of three-variable charts based on regression equations. Here we show a chart based on the above four-variable equation. Through its use predictions can be made without actually solving the equation for each student. For example, a student with a high-school rank of 50, an Iowa score of 160, and a mathematics score of 75 would have a predicted chemistry grade of "middle" B. Solving the eqnation gives 85. This set of values is read on the chart as follows: Start with a high-school rank of 50 on the scale at the lower right of the chart, then read horizontally to the left as far as the diagonal line marked 160 on the Iowa scale at the lower left of the chart; from that point read straight up to the point directly opposite the value of 75 on the mathematics scale at the upper left of the chart. The point thus reached falls in the "middle" B zone. .
.
..
CHARTaon EREDICTINGCHEMISTRY 3 Gnms From Scores on Wperative Mathematics Test and Iowa Chemistry Training Test and HighSchool Rank (in sigma).