2003 Joint Statistical Meetings - Section on Statistical Education

STUDENTS’ ATTITUDES:
THE “OTHER” IMPORTANT OUTCOME IN STATISTICS EDUCATION

Candace Schau
CS Consultants, Albuquerque, NM 87111

KEY WORDS: Statistics Education, Attitudes, assess attitudes toward statistics and assess them
Achievement, Assessment well.
In this paper, I will begin to address the
The ultimate goal of statistics education is to following six questions about students’ attitudes
produce adults who appropriately use statistical toward statistics:
thinking. Most college students take only one
statistics course, the introductory course. This 1. What are attitudes, especially attitudes
course, then, is where we, as statistics toward statistics?
instructors, do or do not motivate students to 2. Are attitudes toward statistics important?
apply the statistics that they have learned in their 3. How do we measure attitudes toward
jobs and in their lives. statistics?
Yet, Butler (1998) entitled an AmStat 4. What do we know about students’ attitudes
Forum article “On the Failure of the Widespread toward statistics?
Use of Statistics.” He suggested that, in spite of 5. Are attitudes toward statistics and statistics
the increasing numbers of adults who complete course achievement causally related?
introductory statistics courses, these adults often 6. How can we influence students’ attitudes
do not use statistical methods in their jobs and, toward statistics?
when they do try, “the results are shambles” (p.
84). What Are Attitudes?
The appropriate use of statistical The construct of attitudes plays an important role
understanding requires persistence. Students, of in social psychology. In spite of this role,
course, first need to complete their introductory however, there are a variety of definitions of
statistics course successfully, rather than drop attitudes with no accepted consensus. Attitude
out. In their lives outside of class, they then theorists do agree that the defining characteristic
need to be able to recognize when they require of an attitude is its evaluative aspect. Ajzen
additional statistical knowledge and skills; obtain (1989) uses a global definition that works well
this additional statistical understanding or better when considering students’ attitudes toward
yet enlist the aid of a statistician; and accurately statistics: “an attitude is an individual’s
use the skills they possess. disposition to respond favorably or unfavorably
The accomplishment of these goals requires to … any … discriminable aspect of the
more from students than a good grade in a individual’s world” (p. 241). In our case, the
statistics course. Students who will use their “world” is anything associated with statistics.
statistical knowledge appropriately must: This definition helps us think about
attitudes toward statistics but we still need to
Think that statistics is useful in their measure them. Once a measure assessing
professional and personal lives, students’ attitudes toward statistics is created and
Believe that they can understand and use used, we then have an operational definition of
statistics, and this construct, one that is useful for identifying
Know that they don’t understand everything and dealing with students’ attitudes.
they might need based only on what they
learned in their introductory statistics Are Attitudes Toward Statistics Important?
course. Many statistics educators and most statistics
students believe that attitudes toward statistics
These statements describe attitudes about are important. Students who hold and express
statistics, the “other” important outcome in negative attitudes can create an uncomfortable
statistics education (Gal, Ginsburg, & Schau, classroom climate (Gal, Ginsburg, & Schau,
1997; Garfield, Hogg, Schau, & Whittinghill, 1997). In addition, many of us believe that
2002). As with any other important educational attitudes impact students’ achievement, course
goal, such as learning, we need to be able to completion, future course enrollment, and

3673
 2003 Joint Statistical Meetings - Section on Statistical Education

statistical thinking (or lack thereof) in their lives Statistical Anxiety Rating Scale (STARS;
outside the classroom. Cruise, Cash, & Bolton, 1985) was designed to
A variety of educational and cognitive assess statistics anxiety, which is only one part
theories propose that attitudes are important in of one component of attitudes toward statistics.
course achievement and persistence and in the For a description of these (and other) surveys,
use of course-learned information outside of the see Sorge (2001).
classroom (see Sorge, 2001, for a brief These authors essentially originated survey
description of some of these models). research into students’ attitudes toward statistics.
Expectancy-value models of behavior are However, their measures, as well as other less
especially useful in mathematics and statistics frequently used measures, do not present a
education. Eccles and her colleagues have taken consistent picture of students’ attitudes toward
these expectancy-value models and applied them statistics. The creators of these measures
to mathematics attitudes and achievement in K- disagreed about such fundamentals as what
12 students (see, for example, Eccles, Adler, components and how many components
Futterman, Goff, Kaczala, Meece, & Midgley, comprise attitudes toward statistics. The
1983, and Eccles & Wigfield, 1995). We, in turn, component names are often misleading and
have taken their model and applied it to statistics simply incorrect in some of the surveys. The
attitudes and achievement. items themselves suffer from a number of
Eccles and colleagues believe attitudes are problems. The most fundamental is that some
multi-dimensional, that is, that attitudes are items appear to be misplaced in regard to the
composed of constructs or factors that, although component they are supposed to be measuring.
related, are distinct. The three expectancy-value Using an extensive development and testing
factors of most use to us in statistics education process, I created the Survey of Attitudes
include: Toward Statistics or SATS©. The current
version of the SATS consists of 28 items
1. Expectancies for Success - students’ self- measuring four components of students’ attitudes
concepts regarding their ability to do toward statistics. These components and
statistics successfully, example items from the pretest version follow:
2. Task Difficulty - students’ perceptions of the
difficulty of statistics, and Affect (6 items) – students’ feelings concerning
3. Task Value - students’ perceptions of the statistics
value of doing statistics successfully. “I am scared by statistics.”

Each of these three factors suggests an important Cognitive Competence (6 items) – students’
component to attitudes toward statistics. In attitudes about their intellectual
addition, students’ perceptions of their past knowledge and skills when applied to
academic performances (in math and in statistics, statistics
if they have had previous experience in the later) “I can learn statistics.”
influence each of these three factors.
Value (9 items) – students’ attitudes about the
How Do We Measure Attitudes Toward usefulness, relevance, and worth of
Statistics? statistics in personal and professional
There are a variety of ways to measure students’ life
attitudes toward statistics. See Gal, Ginsburg, “I use statistics in my everyday life.”
and Schau (1997) for a description of some of “Statistics is not useful to the typical
these approaches. However, the most common professional.”
approach by far, especially in post-secondary
statistics courses, is to use a Likert survey. This Difficulty (7 items) – students’ attitudes about
approach is easy and quick to use. the difficulty of statistics as a subject
In late 1980’s, there were two commonly “Most people have to learn a new way
used surveys purporting to assess post-secondary of thinking to do statistics.”
students’ attitudes toward statistics. They
included the Statistics Attitude Survey (SAS; The four components in the SATS are
Roberts & Bilderback, 1980, and Roberts & consistent with our application of Eccles and
Saxe, 1982), and Attitudes Toward Statistics colleagues’ three expectancy-value factors to
(ATS; Wise, 1985). A third measure, the statistics education (although Eccles and

3674
 2003 Joint Statistical Meetings - Section on Statistical Education

colleagues included affective perceptions within

their factor called Task Value, we included Affect (17 values from 9 studies) .80 to .89,
Affect as a separate attitude component). These Cognitive Competence (16 values from 8 studies)
same four components also are found in a variety .77 to .88,
of other theories concerned with the multi- Value (17 values from 9 studies) .74 to .90,
dimensionality of attitudes. Difficulty (16 values from 8 studies) .64 to .81.
All of the items in the four components use
a 7-point Likert response scale (1 = Strongly The Difficulty component tends to exhibit the
Disagree, 4 = Neither Disagree nor Agree, 7 = lowest level of internal consistency, but that
Strongly Agree). Although some of the items level is considered at least adequate.
are written negatively, responses are reversed Two kinds of score validity information are
before scoring so higher responses always mean available for the SATS. The first kind concerns
more positive attitudes. the score validity of the four-component
The SATS also contains items that assess structure. Two sets of confirmatory factor
students’ academic backgrounds and analyses indicate that the four-component
demographic information (e.g., gender, age), as structure fits responses to the SATS well and that
well as the grade they expect to receive in their the items fit into their hypothesized components
statistics course. In addition, the SATS includes (Dauphinee, Schau, & Stevens, 1997; Hilton,
three global items that assess students’ attitudes Schau, & Olsen, in press; Schau, Stevens,
regarding Math Cognitive Competence, Statistics Dauphinee, & Del Vecchio, 1995). These
Cognitive Competence, and Career Value of findings imply that scores from the SATS have
statistics. The SATS can be viewed at the same meaning for both genders at both
http://www.unm.edu/~cschau/infopage.htm. administration times; that is, mean scores by
Students, on average, spend about ten gender, by time of administration, and by their
minutes responding to the items on the SATS interaction can be compared.
administered in a paper-and-pencil or a Web The second kind of score validity
format. The SATS is easy to administer in either information often is called concurrent validity.
format. Scores have concurrent validity if they interrelate
Before using any kind of measure, it is as expected with other measures of similar
important to examine the measurement quality of constructs. There is evidence of concurrent
its scores. Usually, both score reliability and validity for the SATS component scores of
validity are examined. A variety of research Affect, Cognitive Competence, and Value. SATS
evidence indicates that scores from the SATS Affect scores correlated strongly with scores
have good measurement properties. from Wise’s ATS Course scale, which also
For surveys, reliability usually is assessed measures students’ affective feelings about
as the internal consistency of the items statistics (Cashin & Elmore, 2000; Schau,
composing each scale, that is, the degree of Stevens, Dauphinee, & Del Vecchio, 1995).
interrelationship among students’ responses to Scores from the SATS Value component
the scale’s items. Cronbach’s coefficient alpha correlated strongly with scores from the ATS
often is used for this assessment. The SATS Field scale, which measures students’ attitudes
component scores generally exhibit reasonably about the value of statistics (Cashin & Elmore,
high alpha values indicating good internal 2000; Schau, Stevens, Dauphinee, & Del
consistencies. These values show a consistent Vecchio, 1995).
picture within each attitude component across In addition, the expected relationships
studies that vary in terms of student, course, and between the SATS global attitudes items and the
instructor characteristics and time of corresponding component scores were found.
administration of the SATS (Cashin & Elmore, Correlations greater than +. 5 were found
2000; Faghihi & Rakow, 1995; Hilton, Schau, & between students’ pre-test Cognitive Competence
Olsen, in press; Mayer, 1999; Mills, 2002; component scores and their pre-test responses to
Schau, Dauphinee, & Del Vecchio, 1992; Schau, the single global Cognitive Competence item, as
Dauphinee, & Del Vecchio, 1993; Schau, well as between their pre-test Value component
Stevens, Dauphinee, & Del Vecchio, 1995; scores and their pre-test responses to the single
Schutz, Drogosz, White, & Distefano, 1999; global Career Value item.
Watson, Lang, & Kromrey, 2002; Wisenbacker
& Scott, 1995). The range of alpha values by What Do We Know About Students’ Attitudes
component includes: Toward Statistics?

3675
 2003 Joint Statistical Meetings - Section on Statistical Education

There is not much research on students’ attitudes post-test received grades of about C+ (2.45)
toward statistics. Faculty in education have while those who took both received grades of
conducted most of the research that does exist. about B (2.88), a difference of .4 points on the
They usually study the students in their own usual numeric grading scale. Whenever
courses; many of their students are education possible, pre-test analyses were conducted on
majors and may be advanced undergraduate or both sets of students: all 580 students who took
graduate students. We know little about the the pre-test and the 287 students from this group
attitudes of undergraduates who are enrolled in who took both the pre- and post-tests.
introductory statistics courses and even less To date, my analyses have yielded four
about students in these statistics courses offered broad findings of interest, which are summarized
by mathematics or statistics departments. below. I have not included the statistical
In this paper, I emphasize findings from a significance tests in this summary. Instead, I
sample of the students who participated in the emphasize the size of relationships.
development and testing of the SATS. These
findings come from data collected from 1. Students’ spoken attitudes were more
undergraduates who were enrolled in the negative than were responses to the SATS.
introductory statistics course offered by the Because of the predominance of strong
Mathematics and Statistics Department of a negative words and phrases created during the
major Southwestern research university. The development phase of the SATS, I expected
Psychology, Sociology, Engineering, and students' responses to the SATS survey would be
Business Departments at this University offered at least somewhat negative too. Thus, the results
their own introductory statistics courses so were unexpected. For both sets of pre-test
students with these majors are not well analyses and the post-test analyses, average
represented in these findings. We collected Cognitive Competence scores were somewhat
SATS attitudes and course achievement data in a positive (about 1 point above neutral), as was the
total of 11 sections of this introductory statistics mean Value score at the pre-test administration.
course across two consecutive semesters. The mean post-test Value score was slightly
Five hundred eighty students completed the positive, about ½ point above neutral. The
SATS within the first two weeks of the Difficulty scale was the only scale to yield mean
beginning of their course (the pre-test negative attitudes, and these means were only
administration); 287 of these students also slightly negative for both pre- and post-tests
completed it within the last two weeks (the post- (about half a point below neutral). See Table 1.
test administration). Only one student took the These means generally are similar to those found
SATS during the post-test administration but not in research using the SATS with other samples
during the pre-test administration; that student’s of students (e.g., Cashin & Elmore, 2000;
data were not included in the analyses. Faghihi & Rakow, 1995; Hilton, Schau, &
Of the 293 students who took the pre-test Olsen, in press; Mayer, 1999; Mills, 2002;
but not the post-test, 201 (69%) did not receive a Schau, Dauphinee, & Del Vecchio, 1992; Schau,
letter grade (A through F); since they had Dauphinee, & Del Vecchio, 1993; Schau,
withdrawn from the course, they could not have Stevens, Dauphinee, & Del Vecchio, 1995;
participated in the post-test data collection. Schutz, Drogosz, White, & Distefano, 1999;
Their mean pre-test attitude scores were lower Watson, Lang, & Kromrey, 2002; Wisenbacker
than those of the students who took both the pre- & Scott, 1995).
and the post-tests by .1 point or less (less than It appears that spoken attitudes are more
2% differences on this scale); clearly these negative than those recorded on a survey.
differences were small. Perhaps students who hold negative attitudes are
Participation rates were high; usually, every more verbal than those who hold neutral and
student present on the day of SATS data positive ones.
collection participated, with the occasional
exception of one or two students. Thus, it is 2. Students attributed their attitudes to their
likely that most of the 92 students who received achievement and to instructors.
a letter grade but did not take the post-test SATS Students in two sections of a required
were absent the day we collected the post-test introductory graduate-level statistics course
data. Unfortunately, these students could not taught in a College of Education were given an
afford to miss class. On average, the students extra-credit opportunity to write brief statements
who took the SATS at the pre-test but not at the about their attitudes and the sources for these

3676
 2003 Joint Statistical Meetings - Section on Statistical Education

attitudes regarding mathematics and statistics to .9 points for Affect (about a 15% difference)
and courses in these disciplines. Although they for students who took the SATS at both
cited a variety of sources for their feelings, they administration times. See Table 2.
most often mentioned two general themes: their At the post-test, the differences between
achievement and teacher (and class) mean scores in the sections with the highest and
characteristics. At the beginning of the classes, lowest means were large for all four components,
these students attributed positive attitudes to much larger than they were at pre-test. As
good math achievement that created positive occurred at the pre-test, mean Affect scores
math self-concepts. Students attributed negative showed the largest section differences of almost
attitudes at the beginning of classes to poor 2 points (about 32%), twice as large as the mean
teaching that led to poor mathematics self- pre-test difference. The smallest mean post-test
concepts and poor achievement. Many students difference was for Value; this post-test mean
also attributed positive change in their attitudes difference was over one point (about 20%), again
across their statistics course (as well as high almost twice as large as the Value difference that
achievement) to teacher characteristics. occurred at the pre-test. See Table 3 for the
As one student wrote, “Instructors make a highest and lowest raw mean post-test section
large difference.” See the chapter by Gal, scores.
Ginsburg, and Schau (1997) and papers by These results suggest that the section in
Onwuegbuzie, Da Ros, and Ryan (1997) and which the student was enrolled is important in
Watson, Kromrey, Lang, Hess, Hogarty, and regard to their attitudes. To explore the
Dedrick (2003) for additional discussion of this contribution of section to post-test attitude score
important topic. variability, analysis of covariance was used to
adjust component post-test responses for
3. Mean attitudes varied. corresponding pre-test responses; section was the
Attitudes varied, depending primarily on the predictor variable. Pre-test attitude scores were
component being measured and the section in important in post-test score variability. Pre-test
which the student was enrolled. Mean gender attitude scores shared from 11% (Affect) to 22%
and ethnic attitude differences were small, when (Cognitive Competence and Value) of the
they existed at all, although there was some variance in post-test attitude scores, depending
indication that these differences were larger at on the attitude component being studied. See
the end than at the beginning of the course. Table 4.
Similarly, attitudes did not change much from Section, controlling for pre-test scores, also
the beginning to the end of the course. Value was an important factor in students’ post-test
exhibited the largest change; Value mean scores attitudes for all four components. Section shared
decreased by almost twice as much as mean from 11% (Value) to 21% (Affect) of the
scores on the other three components (.4 points variance in post-test attitude scores, depending
which is a decrease of about 7% on the scale). on the attitude component being studied. See
However, differences among mean scores on the Table 4.
four attitude components were large. As These findings support the idea of the
indicated above, on average, students’ Cognitive importance of the class experience involving
Competence and Value attitudes were highest interactions among the course instructor and the
and positive. Affect attitudes were neutral. group of students in the class in terms of
Difficulty attitudes were slightly negative. See students’ attitudes, as well as the importance of
Table 1. the attitudes students bring to this course. It also
Mean attitudes differences among sections suggests that some sections of students are easier
were large at the beginning of the semester and to work with than other sections.
even larger at the end. With 11 sections
measured twice on each of four attitude 4. Students’ attitudes were positively related to
components, we have 88 means. To make this their achievement
task manageable, I first looked at the mean Like many others, I believe that statistics
difference between the sections with the best and attitudes and achievement are positively related.
the worst attitudes. However, research evidence supporting this
The pre-test differences between the belief is not yet well established. Until recently,
highest and lowest mean section scores ranged studies exploring attitudes toward statistics have
from about 1/2 point for Difficulty (about an 8% focused on a small part of the complex
difference) for all students who took the pre-test relationships between attitudes and achievement.

3677
 2003 Joint Statistical Meetings - Section on Statistical Education

These studies often have explored these students who enter our classes already possess
relationships by correlating attitude and attitudes toward statistics and learning that will
achievement scores. impact their course performances. Attitudes and
The only achievement variable available for Course Achievement also are endogenous
my sample was letter grade; it was converted to variables that impact each other throughout the
the usual numeric grading scale and standardized course and that are impacted by both Prior
within section due to differences in instructors’ Attitudes and Prior Achievement. In my model,
grading standards across sections. Attitude the four endogenous Attitude constructs match
component scores also were standardized within the four components of the SATS. The direction
section. Table 5 presents correlations from this of the impacts among the Attitude components is
sample. The relationships among the pre-test based on work by Eccles and Wigfield (1995).
attitude components and course grade were See Figure 1.
small, .20 or below. The relationships among This global model is not testable without
the post-test attitude components and course refinement. However, using data from
grade were larger than the pre-test relationships, engineering undergraduate students in a required
but they still were not large. These results are introductory engineering statistics course, my
consistent with those found in other research (see colleague Carmen Sorge (2001) and I were able
Sorge, 2001 for a summary of this research). to test a part of this model. With some
A hierarchical regression, entering the four modifications, the data fit the model adequately.
standardized pre-test attitude component scores The post-test attitude components together
in the first block and the four standardized post- accounted for about 1/3 of the variation in course
test component scores in the second block, was achievement (pre-test attitudes could not be
used to predict standardized grade. The pre-test included in this variation of the model). Prior
block shared only 3% of the variance in grade. achievement accounted for the remaining 2/3. I
The post-test block, controlling for the pre-test believe that these variance percentages
block, shared an additional 16% of the variance associated with course achievement (1/3 with
in grade. Together, they shared 20%. attitudes and 2/3 with prior achievement) will
These relationships, although adequate, generalize to other samples.
were not strong. There are at least three reasons My model isn’t the only reasonable model
these relationships were not stronger. First, relating statistics attitudes and achievement. See
letter grade is not the best measure of course the work by Harlow, Burkholder, and Morrow
achievement to use in analyses due to its limited (2002) and by Wisenbaker and colleagues
number of possible values. Total course points (e.g.,Wisenbaker and Scott 1997; Wisenbaker,
would be a better measure. Second, students Scott, & Nasser, 1999; Wisenbaker, Scott, &
who didn’t participate in the post-test, on Nasser, 2000).
average, received lower grades than those who The attitude components assessed by the
did participate, thus likely restricting the size of SATS, coupled with students’ prior achievement,
the relationships. Third, simple correlations and are not all of the important student inputs into
regressions are unlikely to represent the their work in statistics courses. We currently are
complexity of the interrelationships among adding two other components to the SATS.
attitudes and achievement. These include:

Are Attitudes Toward Statistics and Statistics Interest – students’ self-reported level of
Course Achievement Causally Related? individual interest in statistics, and
I believe that attitudes toward statistics and Effort – amount of work students say they
course achievement causally impact each other, expend to learn statistics.
and that these relationships can be represented in
a model. Models often contain constructs that It is not clear, however, if these two constructs
are internal to the model (endogenous constructs) are components of attitudes. Other important
and constructs that are external to the model inputs and outcomes include students’ goals for
(exogenous constructs). These models represent studying statistics and the metacognitive
the researcher’s idea about the causal approaches they use in doing so.
relationships among the constructs while taking
the exogenous constructs as “givens.” How Can We Influence Students’ Attitudes
In my model, Prior Attitudes and Prior Toward Statistics?
Achievement are related exogenous variables;

3678
 2003 Joint Statistical Meetings - Section on Statistical Education

There are many things that we as instructors can Conclusion

do to try to influence our students’ attitudes and I began this paper by indicating that many of us
to help them at least complete our courses. want our students to be intelligent users of
Unfortunately, there is little research available on statistics in their lives. This outcome may have
the effectiveness of these approaches. My little to do with course achievement and
suggestions, which aren’t exhaustive, are based everything to do with their attitudes toward
on educational and cognitive theories, what I’ve statistics. The SATS is a simple measure that
tried, and others’ suggestions that sound assesses these attitudes. It is easy to use, score,
reasonable to try. Whatever you choose to do, and explain.
however, must be comfortable for you and fit We need to better understand students’
into your vision of yourself as an introductory attitudes toward statistics and their
statistics course instructor. See also Harris and interrelationships with achievement and eventual
Schau (1999). use in life, and we need to find more methods for
promoting positive attitudes. I believe that
1. Encourage students with debilitating assessing our students’ attitudes and creating,
negative attitudes to see a counselor. considering, and testing models such as the one
2. Stress that your statistics course is not a I’ve presented will help us accomplish these
math course (unless it is). tasks.
3. Bring positive attitudes to your course.
4. If you believe that students’ attitudes are References
important, acknowledge their importance. Ajzen, I. (1989). Attitude structure and
5. If engendering positive attitudes is one of behavior. In A. R. Pratkanis, S. J. Breckler, &
your course goals, assess attitudes twice A. G. Greenwald (Eds.), Attitude Structure and
(pre- and post-tests) to evaluate your success Function (pp. 241-274). Hillsdale, NJ:
in achieving this goal. Lawrence Earlbaum.
6. Use activities that will help students identify Butler, R. S. (1998, March). On the
and acknowledge their attitudes. failure of the widespread us of statistics. Amstat
7. Provide a great deal of structure in your News (p. 84), Alexandria, VA: American
course. Statistical Association.
8. Use humor but not sarcasm. Cashin, S. E., & Elmore, P. B. (October,
9. Let students know that it is likely that both 2000). The Survey of Attitudes Toward Statistics
you and they will make mistakes sometime scale: A construct validity study. Paper
during the course; use those mistakes as presented at the meeting of the Mid-Western
“teaching moments” for content, process, Education Research Association, Chicago, IL.
and attitudes. Cruise, R. J., Cash, R. W., & Bolton, D. L.
10. Allow students to use so-called “cheat (1985). Development and validation of an
sheets” on exams. instrument to measure statistical anxiety. 1985
11. If possible, use more than in-class tests for Proceedings of the American Statistical
assigning grades. Association, Statistics Education Section (pp. 92-
97). Alexandria, VA: American Statistical
There are at least two additional sources of Association.
help. Freda Watson, for her doctoral dissertation Dauphinee, T. L., Schau, C., & Stevens, J.
at the University of South Florida - Tampa, is J. (1997). Survey of Attitudes Toward Statistics:
creating a multi-media program called EncStat to Factor structure and factorial invariance for
identify students with poor attitudes toward females and males. Structural Equation
statistics and to help them develop more positive Modeling 4, 129-141.
attitudes. A second phase of the project (EncStat Eccles, J., Adler, T. F., Futterman, R., Goff,
– Professor) will provide statistics instructors S. B., Kaczala, C. M., Meece, J. L. and Midgley,
with information about statistics anxiety and how C. (1983). Expectancies, values, and academic
to help students cope with it. In addition, behaviors. In J. T. Spence (Ed.), Achievement
Anthony Onwuegbuzie (also at the University of and Achievement Motivation (pp. 75-146). San
South Florida – Tampa) does a prodigious Francisco: W.J. Freeman.
amount of research on the correlates of statistics Eccles, J. S., & Wigfield, A. (1995). In the
anxiety, one part of the Affect component of mind of the actor: The structure of adolescents’
attitudes toward statistics. achievement task value and expectancy-related

3679
 2003 Joint Statistical Meetings - Section on Statistical Education

beliefs. Personality and Social Psychology attitude survey. Educational and Psychological
Bulletin 21, 215-225. Measurement, 40, 235-238.
Faghihi, F., & Rakow, E. A. (1995, Roberts, D. M., & Saxe, J. E. (1982).
November). The relationship of instructional Validity of a statistics attitude survey: A follow-
methods with student responses to the survey of up study. Educational and Psychological
attitudes toward statistics. Paper presented at the Measurement, 42, 907-912.
meeting of the Mid-South Educational Research Schau, C. G., Dauphinee, T., & Del
Association, Biloxi, MS. (ERIC Document Vecchio, A. (1993, April). Evaluation of two
Reproduction Service no. ED 392358) surveys measuring students' attitudes toward
Gal, I., Ginsburg, L., & Schau, C. (1997). statistics. Paper presented at the meeting of the
Monitoring attitudes and beliefs in statistics American Educational Research Association,
education. In I. Gal & J. B. Garfield (Eds.), The Atlanta, GA.
Assessment Challenge in Statistics Education Schau, C. G., Dauphinee, T., & Del
(pp. 37-51). Netherlands: IOS Press. Vecchio, A. (1992, April). The development of
Garfield, J., Hogg, B, Schau, C., & the Survey of Attitudes Toward Statistics. Paper
Whittinghill, D. (2002). First courses in presented at the meeting of the American
statistical science: The status of educational Educational Research Association, San
reform efforts. Journal of Statistics Education, Francisco, CA.
10, (2), Schau, C., Stevens, J., Dauphinee, T. L., &
http://www.amstat.org/publications/jse/v10n2/ga Del Vecchio, A. (1995). The development and
rfield,html. validation of the Survey of Attitudes Toward
Harris, M. B., & Schau, C. (1999). Statistics. Educational and Psychological
Successful strategies for teaching statistics. In S. Measurement, 55, 868-875.
N. Davis, M. Crawford, & J. Sebrechts (Eds.), Schutz, P.A., Drogosz, L. M., White, V. E.,
Coming into her Own: Educational Success in & Distefano, C. (1999). Prior knowledge,
Girls and Women (pp. 193-210). San Francisco: attitude, and strategy use in an introduction to
Jossey-Bass. statistics course. Learning and Individual
Harlow, L. L., Burkholder, G. J., & Differences, 10, 291-308.
Morrow, J. A. (2002). Evaluating attitudes, Sorge, C. (2001). Impact of engineering
skill, and performance in a learning-enhanced students’ attitudes on achievement in statistics:
quantitative methods course: A structural A structural equation model analysis.
modeling approach. Structural Equation Unpublished doctoral dissertation, University of
Modeling, 9, 413-430. New Mexico, Albuquerque.
Hilton, S. C., Schau, C., & Olsen, J. A. Watson, F. S., Kromrey, J. D., Lang, R.,
(in press). Survey of Attitudes Toward Hess, M. R., Hogarty, K. Y, & Dedrick, R. F.
Statistics: Factor structure invariance by gender (2003, April). Multifaceted foci: The
and by administration time. Structural Equation antecedents of statistics anxiety and negative
Modeling. attitudes toward statistic. Paper presented at the
Mayer, A. M. (1999). The relationship American Educational Research Association,
between the use of multimedia instructional Chicago.
software and students attitudes toward statistics. Watson, F. S., Lang, T. R., & Kromrey, J.
Unpublished master’s thesis. University of D. (2002, November). Breaking ground for
Minnesota, Minneapolis. EncStat: A statistics anxiety intervention
Mills, J. A. (2002, April). Using program. Paper presented at the meeting of the
categorical data analysis to describe student Florida Educational Research Association,
attitudes toward statistics. Paper presented at Gainesville, FL.
the meeting of the American Educational Wise, S. L. (1985). The development and
Research Association, New Orleans, LA. validation of a scale measuring attitudes toward
Onwuegbuzie, A. J., Da Ros, D., & Ryan, statistics. Educational and Psychological
J. M. (1997). The components of statistics Measurement, 45, 401-405.
anxiety: A phenomenological study. Focus on Wisenbaker, J., & Scott, J. (1995, April).
Learning Problems in Mathematics, 19(4), 11- Attitudes about statistics and achievement
35. among students taking introductory statistics
Roberts, D. M., & Bilderback, E. W. courses. Paper presented at the meeting of the
(1980). Reliability and validity of a statistics American Educational Research Associates, San
Francisco, CA.

3680
 2003 Joint Statistical Meetings - Section on Statistical Education

Wisenbaker, J. M., & Scott, J. S. (1997,

April). Modeling aspects of students’ attitudes
and achievement in introductory statistics
courses. Paper presented at the meeting of the
American Educational Research Association,
Chicago, IL.
Wisenbaker, J. M., Scott, J. S. and Nasser,
F. (1999). A cross-cultural comparison of path
models relating attitudes about and achievement
in introductory statistics courses. Paper
presented at the 52nd Session of the International
Statistical Institute, Helsinki, Finland.
Wisenbaker, J. M., Scott, J. S. and Nasser,
F. (2000). Structural equation models relating
attitudes about and achievement in introductory
statistics courses: A comparison of results from
the U.S. and Israel. Paper presented at the Ninth
International Congress on Mathematics
Education, Tokyo, Japan.

3681
 2003 Joint Statistical Meetings - Section on Statistical Education

Table 1. SATS Attitude Component Mean Scores (Standard Deviations) at Pre- and Post-Test

Affect Cognitive Value Difficulty

Competence

Pre-test* 4.03 4.91 4.86 3.62

(1.14) (1.09) (1.01) (0.76)
Pre-test ** 4.12 5.01 4.96 3.62
(1.13) (1.09) (0.97) (0.78)
Post-test 3.95 4.84 4.57 3.49
(1.45) (1.27) (1.21) (1.15)

* Pre-test scores for students who took the pre-test (n=580).

** Pre-test scores for students who took the post-test also (n=287).

Table 2. Lowest and Highest SATS Pre-test Component Mean Scores by Section (Section Number)

Students Who Took Pre-test Students Who Took Pre- & Post-test
Lowest Highest Difference Lowest Highest Difference

Affect 3.57 (3) 4.41 (2) 0.84 3.60 (3) 4.50 (4) 0.90
Cognitive 4.58 (3) 5.21 (6) 0.63 4.65 (3) 5.34 (4) 0.69
Competence
Value 4.60 (8) 5.27 (2) 0.67 4.60 (8) 5.33 (6) 0.73
Difficulty 3.31 (9) 3.78 (4) 0.47 3.31 (9) 3.96 (4) 0.65

Table 3. Lowest and Highest Raw and Adjusted SATS Post-test Component Mean Scores by Section
(Section Number)

Raw Adjusted
Lowest Highest Difference Lowest Highest Difference

Affect 2.85 (10) 4.74 (8) 1.89 2.90 (10) 4.77 (8) 1.87
Cognitive 3.79 (10) 5.53 (8) 1.74 3.90 (10) 5.49 (8) 1.59
Competence
Value 4.09 (10) 5.26 (5) 1.17 3.96 (6) 5.29 (5) 1.33
Difficulty 2.67 (2) 4.05 (8) 1.38 2.69 (2) 4.05 (8) 1.36

Table 4. Percent Variance in Post-test Attitude Scores Associated with Pre-test Attitude Scores and with
Section Membership by attitude component

Attitude Component Pre-test Section Total

Affect 11% 21% 32%

Cognitive Competence 22% 18% 40%
Value 22% 11% 33%
Difficulty 14% 14% 28%

3682
 2003 Joint Statistical Meetings - Section on Statistical Education

Table 5. Correlations among SATS© Attitude Component Scores and Grades*

Affect Cognitive Value Difficulty

Competence

Pre-test** .12 .20 .08 .09

Pre-test*** .04 .14 .06 .03
Post-test .35 .36 .30 .17

* Attitude scores and grades were standardized within section.

** Pre-test scores for students who took the pre-test (n=360).
*** Pre-test scores for students who took the post-test also (n=268).

Figure 1. Global Causal Model Interrelating Students’ Attitudes Toward Statistics and Course
Achievement.

Prior Cognitive Cognitive

Competence Competence

Prior Prior Difficulty Value

Difficulty Value

Prior Affect
Affect

Prior Course
Achievement Achievement

3683