economy

Research Summaries: The purpose of this assignment is to introduce you to peer-reviewed academic literature.

Use Google Scholar to find a peer-reviewed journal article related to the topic we will be discussing over the next few weeks: Higher education.

In bullet point format, summarize the main points of the article in your own words to the best of your ability. If you desire, feel free to include your thoughts on the article in the final bullet points.

Format (an example is attached):

On the top line of the first page, insert the citation of your chosen paper in

APA reference style format

.

Bullet points shall be double-spaced from each other, but each bullet point shall be single-spaced. (Again, see attached example.)

12 pt., Times New Roman font.

This is a low-stakes assignment, but you are required to format your summary as instructed.

Suggestions:

· If you must do a research project/paper for another course, then note you are welcome to use this assignment to augment that effort.

· It is highly unlikely you will understand the entirety of 99% of peer-reviewed journal articles (that are worth reading, at least).

· Try using varieties of search terms. Google Scholar is sensitive to even the slightest variations in search terms. For example: Let’s say you want to find a paper on the effect of marijuana usage on college students’ academic performance. Searching “marijuana college students” returns 95,500 results and the top hit is “The residual cognitive effects of heavy marijuana use in college students”. Searching “marijuana university students” returns 112,000 results and the top hit is “Alcohol and drug use in UK university students”.

· Find an article title that seems interesting? I suggest you follow these steps:No more than two pages. Upload your summary on Blackboard.

· Okay, now read the abstract. If it fails to interest you, or seems too “mathy” or otherwise unintelligible, then keep searching. There are tens of thousands of articles on any topic you can fathom that you should find fascinating, somewhat intelligible, and relevant to your interests. You are significantly more likely to enjoy this assignment and learn exponentially more if you choose your articles with care.

· Note the number of citations–Google Scholar results report “Cited by ___” and automatically presents search results with the most highly cited papers up top. More citations mean the journal is highly respected, the article’s findings are very influential, and professors are more likely to recognize the journal/article/author(s)—undergraduates are forgiven for citing lame articles, but usually highly rewarded when professors realize you’re grappling with the highest caliber of respected research.

· Note the date of publication. Whether the date of publication matters depends on context. An article presenting statistics on marijuana use amongst college students conducted in 1985 is probably not relevant, but an article on the “influence of spirituality on substance abuse by college students” from 2001 is still very much relevant. Use common sense.

· If using Google Scholar, click “Cited by” and “Related articles” and quickly scan the top results. Very often you’ll find better articles published more recently. Note that newer articles usually have fewer citations, which makes sense. If a “Cited by” or more “Recent article” has more citations, that’s powerful signal it’s a much better and more influential paper. Check it out.However, the most important criteria I suggest for picking papers: Is it interesting—does it immediately appeal to your interests? Can you understand most of the abstract? Do the main findings sound important, and perhaps contrary to your current beliefs?

· You’ve chosen a paper. Now scan the article and make sure it’s not 90% equations that you’d need a PhD to comprehend and it isn’t ridiculously long. (Top journals usually limit papers to 40 pages—don’t freak out, that’s an upper bound and most papers have lots of graphs, tables, endnotes, etc.)

· Now, read the introduction, then skip to the conclusion. A quality paper gives you all important points in these two sections.

· Now, read the rest of the article to the best of your ability. On the first pass just read without stopping even if you don’t understand most of it. Scholars often rely on specialized methods only very few specialists can comprehend. Don’t sweat it, just plow through. Exposure to this portion of the paper is one of my main objectives for this assignment, which I will explain several times during class. Now, do something else to take your mind of the paper for a while. After at least a few hours, take another shot at reading the guts of the paper. You’ll be surprised to discover how much you can learn from high-level research it you have faith in yourself and use a disciplined approach.

· Download the paper.

· Now, open a Word document. Make sure it’s set to Times New Roman, 12-point font If you use any other font or style I will set you on fire.

· Click the “References” tab, select “APA”, click “Manage Sources”, then “New”, and then “Journal Article”. Fill out the fields using information from the article.

· At the top of the first page, insert your citation by clicking “Bibliography” and “Insert Bibliography”.

· In bullet point format and using your own words, summarize the article using the same order as the article. Underline one or two main bullet points that you think are the most interesting and/or important, please—this is for my benefit.

· The last bullet point is an opportunity for you to tell me your opinion of the article, its value (if any), and what you learned from the article and the assignment. Please be honest. I often use honest feedback to recalibrate assignments, and even the harshest criticism is welcome if it’s clearly articulated and supported.

American Economic Association

Grade Inflation and Course Choice
Author(s): Richard Sabot and John Wakeman-Linn
Source: The Journal of Economic Perspectives, Vol. 5, No. 1 (Winter, 1991), pp. 159-170
Published by: American Economic Association
Stable URL: http://www.jstor.org/stable/1942708
Accessed: 26-02-2018 13:45 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide

range of content in a trusted digital archive. We use information technology and tools to increase productivity and

facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

Terms and Conditions of Use

American Economic Association is collaborating with JSTOR to digitize, preserve and extend
access to The Journal of Economic Perspectives

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

Journal of Economic Perspectives- Volume 5, Number 1-Winter 1991-Pages 159-170

Grade Inflation and Course Choice

Richard Sabot and John Wakeman-Linn

T 4he number of students graduating from American colleges and univer-
sities who had majored in the sciences declined from 1970-71 to

1984-85, both as a proportion of the steadily growing total and in

absolute terms (U.S. Department of Education, 1987). This decline has
prompted forecasts of a nation of scientific illiterates and a loss of economic

competitiveness. The Director of the National Science Foundation, Ernest

Bloch, put it this way in a speech at Carleton College (July 13, 1988):

The nation depends upon undergraduate education to prepare not only

the small number of students who will become research scientists and

engineers, but also the many other students who will have to function

effectively in an increasingly technological world. That is a difficult and

very important task. … The college age population is shrinking. Declines

(in science enrollments) are inevitable unless the proportion of students
pursuing science and engineering increases-and there is little evidence

of that. Somehow, we must persuade more students to study science and
engineering.

Other trends in student course choice, like the rise in enrollments in “voca-

tional” courses, have also elicited concern. The most common response by
faculty and administration concerned with these patterns of demand has been
to tighten quantitative restrictions: distribution requirements have been altered
with the aim of bolstering enrollments in the sciences.

* Richard Sabot is Professor of Economics and John Wakeman-Linn is Assistant
Professor of Economics, both at Williams College, Williamstown, Massachusetts. Sabot is

also Senior Research Fellow at the International Food Policy Research Institute, Wash-
ington, D.C.

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

160 Journal of Economic Perspectives

However, faculty bear some responsibility for these patterns of student

choice which they bemoan. Students make their course choices in response to a

powerful set of incentives: grades. These incentives have been systematically

distorted by the grade inflation of the past 25 years. As a consequence of

inflation, many universities have split into high- and low-grading departments.

Economics, along with Chemistry and Math, tends to be low-grading. Art,

English, Philosophy, Psychology, and Political Science tend to be high-grading.

As a Yale senior, interviewed by The New York Times (1988) for an article on

honors and grade inflation, explained, “It’s pretty hard to get below a B – in

most humanities courses. I hear it’s a little different in science classes, though.

There aren’t really F’s anymore. People look at a C and think it’s an F.”

Another Yale student, who switched from a science major to English, said in a

Wall Street Journal (1990) story attributing science dropouts to low grades, “In

other classes, if you do the work, you’ll get an A. In science, it just doesn’t work

that way…”

One result of varying rates of grade inflation between departments is that

grades as a signal of relative strengths and weaknesses become more difficult

for students to interpret. Grades therefore contribute less to students’ assess-

ment of their comparative advantage. But even if grades fail completely to

perform this function-even if a high mark is less an indication of the student’s

strength than of the weakness of the instructor’s resolve-grades, or more

precisely the expectation of grades, are still likely to influence course choice. A

conflict exists between the incentives offered to students and the institutional

goal of increased science and math education.

This paper presents evidence from nine colleges and universities that

grade inflation has led to a divergence among departments in grading policies.

We then discuss the results of an econometric study we conducted at Williams

College of the influence of grading policies on course choice. The impact that

differences in grading policies across departments have on the distribution of

enrollments was also estimated, and policy implications of the findings are
discussed.

Evidence of Divergent Grading Policies

Grade inflation and a widening gap between low and high grading depart-

ments are a nationwide phenomenon. We begin with a close examination of
what has happened to grades at Williams College, and then compare grades at
Williams with those at a diverse group of colleges and universities.

Table 1 shows that grade inflation at Williams has been substantial. The

mean grade in the introductory courses of eight large departments at Williams
has risen from 2.49 on a 4-point scale (a bit above C + ) in 1962-63 to 2.93
(roughly B) in 1985-86; the proportion of students receiving less than B – has
fallen from 47 percent to 26 percent and the proportion receiving more than

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

Richard Sabot andJohn Wakeman-Linn 161

Table I

Mean Grades and Their Distributions in Introductory Courses

in Eight Departments, Williams College, 1962-3 and 1985-6

1962-3

Mean Standard

Grade Deviation % Above B + % Below B –

Departments:

Art 2.62 .7033 9% 32%
Economics 2.40 .9600 17% 49%
English 2.58 .6767 10% 48%
Math 2.09 1.063 9% 65%
Music 2.74 .6600 14% 37%
Philosophy 2.38 .7200 13% 46%
Poli. Sci. 2.43 .6967 8% 55%
Psychology 2.64 .7933 15% 44%

Aggregate Average 2.49 .7857 11.9% 47.0%
Standard Deviation .1916

1985-86

High Grading

Departments:

Art 3.00 .6500 23% 20%

English 3.13 .5467 25% 12%
Music 3.26 .5733 28% 17%
Philosophy 2.94 .6067 20% 17%
Poli. Sci. 3.10 .5300 17% 19%

Average 3.09 .5800 22.6% 17%
Std. Dev. .1105

Low Grading

Departments:

Economics 2.67 .7333 15% 42%
Math 2.61 1.0033 20% 44%

Psychology 2.71 .8733 17% 37%

Average 2.66 .8700 17.3% 41%
Std. Dev. .0411

Aggregate Average 2.93 .6887 20.6% 26%
Standard Deviation .2239

B + has risen from 11.9 percent to 20.6 percent over the same period. This

pattern is manifested by smaller departments as well.

More central to our concern is the variation in the pace of inflation. In

some departments the rate of inflation is high; in Political Science the mean

grade has risen by .67 and the proportion receiving below B – has fallen by

nearly two-thirds. In others, the increase has been modest; the mean grade has

risen only .27 in Economics, while the proportion receiving below B – declined

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

162 Journal of Economic Perspectives

by one-seventh. As a consequence of this differential grade inflation, the

variation of mean grades across departments has increased. In 1962-63, with

the exception of the unusually low grading Math department, there was little

difference across departments either in mean grades or in the distribution of

grades. After 25 years of grade inflation, the situation is markedly different.

Williams College now divides itself into low-grading departments, in which

the mean grade is 2.66 and 41 percent of students receive less than a B – , and

high grading departments, in which the mean grade is 3.09 and only 17

percent of students receive less than B – . This difference in means between

high- and low-grading departments is significant at the 1 percent level. Fur-

ther, in those departments in which the grade distribution shifted higher, given

the fixed limit on the highest possible grade, the distribution became more

compressed. Within a typical low-grading department, the dispersion of grades

is about the same as in 1962-63. But in the typical high grading department,

dispersion is much less. The correlation between mean grade and the standard

deviation of grades within departments is -.886, and is statistically significant.

The difference in grades between these two groups of departments cannot

be explained simply by a difference in the quality of students; there is no

significant difference between students in high- and low-grading departments

in either SAT scores or grades in other courses.

We compared the grades at Williams to grades at Amherst College, Duke

University, Hamilton College, Haverford College, Pomona College, the Univer-

sity of Michigan, the University of North Carolina and the University of

Wisconsin. This sample is admittedly small, but was selected so as to include

private and state schools, large universities and small colleges, and Eastern,

Southern, Midwestern and Western schools. We promised these schools that we

would publish the data only in the relatively anonymous way that it appears in

what follows.

Seven of these eight schools have experienced substantial grade inflation.

Table 2 provides data (unweighted averages) for these seven schools compara-

ble to that in Table 1. Grades were relatively low and very similar across

departments in 1962-63.’ In 1985-86, grades were higher and all seven
exhibited the same phenomenon that we have described at Williams: each

school is now divided into low- and high-grading departments. Averaging

across the schools, 32 percent of all students in high-grading departments
receive grades above B + , while in low-grading departments only 19 percent

do. Only 19 percent of students in high-grading departments receive grades
below B – , while 40 percent of students in low-grading departments do.

In four of the seven schools, interdepartmental differences in grading

policies are about equal to Williams, while in the others the differences are even

more marked than at Williams. In one case the proportion of students receiv-

‘We lack 1962-63 data for two of these schools. As at Williams, the Math and Chemistry
departments are an occasional exception.

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

Grade Inflation and Course Choice 163

Table 2

Mean Grades and Their Distributions in Introductory Courses
in Seven Other Colleges, 1962-63 and 1985-86

1962-63a

Mean Standard % Above B + % Below B –
Grade Deviation

Departments:

Art 2.45 .8312 11% 53%
Biology 2.22 1.045 11% 56%
Chemistry 2.19 1.032 12% 60%
Economics 2.23 .9421 11% 61%
English 2.30 .7517 5% 60%
Math 2.21 1.199 18% 57%
Music 2.67 .9123 20% 35%
Philosophy 2.48 .8302 11% 51%
Poli. Sci. 2.51 .7833 13% 50%
Psychology 2.59 .8142 20% 45%

Aggregate Average 2.38 .9141 13.4% 52.7%
Standard Deviation .1682

1985 -86 b
High Grading
Departments:

Art 2.95 .7223 299% 24%
English 3.12 .5437 27% 12%
Music 3.16 .6657 44% 21%
Philosophy 2.99 .6698 29% 21%
Poli. Sci. 2.95 .7115 24% 23%
Psychology 3.02 .6879 28% 23%

Average 3.03 .6668 30.2% 20.8%
Std. Dev. .0809

Low Grading
Departments:

Chemistry 2.66 .9847 17% 44%
Economics 2.81 .8905 20% 31%
Math 2.53 1.042 22% 46%

Average 2.67 .9722 19.9% 40.3%
Std. Dev. .1126

Aggregate Average 2.91 .7686 26.8% 27.3%
Standard Deviation .1936

aThe data are not all from the same semesters. For one school the data is from 1962-63, for two
the data is from Fall 1962, for one it is from Fall 1963 and for one it is from Fall 1969. We lack
1960s data for two schools.

bFor four schools the data is from 1985-86 and for three schools the data is from Fall 1985.

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

164 Journal of Economic Perspectives

ing below B – ranges from 9 percent to 60 percent; in another the share below

B – ranges from 0 percent to 39 percent, while the proportion above B +

ranges from 10 percent to 57 percent. In addition, the departments that grade

low and high display a consistent pattern. Economics, Chemistry, and Math are

consistently low-grading departments, while Art, English, Music, Philosophy,

Psychology and Political Science are almost always high-grading departments.

Only Biology varies, being a low-grading department in just over half the cases.

The one school in our sample which did not experience grade inflation is

the exception that proves the rule. At this school, mean grades rose only .022

on a 4-point scale. It is also the one school that has managed to maintain

uniform grades; in 1985-86, the standard deviation of mean grades across

departments was only .1508, which is actually less than it was in 1962-63.

Some Intuition about Course Choice

An individual’s choice among courses can be viewed in a utility maximizing

framework.2 There are two aspects of course choice as it pertains to utility

maximization; one involves the intrinsic and extrinsic satisfaction derived from

taking the course and from the grade received. The other involves students’

knowledge of their learning abilities.

Learning can be intrinsically satisfying. Or the course may not be much fun

in itself but still deemed useful; organic chemistry and microeconomics are

often considered a pain worth tolerating because they lead to future courses or

careers that are expected to be highly pleasurable or profitable. In the same

way, good grades yield intrinsic satisfaction (the A that brings the warm glow of

achievement) and extrinsic satisfaction (Dean’s list, good jobs, and graduate

scholarships). Bad grades, of course, may result in disappointment, restrictions

on participation in sports, academic probation, and parental disapproval.
In addition to entering the student’s utility function directly, grades have

an indirect influence as signals of the student’s strengths. Students do not

typically know which subjects they learn most efficiently. Grades signal to
students their relative strengths and weaknesses. They can be an integral part

of the educational process, a feedback mechanism which helps the student
define her comparative advantage and choose courses on that basis. They can

reveal whether a student is good at Physics or English, poor at History or

Economics.

If grading policies are uniform across departments, maximizing grades

and exploiting comparative advantage are mutually consistent; by choosing
those subjects in which she is good, a student will both learn more and get
better grades. Departure from a uniform grading policy doesn’t necessarily

2The analysis of this section is based on the activity choice model (Winston, 1982), which is an
extension of the familiar Becker (1965) analysis.

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

Richard Sabot and John Wakeman-Linn 165

obscure signals to students of their comparative advantage or alter course

choice. A student who receives a higher grade in English than in Economics,
but whose higher English grade is lower relative to her classmates than is her
Economics grade, may correctly conclude that her comparative advantage is in

Economics. She may then go with her relative strength, and choose a second

course in Economics over a second course in English. Or she may choose a

second course in the subject in which she is an inefficient learner, English, over

a second course in Economics, because of the expected consequences of that

choice for her grade point average. So, whether course choice is influenced by

differences across departments in grading policy depends on the weight stu-

dents give to grades as signals of comparative advantage, relative to the weight

they give to grades as rewards.

Course Choice at Williams College

To investigate how grades affect student course choices, we studied a

representative sample of 376 students enrolled at Williams College during the

academic year 1985-86. Our data on these students were from three sources:

student transcripts, student files including application forms, and a survey

which we administered to these students which yielded a measure of the

student’s “need for achievement” (Gough, 1952). For each student, we had

data on all courses taken and grades received through June 1987: 6842 total

course choices. In addition, we had demographic and family background data,

indicators of abilities, cognitive skills and academic performance prior to

enrolling at Williams, as well as indicators of course preference and academic

motivation. This panel data set is rich, but not unique. Similar data are to be

found in the Registrars’ Offices of all colleges and universities. While this study

represents one of the first attempts by economists to exploit these data,

replication and extension of this work at other institutions should be quite

simple.

We chose for our analysis, from our sample, the five departments with the

largest enrollments in their introductory course: Economics, English, Math,

Political Science and Psychology. We used probit functions to measure the

influence of the grade received by a student on the probability of that student

taking a second course in the same department.3
In the two departments with the greatest number of observations-

Economics, a low-grading department, and English, a high-grading depart-

3We derived maximum likelihood estimates of the parameters in the reduced form equation,
Prob(Y = 1) = D(X’B), where Y is a dichotomous variable which takes the value 1 when the
individual has taken a second course in the discipline and 0 when she has not, X is a vector of
exogenous variables (discussed below), and D(X’B) is the cumulative normal distribution function.
We focussed on the decision to take one more course, rather than on the number of courses taken
(or the choice of major), because the decision to take a third course (or to major in a department)
depends on the grade received in the second and subsequent course(s).

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

166 Journal of Economic Perspectives

ment-we also assessed the independent influence on course choice of a

measure of comparative advantage. Our hypothesis was that in assessing com-

parative advantage, students will deflate the grades they receive in high-grad-

ing departments. Thus, controlling for the absolute level of the grade received

in Economics 101, students should be more likely to take additional economics

courses the higher is their rank in the distribution of Economics grades relative

to their rank in other courses.

When measuring the influence of grades we controlled for such other

influences on course choice as intrinsic interest in the discipline, beliefs con-

cerning the level of rewards associated with different disciplines, prior evidence

of comparative advantage, and the student’s need for achievement. To do this,

we included a variable which reflected whether the student intended to major

in that department, a gender dummy and a measure of the student’s need for

achievement, as well as the grade received. Details of the empirical work and

our results can be found in Sabot and Wakeman-Linn (1988).

We first estimated course choice functions that did not include measures of

comparative advantage. F-tests for four of the five equations are significant at

the 1 percent level. We found that in Economics, English and Math, the

probability of taking a second course declines significantly as the grade the

student received declines.4 Political Science and Psychology also fit this pattern,

although the grade variables were not statistically significant in these depart-

ments. However, these two departments had the fewest observations, and the

significance levels of the coefficients on grades are sensitive to the number of

observations.5

The probabilities of taking an additional course in Economics, the largest

low-grading department, and English, the largest high-grading department,

are revealing. Of the students in Economics 101 who do not intend to major in

the subject (the large majority) and who are male (also the majority), the

probability of taking a second course is 18.2 percent less if they received a B

than if they received an A, and 27.6 percent less if they received a C than if they
received an A. Responsiveness to grade is lower for those who intend to major

in Economics than those who do not; likewise, it is lower for males than for

females in Economics 101.

Students in English 101 are also responsive to their grade, though some-
what less so than students in Economics 101. Of those who do not intend to

major in English (the large majority) and are male (again the majority) the

4Our results also show that intended majors are more likely than other students to take a second
course in the department. Gender has no consistent effect on course choice. Women are less likely
to take a second course in Economics or Political Science than men, but gender made no difference
in the other departments. Need for achievement significantly influences course choice only in
Economics.

5Course choice functions were estimated using random sub-samples of the Economics and English
students in our sample. The coefficients on the grade variables became insignificant as N declined
from 375 to 200. N was less than 200 in both Political Science and Psychology.

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

Grade Inflation and Course Choice 167

probability of taking a second course in English is 14 percent less if they
received a B than if they received an A, and 20.3 percent less if they received a

C than if they received an A. As in Economics, responsiveness to grade is lower
for those who intend to major in English than for those who do not. In contrast

to Economics, males and females in English 101 do not differ in their degree of
responsiveness to grades. The structure of predicted probabilities for students
in introductory courses in Math, Political Science and Psychology are similar to

those for students in Economics and English.

Does the comparative advantage signal contained in grades influence

course choice? To answer this, we added to our list of variables a continuous
variable signifying the difference between the student’s relative performance in
the introductory course and relative performance in all courses, as measured by
grade point average up to and including the semester in which the introduc-
tory course is taken. Our analysis focussed on Economics and English, the two
departments with more than 300 observations.6

There are two notable findings. First, comparative advantage does influ-

ence course choice in both Economics and English. As students’ rank in the
introductory class increases relative to their grade point average rank, their

probability of taking a second course increases. Despite discrepancies across

departments in grading policies, students are able to derive a signal of compar-

ative advantage from their grades, and they respond to that signal.

Second, accounting for signals of comparative advantage only marginally

reduces the incentive effects of grades. While students do consider comparative

advantage, the incentive effects of absolute grades on course choice are far

more powerful. Changing a student’s grade in Economics from B to A would
increase his indicator of comparative advantage in Economics, and as a conse-

quence would increase his probability of taking another course by about 4.5

percent. That same change in grade would increase his grade incentive to take
a second Economics course, increasing the probability of doing so by about 15

percent.

Simulations and Implications for Altering Enrollments

What are the implications of differences in grading policies across depart-

ments for enrollments in courses beyond the introductory level? To answer this

question, we conducted simulations with the probabilities generated by our
course choice functions. Details of the simulation procedure are available in
Sabot and Wakeman-Linn (1988). In particular, we address the question of how

many more students would enroll in post-introductory courses in low-grading

6The sample size necessary for significant results is increased by the inclusion of the comparative
advantage variable; exercises with the largest departments indicate sample sizes below 300 produce
insignificant results. As with our earlier assessment of sensitivity to sample size, this exercise
involved a random exclusion of cases.

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

168 Journal of Economic Perspectives

departments, like Economics, if that department adopted for its introductory

courses the grading policy followed in a high-grading department, like English.

The results that follow do not use the comparative advantage variable, but its
inclusion has very little effect.

Our simulation indicated that if Economics 101 grades were distributed as
they are in English 101, the number of students taking one or more courses

beyond the introductory course in Economics would increase by 11.9 percent.

Conversely, if grades in English 101 were distributed as they are in Economics

101, the simulation indicated that the number of students taking one or more

courses beyond the introductory course in English would decline by 14.4

percent. The results of applying this method to-the Math department, which

had the lowest mean grade and the highest dispersion of grades, are more

striking. If the Math department adopted in its introductory course the English

101 grading distribution, our simulation indicated an 80.2 percent increase in

the number of students taking at least one additional Math course! Alterna-

tively, if the English department adopted the Math grade distribution, there

would be a decline of 47 percent in the number of students taking one or more

courses beyond the introductory course in English.

There are two reasons why exchanging Math and English grade policies

produces greater impact than does the exchange of Economics department and

English department grading policies. Grades in Math are substantially lower

than grades in the introductory Economics course, hence the direct impact of a

change to the English 101 grade distribution is greater in Math. Moreover, for
reasons discussed below, Math students are more responsive to grades than are
Economics students, which implies a greater increase in enrollments.

Although these results are striking, they probably underestimate the influ-

ence of grades in introductory courses on enrollments in advanced courses.

First, our simulation method assumed that the probability of an A student

taking another course was unaffected by the distribution of grades, while the

probability for other students was allowed to vary with the grade distribution.

That A students are unaffected is unlikely. If the Economics department

adopted the grade distribution of English 101, the proportion of A students

taking another Economics course would increase; making high grades easier to

obtain increases the incentive to take courses in that department. Second, our

simulations ignored the impact of grading distributions on the original decision
to take the introductory course. This effect may well be substantial.

Policy Implications and Conclusion

The division of colleges and universities into high- and low-grading depart-
ments was not conscious policy but the result of uncoordinated decisions by

individual departments and instructors. The consequent impact on the pattern
of enrollments, which we have documented, is an unintended side effect of this

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

Richard Sabot and John Wakeman-Linn 169

unplanned division. There are conflicts between implicit grading policies and

the explicit policies of these institutions. Since science departments are typically

among the low-grading departments, the skew in enrollments resulting from

divergent grading policies is in direct opposition to attempts to increase

enrollments in the sciences. Moreover, most colleges and universities would not

wish for students to be lured away from their areas of comparative advantage

by arbitrary differences in departmental grading. The policy implication seems

clear: such arbitrary differences in grading policies among departments should

be eliminated (although planned differences in grading policies may be desir-

able).

The findings of the Williams study are not neutral, however, with respect

to whether low-grading departments should raise their grades or high-grading

departments should lower their grades. Students in high-grading departments

are consistently less responsive to grades than students in low-grading depart-

ments. This appears to be a consequence of the more compressed distribution

of grades in high grading departments; the compression results in grades that

provide less accurate signals of comparative advantage and are more random.

Two additional facts support this conclusion. First, grades in high grading

departments are less accurate predictors of subsequent performance. The

average correlation between grades received in the first and second courses at

Williams is .6147 and highly significant in three low-grading departments, but

only .3681 and occasionally insignificant in five high-grading departments.

Second, various indicators of ability, prior level of skill, and motivation are

poor predictors of grades in high-grading departments. Sabot and Wakeman-

Linn (forthcoming) estimate production functions to determine what factors

contribute to success in introductory courses at Williams. In low-grading

Economics, math and verbal SAT’s, parents’ education, the student’s need for

achievement, performance in high school and sibling rank all have a significant

influence on performance. Together, they explain between a third and a half of

the variance in Economics 101 grades. By contrast, verbal SAT’s are the only

significant variable in predicting introductory English grades, and all the

variables together can explain only between 5 and 10 percent of the grade
variance.

Compressed grading distributions in high-grading departments convey

cruder signals. One reason is that instructors with fewer grading categories

must make cruder distinctions. In addition, if there is little difference in the

grade received by the top, middle, and bottom students in the class, there is less

incentive for the instructor, and less pressure from students, to make accurate

distinctions among students.

If the aim of grading is to convey information to students about their

relative strengths and weaknesses, then grade distributions with more disper-

sion and a lower average will be preferable. In addition, our results indicate

that a uniform grading policy might be an effective response to Ernest Bloch’s

entreaty to “persuade more students to study science and engineering.”

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

170 Journal of Economic Perspectives

* We are grateful to Gordon Winston for his numerous contributions to the paper, to

David Ross for econometric advice, and to the participants in the economics seminar at

Williams College for useful comments. We would like to thank the President of Williams

College for providing research funds for the study, the Registrar of Williams for

legitimizing our requests to other institutions for detailed data on grades, and the
Registrars at Amherst, Duke, Hamilton, Haverford, Pomona, and the Universities of

Michigan, North Carolina and Wisconsin for the provision of that data. Finally, we

would like to thank the editors for unusually helpful interventions.

References

Becker, Gary S., “A Theory of the Alloca-

tion of Time,” Economic Journal, 1965, 75,
493-517.

Gough, Harrison G., “The Adjective Check-

list.” Palo Alto: Consulting Psychologists Press,

1952.

Milbank, Dana, “Shortage of Scientists Ap-

proaches a Crisis as More Students Drop Out

of the Field,” The Wall Street Journal, Septem-
ber 17, 1990, p. Bi.

Ravo, Nick, “Yale Moves to Make Cum

Laude Mean More,” New York Times, May 22,
1988, p. 26.

Sabot, Richard, and John Wakeman-Linn,
“Performance in Introductory Courses: A Pro-

duction Function Analysis,” Journal of Eco-
nomic Education, forthcoming.

Sabot, Richard, and John Wakeman-Linn,
“Grade Inflation and Course Choice,”

Williams College Research Paper, November,

1988.

U.S Department of Education, “The Condi-
tions of Education,” Center for Educational

Statistics, Washington, D.C., 1987, pp.
104- 105.

Winston, Gordon C., The Timing of Economic
Activities: Firms, Households, and Markets in

Time-Specific Analysis. Cambridge: Cambridge

University Press, 1982.

This content downloaded from 147.97.138.196 on Mon, 26 Feb 2018 13:45:16 UTC
All use subject to http://about.jstor.org/terms

• Contents

image 1
image 2
image 3
image 4
image 5
image 6
image 7
image 8
image 9
image 10
image 11
image 12

• Issue Table of Contents

The Journal of Economic Perspectives, Vol. 5, No. 1, Winter, 1991
Front Matter [pp. 1 – 2]
Symposium: Intellectual Property
An Introduction to the Law and Economics of Intellectual Property [pp. 3 – 27]
Standing on the Shoulders of Giants: Cumulative Research and the Patent Law [pp. 29 – 41]
A Patent System for Both Diffusion and Exclusion [pp. 43 – 60]
Some Economics of Trade Secret Law [pp. 61 – 72]
Taking Stock: A Critical Assessment of Recent Research on Inventories [pp. 73 – 96]
Institutions [pp. 97 – 112]
Efficient Transportation Infrastructure Policy [pp. 113 – 127]
Is Probability Theory Relevant for Uncertainty? A Post Keynesian Perspective [pp. 129 – 143]
The When, the How and the Why of Mathematical Expression in the History of Economics Analysis [pp. 145 – 157]
Grade Inflation and Course Choice [pp. 159 – 170]
Distinguished Fellow: An Appreciation of Guy Orcutt [pp. 171 – 179]
Policy Watch: Cutting Capital Gains Taxes [pp. 181 – 192]
Anomalies: The Endowment Effect, Loss Aversion, and Status Quo Bias [pp. 193 – 206]
Recommendations for Further Reading [pp. 207 – 211]
Correspondence [pp. 213 – 216]
Notes [pp. 217 – 223]
Back Matter [pp. i – viii]

Nunn,N. (2008). The Long Term Effects of Africa’s Slave Trades. Quarterly Journal of Economics, 123(1), 139-176.

Introduction:

· Nunn’s article begins by stating the need for the research. He states that parts of Africa have continuously faced economic underdevelopment (a trend that still continues to this day).

· This economic underdevelopment has typically been attributed to Africa’s colonialist past (especially European colonialism) and Africa’s many waves of slave trading.

· Citing a lack of empirical analysis on the latter topic, Nunn proceeds with the thesis of his paper: African countries more affected by the slave trades are more likely to face current economic underdevelopment and stagnation.

· Nunn recognized that this negative relationship between countries most affected by slave trade and current GDP may not have a causal connection. For this reason, he goes on to rule out any other intervening variables that could serve as a better explanation.

· One of the main possibilities of an alternative casual explanation is that the countries most affected were countries that were already struggling (before 1400 A.D.)

· To rule out this rival hypothesis, Nunn consulted two different sources of information: (1) African Historical experts on why certain areas were selected over others to be involved in the slave trades, and (2) historic population data for various African countries.

· Both seem to disprove the rival hypothesis because data shows that the areas most affected by the slave trades are the areas that were the most developed previously.

Historical Background:

· The African Slave Trades lasted for about half of a millennium (1400-1900 approximately) and essentially cut the population of Africa in half.

· There were four main waves of slave trading focusing on different regions at different times, but the most recognized is known as the Trans-Atlantic slave trade. (Slaves shipped from West/West Central Africa to Europe and the Americas)

· The three other waves include: the trans-Saharan slave trade, the Red Sea slave trade, and the Indian Ocean slave trade.

· The African slave trades are the most massive in history with approximately 12 million people exported in just the trans-Atlantic trade (6 million in the 3 others combined) (figures not including those who died in raids or travel).

· Another reason that Nunn stated for the uniqueness of these trades is that this was the first time that certain ethnicities began to enslave their own people.

· Nunn stated that this is a direct cause of increased “social and ethnic fragmentation” that weakened states and pre-trade institutions. (This also highly contributed to rampant amoral familism and the destruction of the African tribal system (C.f. George Ayittey))

· The slave trades caused much infighting between different African groups. This lead to a further deterioration of prior social, political, and judicial institutions. Corruption grew uncontrollably.

Slave Export Data:

· Nunn states that his mode of measurement for this study is the number of slaves taken from each country during the years 1400-1900.

· To obtain the total number of slaves exported from Africa, Nunn used shipping data from 34,584 voyages from 1514-1866 as well as other databases that have been extensively compiled over several years.

· In order to locate where the slaves originated from in Africa, Nunn looked at data that reported their ethnic identities. (Data from “records of sale, slave registers, slave runaway notices, court records, church records, and notarial documents.”)

· Nunn then created an algorithm to try and accurately show (as much as humanly possible) in which region each slave originated.

Basic Correlations: OLS Estimates

· This section is very math intensive, but basically Nunn runs several linear regression models showing that there is indeed a relationship between countries with a history of high slave trade and low GDP (2000 was the year used). He uses the same methods to show how other rival hypotheses are not adequate in addressing causality.

Conclusions:

· The African slave trade is a significant contribution to current economic underdevelopment in the region. Ethnic fractionalization and the destruction of institutions are byproducts that continue to manifest themselves in modern day Africa. Data (qualitative and quantitative) also shows that more affected areas were previously more developed. The slave trade definitely contributed to stunted growth.