Watkins SOR Instrukcja Użytkownika Pobierz pdf (Strona 3)

Journal of College Science Teaching

philosophy (Bigelow, Butchart, &

Handeld, 2006), and mathematics

(Miller, Santana-Vega, & Terrell,

2006).

Methods

In this paper, we examine data from

students in the rst semester of an

introductory calculus-based phys-

ics course that is two semesters long

for nonmajors at Harvard University

during the period 1990–1996. Each

year, the course met for 1.5 hours,

twice a week, in a large lecture hall,

while smaller sections (15–20 stu-

dents) were led by teaching assis-

tants once a week for 1–2 hours. In

1990, the course was taught using

traditional lecture-based methods.

In 1991, PI was introduced during

the twice-weekly, whole-class meet-

ings. Also at this time, reading quiz-

zes were implemented to encourage

students to read before class and,

in 1995, the courses began using

a research-based textbook devel-

oped by Mazur (2014). In all years,

the weekly homework consisted of

traditional quantitative problems.

Starting in 1991, exams included

conceptual problems as well as tra-

ditional quantitative problems to re-

inforce the importance of conceptual

understanding in learning physics.

As PI had not yet been widely dis-

seminated, the introductory physics

course was the only course in which

students were exposed to PI.

Students were asked at the begin-

ning of their introductory physics

courses to indicate their major (rst-

year students were asked to indicate

in which subject they planned to

major). We linked these data to stu-

dents’ majors recorded at graduation.

We then analyzed the relationship

between pedagogy and the fraction

of the students who initially indicate

that they intend to major in a STEM

discipline and then later switch to a

non-STEM major.

Our study sample included 105

students in the traditionally taught

1990 course, 101 of whom indicated a

STEM major at the start of the course.

There were 1,072 PI-taught students

in our sample; 997 indicated they

were majoring in or intended to major

in a STEM discipline. No students

in either the traditional course or the

PI courses indicated that they were

majoring in physics.

In our analysis, we used a chi-

square test to compare the proportions

of students switching out of STEM

majors in the traditional lecture-based

course and those courses using PI. We

then controlled for differences in stu-

dents’ background and demographics

using regression analysis. However,

we could not use linear regression,

because our dependent variable could

only take on the values of 0 or 1—stu-

dents either stay in a STEM major or

switch out of it. Instead, we used lo-

gistic regression analysis (Hosmer &

Lemeshow, 2000), which uses the log

of the odds of switching out of STEM

majors as the dependent variable

(

, where P is the prob-

ability of switching out of a STEM

major), as this transformation allows

for a linear relationship with the inde-

pendent variables. From this analysis

we obtained the estimated probability

of switching out of a major, given

different background characteristics

of students and whether they took the

traditional or PI courses.

Results

Table 1 shows the percentage of

students who switch out of a STEM

major, separated by course pedago-

gy. The proportion of students who

were enrolled in the traditionally

taught introductory physics course

and switched out of a STEM major

is more than twice that of students

enrolled in the courses taught using

PI (χ

= 5.1, p = .02). Furthermore,

the impact of pedagogy on STEM

major retention is consistent across

both genders.

Figure 2 shows the uctuations in

the percentage of students switching

out of STEM majors from year to

year. Compared with 1990, when

the course was traditionally taught,

the percentage of students switching

out of STEM majors after taking the

PI course is more than 50% smaller.

The gure suggests that the results

from 1990 are not simply due to

yearly uctuations, but also point to

the need to account for the groupings

of students by year in our regres-

sion models. By using multilevel

modeling in the logistic regression

analyses, we take into account the

random variability between courses

in addition to the variability between

individual students.

In Figure 3 we graphically repre-

sent the results from logistic regres-

sion analysis by plotting the tted

probabilities of switching out of a

STEM major. As the gure shows,

when we control for pedagogy and

SAT math scores, the odds of fresh-

men switching out of a STEM major

are predicted to be about 10 times

those of upperclassmen (p < .001).

Furthermore, students with higher

SAT math scores are less likely to

TABLE 1

Percentage of students who switch out of STEM majors, by pedagogy

and by gender (N

trad

= 101; N

= 997).

Instruction Total Male Female

Traditional 0.11 0.11 0.10

PI 0.05 0.06 0.05

Note: PI = Peer Instruction.

1 2 3 4 5 6

Komentarze do niniejszej Instrukcji

Brak uwag

Watkins SOR Instrukcja Użytkownika Strona 3

Komentarze do niniejszej Instrukcji

Powiązane produkty i podręczniki dla Nie Watkins SOR