Subjects and Experimental Design
All students enrolled in PY-261: Learning at Rollins College starting from the fall term of 1982 and including those enrolled during the fall term of 1993 served as subjects for this investigation. The total enrollment across the 17 individual course offerings was 357 students. Eighty-eight students chose to participate in the contract. The pre-contract performance of these students establishes a baseline by which effects of contracting may be compared for each individual subject, thereby establishing subjects as their own control. The remaining 269 students who chose to continue the course as previously structured (the non-contract group) also serve as a control for group comparisons of any effects which might have been experienced by all students, but only during the contracting period of the course.
 

Data Analyses
Results from this investigation were explored from several alternative perspectives. First, we compared average mid- term (pre-contract) letter grade levels to average final letter grade levels for both groups. Second, we evaluated student performance /grade profiles with respect to students who chose to contract vs. those who did not contract. An analysis of performance, in the form of mid-term grades vs. final grades was also conducted individually for those students contracting vs. those not contracting. A comparison was subsequently made of point percentage performances (as opposed to letter grades) for total performance measures and test performance measures on each group to adjust for potential effects of unequal point-to-letter grade conversion scales. Finally, we evaluated term-by-term pre-course vs. post-course GPAs to ascertain whether the course seemed to have any generalization effects. Most forms of analysis were submitted to simple t-tests with unequal n's to evaluate differences between means for each group. Alpha levels obtained are reported for each test, and a maximum of .05 was the criterion for significance/non-significance.
 
 

I. Letter Grade Differences from Mid-Term to End of the Course. Differences were investigated between contracting and non-contracting groups with respect to mid-term letter grades, based on all possible grading points (including tests, laboratory reports, and written assignments) before contracts were offered and the final course letter grade, based on all available data. Such an analysis addresses the degree of change from mid-term to final grade for each of the two groups, ignoring the fact that the contracting group attained their final grades through different evaluative criteria.

To address this question, all possible variations in letter grades were converted into numeric equivalents using an 11 point scale (with F = 0, D- = 1, D = 2, D+ = 3, C- = 4, C = 5, C+ = 6, B- = 7, B = 8, B+ = 9, A- = 10, and A = 11). Results demonstrate that students who did not contract improved their mean grades from mid-term to final grade by .5 grade scale points, while students who contracted improved by an average of 3.4 grade scale points. Students who contracted thus showed a significantly greater increase in grade performance, t (355) = 18.07785, p < . 00 1, than students who did not contract.

II. Group Characteristics: Letter Grades at Mid-term In this analysis, we asked the question, "are students who contract different in their performance prior to contracting than those students who choose not to contract?" Analysis of letter grade levels at mid-term for both groups demonstrated a significant difference (4355) = 17.98, p < .001) between contracting and non-contracting students. On the 11 point scale, non-contracting students averaged 7.5 (which rounds to a B letter grade) and contracting students averaged 4.06 (a C-) at mid-term. The average grade value of contracting students represents the 15th percentile for the total student distribution within the course. Clearly, it generally was the low-performing students who chose to contract.

III. Group Characteristics: Letter Grade Differences at Course End. Having determined performance differences between contracting and non-contracting groups prior to the contract phase of the course, we subsequently addressed the question of remaining differences after the contract phase was completed (i.e., at the end of the course). While there is a potential ceiling effect for the B average non-contracting student in comparison to the C- average contracting student, there still seems to be sufficient room for non-contracting students to improve in performance.

Also, it is relevant to remind the reader at this point that any student contracting will either be likely to have his contracted grade reduced by averaging the (generally lower) mid-term grade with the (generally higher) contracted grade, or will have made up the first half's work to the contracted grade's prior performance criteria (i.e., if the student has contracted for an A, all prior performances must be raised to a 96% or better, a B to a 87% or better, and a C to a 76% or better performance level through retakes and/or rewrites.)

An analysis of final letter grade scales showed that a significant difference in performance remained by course end between the contracting and non-contracting groups, t(355) = -2.93, p < .01. Those who did not contract continued to outperform those who contracted, but the difference between groups decreased dramatically. Students who did not contract ended with a mean grade scale of 8.03 (compared to their mid-term average of 7.5, both of which translate into B letter grades), while those who contracted ended the course with a grade scale of 7.489 (which almost meets the 7.5 required value to round to a B, but which, in fact, rounds down to a 13- letter grade). This value compares to a mid-term average of 4.06 (a C-) for the contracting group.

IV. Analysis of Individual Changes for Contracting Students. A matrix illustrating how mid-term grade achievement (rows) related to final course grade achievement (columns) for each individual student in the contract group is offered in Table 2. Inspection of this table reveals that, of the 14 students beginning the contracting phase with a failing (F) letter grade, five ended the course by bringing their work up to C levels, three up to C+ levels, and six actually brought their failing grades up to B levels. The four students who began contracting with D- averages brought their grades up to C, and so on. This table also illustrates that, of the 88 students in the contract group, only 11 had no change in grade level (8 mid-term C students contracted for C final grades, and 3 mid-term B students contracted for final Bs). Further, only one contracting student ended the course with anything lower than a C grade, and that student dropped from a C to eventually fail the course. Put another way, this says that only one student ever failed the course under contract conditions.

A global inspection of Table 2 illustrates the distributional details of our earlier statement that the contract option appeals mostly to the lowest performing students (C and below) as defined at mid-term. This table also reveals that most of the lowest performing students improved by one-to-two letter grade levels by the end of the course. A relative few who were in the low B range accomplished A's by course end. No A level mid-term student ever chanced contracting to guarantee an A by course end.

Table 3 illustrates the same distributional relationships between mid-term grades and final course grades for individual non-contract students. As this table shows, only one student failed the course over this ten-year period under non- contracting conditions, and that student had a D+ at midterm. In contrast to the previously stated fact that most contracting students started with low mid-term grades, and that almost all contracting students dramatically improved their grades, Table 3 confirms that most (but not all) non-contracting students began with relatively high grades (C or better), but that many ended the course with grades lower than they had at mid-term. To be specific, 145 non-contracting students showed grade improvement, 79 stayed at the same grade level, and 45 dropped in grade level. The fact that nearly half of the non-contracting group either declined or stayed at the same grade level suggests that the success of the contract group as a whole is probably not due to the course becoming significantly easier in its demands.

The comparative extent of this individual improvement/ slippage effect is perhaps more easily appreciated by a graphic translation of these two tables. For example, Figure 1 represents the individual amount and direction of change from mid- term grade to final course grade by using a line to represent each student's grade level change between mid-term and final grade. In order to represent each individual's positive or negative grade changes in a more traditional upward/downward fashion, both the mid-term grades and final grades can be read on the ordinate. The abscissa is used to represent the mid- term grade groupings in order to facilitate understanding the amount of change within each grade range, and to easily locate the starting mid-term grade level with respect to the ordinate. Thus, for example, the C mid-term group may be easily located across the abscissa, and the beginning of each change line starts at the C level as depicted on the ordinate. Inspection of this group illustrates that one student dropped from a C to an F, eight students remained at the C level, two students climbed from a C to a C+, and all other students but one (who climbed to an A) climbed to a B from the C mid-term level. Overall, Figure I illustrates that almost all contracting students showed sizable increases in their grade levels between mid-term and final grading (which is reflected by the length of each line in Figure 1).

On the other hand, a similar graphic (Figure 2) depicting changes for non-contracting students illustrates that every midterm grade grouping included students who dropped from initial grade levels, with some doing so substantially. The other obvious difference between Figure I and Figure 2 is the size of positive changes between the groups (i.e., the lengths of the change lines in each). The non-contracting group had only moderate increases in grade levels when the change direction was positive, and many more students remained at the same grade from mid-term to final in comparison to contracting students.
 
 

V. Differences between Point Earnings. The fact that contracting students earned token points which are of a very different character than non-contracting students' points earned by testing and writing, and that token points are used for determining at least the last phase's letter grades for contract students, suggests the need to consider more equivalent point-based performances for comparing between groups, as well as for pre-to-post comparisons within the contract group.

Further, the use of letter grades for group comparisons may distort the differences between groups on another basis. Letter grades in the Learning course are not constructed from equal point ranges. The A range includes six percentage points, the B range includes seven, the C range includes fourteen, and the D range includes eight. For a student to move from a low D to a high B is thus quite a different jump than moving from, say, a high C to a low A, despite the fact that both involve a two grade level change.

A more accurate comparison could rely upon percentage of total available points directly, without translating them into grades. An analysis was thus performed using differences between the total percentage of points earned by each student divided by the total available, based on all possible performance measures combined (tests, laboratory reports, etc.). Students who contracted showed a significant increase in final percentage performance, increasing by an average of 14.86 points, while the non-contracting average increased only 3.2 percentage points. These differences between means are not only highly significant, t(355) = 32.52, p < .001, but also much more telling of the degree of improvement by contracting students. Contracting students improved over 4.6 times more than non-contracting students when compared on equivalent scales.

VI. Test Score Comparisons. Even the total percentage scores just described may be slightly confounded by the inclusion of outside writing assignments, because writings may reflect both intrinsic writing skills as well as the content knowledge of the student. However, test comparisons can eliminate this potential bias because all tests were constructed of the same questions and were graded by equivalent percentage scorings for all students, regardless of contracting status. Figure 3 illustrates a test-by-test comparison of group means for each group on each test (usually two test administrations per condition, but some course offerings used three). Students experiencing the contract show dramatic improvements in test performance during the contracting phase of the course, while non-contracting students remain much the same from pre-contract to contract conditions.

An analysis was performed on the combined percentage scores of all tests taken before contracting against comparable scores from tests taken during the contract phase. To assess the statistical qualities of this comparison, a difference score was calculated between these mid-term and final averages for each student within the non-contract group and the contract group. The non-contract group showed a mean improvement from mid-term to contract phase of 4.8 percentage points compared to a mean test performance improvement of 17.3 percentage points for the contract group. This difference is significant, 0355) = 36.181, p < .005. As it turns out, the contract group showed a greater improvement on test scores than on their combined percentage scores, which included writing assignments -- tests improved 17.3 and combined measures improved 14.9 percentage points.
 
 

VII Potential Generalization Effects. The fact that this contract investigation, in contrast to almost all of its published predecessors, stresses the development of student study and time management skills suggests the relevance of considering how contracting students fare in their academic careers following this experience. Perhaps the effect is highly localized and/or teaches students only to rely upon such formalized contracts for their successes. To address this question, an analysis was performed of subsequent academic transcripts for every individual student included in the study.

From these transcripts, the term GPA of each student was recorded for each term prior to taking the Learning course and for each term following the Learning course. A mean GPA was then established for each of the contract and non-contract groups for each term preceding and following the course. Because students vary greatly with respect to which term they enroll in the course, the size of the group determining the mean for each term also varies. Table 4 depicts the size of the group and the group's mean GPA for each term, and Figure 4 illustrates these data graphically for the five terms preceding and six terms succeeding enrollment in the course. The size of the group for any additional terms either preceding or following is so small (usually only one or two students) as to suggest serious questions of reliability for comparative purposes. Also, because the registrar's office calculates GPAs only on a 4.0 scale for external use, and because conversion to the 11 point scale used within the course would be extremely time consuming, these data are reported on the traditional 4.0 scale.

Contract students began with a GPA of 2.486 five terms prior to taking the Learning course. This compares to a 2.60 GPA for the non-contracting group. By the term immediately preceding the course, contracting students had improved to a 2.65 and non-contracting had a mean GPA of 2.96.

With respect to their respective mid-term averages within the Learning course, both groups were performing lower than their prior term's GPA, but contracting students were considerably lower than non-contracting students. Translating between the two scales using a percentage-of-scale measure, the non-contracting students' mean pre-term GPA of 2.96 is 74% of scale while their mid-term Learning average of 7.5 is 68% of scale. Contracting students' mean pre-term GPA of 2.65 is 66% of scale, while their mid-term Learning average is only 37% of scale. Thus contracting students were 29 percentage points below pre-term performance levels while non-contracting students were 6 percentage points below.

Following the Learning course, both groups continued to improve their GPAs, and in fact seemed to accelerate their rate of improvement. By the seventh or eighth term following the course, most have graduated, but in the sixth term following the course, the non-contract students are now attaining a term GPA of 3.597. Contracting students are attaining a 3.217 during this term. Both groups show a rather systematic trend upward, although the non-contracting students tend to increase their GPA advantage over contracting students.

Improvements in GPA for both groups following the course are not easy to interpret. Students beginning to concentrate in their majors after a few terms are well known for earning higher grades than when taking their first couple of major courses and their required electives. Nevertheless, the growth is there for both groups, and this is especially encouraging with respect to the contract group since they were largely at-risk students within the Learning course.