Students often become disinterested in chemistry when the course involves only their passive note-taking (1). An increase in positive attitudes often results when there is a move away from the lecture, toward a group-work format (e.g., 2, 3). This study originated from discussions between the authors, a veteran instructor who had more than 30 years of experience (MWR), and a chemical educator (VMW). Although the instructor was not aware of educational literature indicating that cooperative learning can lead to statistically significant improvement in academic achievement (4–12), he was convinced that the lecture mode was not achieving the aims of the course and sought a better means of achieving those ends.

His main goals were that the students learn to think about the problems they encounter in quantitative analysis, rather than to memorize solutions, and willingly seek assistance when their thinking does not lead to the problem solution. Over the last 15 years, the instructor had tried a number of techniques involving cooperative learning but wanted more than anecdotal evidence for his decision to abandon traditional lectures. The study looks at a complete replacement of lectures with group problem-solving sessions in a junior-level quantitative analysis class for non-chemistry majors.

the treatment and control sections had reading assignments to complete before each class period. In the treatment section, students were assigned to heterogeneous cooperative groups of four (to encompass a variation in TOLT scores and in gender) and were presented with problems on an overhead projector or in a handout. The students were asked to solve the problems with a full discussion within their group. After one member of the group solved the problem, the group was required to discuss the problem until all members understood it. After all the groups

had finished, a member of the group was randomly selected to begin the solution on the board or the overhead projector or to direct the instructor through the steps to solve the problem.

Sometimes members of a single group completed the solution; other times a member of one group did one part of the solution and a member of another group finished the problem. The rest of the class was encouraged to find flaws and patterns in the reasoning. Problems used in the control and treatment sections were equivalent in content and number. In the event that the treatment section didn’t finish all the problems in one class period, they were asked to finish them as homework before the next class.

Objectives of the Study

The following research questions were investigated.

1. Do students who attend group problem-solving sessions in place of lectures in quantitative analysis learn the

content as well as students in a traditional class?

2. Do the attitudes and dropout rates of quantitative analysis students who attend group problem-solving sessions differ from those of students who attend the traditional lecture? Subjects

The subjects were students enrolled in two sections of a junior-level quantitative analysis course at a major southwestern university. Both sections were taught by MWR. The sections met on the same days of the week at slightly different times. One section was randomly selected to be the treatment section.

### Treatment

The treatment consisted of replacing a traditional lecture with cooperative group problem-solving throughout the

semester. The control section received traditional lectures on the content, during which the instructor worked on sample problems and explained the theory behind each step. Both the treatment and control sections had reading assignments to complete before each class period. In the treatment section, students were assigned to heterogeneous cooperative groups of four (to encompass a variation in TOLT scores and in gender) and were presented

with problems on an overhead projector or in a handout.

The students were asked to solve the problems with the full discussion within their group. After one member of the group solved the problem, the group was required to discuss the problem until all members understood it. After all the groups had finished, a member of the group was randomly selected to begin the solution on the board or the overhead projector or to direct the instructor through the steps to solve the problem. Sometimes members of a single group completed the solution; other times a member of one group did one part of the solution and a member of another group finished the problem. The rest of the class was encouraged to find flaws and patterns in the reasoning. Problems used in the control and treatment sections were equivalent in content and number. In the event that the treatment section didn’t finish all the problems in one class period, they were asked to finish them as homework before the next class.

### Data Collected

Both qualitative and quantitative data were collected. The subjects were given the Test of Logical Thinking (TOLT)

on the first day of class. The TOLT measures the following reasoning abilities: controlling variables, proportional reasoning, combinatorial reasoning, probabilistic reasoning, and correlational reasoning (13). TOLT scores can range from 0 to 10. The scores from the TOLT were used to construct the heterogeneous student groups. To measure content knowledge, exams, quizzes, and the course average were used. Both sections took three common exams given at night: exam 1, exam 2, and the final exam. To ensure that students were prepared for the daily classroom

discussions, frequent “daily” quizzes on the reading assignments and on classroom activities were given to each section.

Quiz questions were similar to those on the hour examinations, but were graded less rigorously. Course averages were calculated using the formula [E1 + E2 + (2 × Final) + (0.5 × sum of quizzes)]/4.5. A 7-point Likert-scaled attitude survey similar to those used in other research studies (14, 15) was constructed. The seven-point scale permitted the desired degree of variation in responses. This 31-item instrument was used at the end of the semester. It asked students about their perception of how they understood the chemistry content, their attitude toward chemistry, and their attitude toward the teaching methods used in their class. (See sample items in the results

section.) Additionally, students were asked to write about what they liked most and least about the course. The full

attitudinal questionnaire can be found online. W VMW observed both sections and made field notes concerning student behaviors, interactions, etc. The treatment-section observations comprised notes on the class as a whole

and on two specific groups.

Notes about the groups included dialogue and interaction of group members. Interviews held during the semester with various group members consisted of both open-ended and leading questions, which are described later. Students completed an open-response midterm evaluation. MWR also kept notes about his impressions throughout the progress of the course. Only students who completed all exams, the midterm evaluation, the end-of-term attitude survey, and the TOLT were included in the quantitative analysis. This restriction left 13 of the 27 students in the control section and 32 of the 52 in the treatment section. Other reasons for the larger number

of subjects in the treatment section will be discussed later.

### Results and Discussion

No content measure (quizzes, exams 1 and 2, the final exam, and the course average) using multiple analyses of

variance showed significant differences between sections at the p < .05 level when the students from whom all data were collected were considered (Table 1). There was also no significant difference in the TOLT scores of the sections (p = .81). Since the reasoning ability of the sections did not differ, we treated them as similar groups. Even using the TOLT score as a covariant showed no differences in the content scores between the sections when reasoning ability was controlled. The course average was 74.38% (SD 17.24) for the lecture section and 77.53% (SD 3.55) for the treatment section; the difference was not significant. Grade cutoffs were the same for both sections: A = 89; B = 76; C = 65; D = 45; F < 45. Figure 1 shows the letter grades for the subjects who completed the course and were included in the original study. The proportion of A’s, B’s, and C’s was 76.9% in the control section (n = 13) and 87.5% in the treatment section (n = 32).

A more interesting finding is the grades for all those who enrolled in the course (27 students in the control and 52 in the treatment section) (Fig. 2). The proportion of A’s, B’s, and C’s was 48.1% of those enrolled in the control section and 67.3% of those enrolled in the treatment section. The withdrawal rates were very different for the two sections (33.3% for the control section and 17.3% for the treatment). It should be noted that the withdrawal system of “Q–drop” allows students only 3 such drops in their academic careers, but a Q–drop also removes all mention of the course from a transcript. Students hoard their Q–drops and are reluctant to use them, except in urgent cases. A χ2 analysis of the Q and I scores versus the A–F grades for each section showed a significant difference between the sections (p = .040). The rationale for combining the Q and I rates is that the I’s are students who quit participating in the class but did not officially drop. Since most institutions report grade distributions as percentages, the difference between a 33.3% and a 17.3% withdrawal rate may be a better way of reporting this finding.

Two of the 31 items on the attitude survey showed significant differences at the p < .05 level when analyzed with multiple analyses of variance (see Table 2). The findings on the first item might reflect a desire of the control section to discuss more since there was little if any, discussion in this section. For the treatment section, the mean on this item is skewed toward a major component of their treatment, learning by listening to another student’s viewpoint. The means for the second item reflect the method by which students were asked to learn in this study. While in some ways this finding isn’t surprising, the factors in the difference between the two sections should be investigated.