16. Computer Software In Statistics Education Student Views On The Impact Of A Computer Package On Affection And Cognition

207
16. COMPUTER SOFTWARE IN STATISTICS EDUCATION:
STUDENT VIEWS ON THE IMPACT OF A COMPUTER PACKAGE
ON AFFECTION AND COGNITION
Gilberte Schuyten and Hannelore Dekeyser
University of Gent
The Data Analysis subdepartment of the department of Psychology and Educational Sciences is
conducting several studies focus on the use of computers in statistics learning. One project investigates the
computer as a learning environment for statistics. In an experimental setting, three different learning
conditions are studied: computer-based independent learning, pen-and-paper-based independent learning,
and lecture-based learning. The results of this research were presented at ICOTS IV Marrakech (Schuyten,
Dekeyser, & Goeminne, 1994).
A second project focuses on student characteristics; that is, individual differences in dealing with verbal
explanations, graphical displays, and formula. Their impact on the learning of statistics is investigated. In
this study, the computer is used not only as a research tool to register the activities of the students in
logfiles, but also as a didactic tool to explore the possibilities of computer-based materials for
differentiation and individualization. Some findings were presented at ISI in Beijing in August 1995
(Schuyten, Goeminne, & Dekeyser, 1995).
A third project, discussed here, focuses on the added value of standard scientific computer packages.
The added value of these packages to the traditional course (consisting of lectures and pen-and-paper
exercises) is discussed. Students' views on the impact of the computer package on affection and cognition
are collected with a structured questionnaire.
INTRODUCTION
The role of computer technology in education was and still is a much debated topic. Computers provide
us with the opportunity to create entirely new learning environments for our students. Computers can be
used in statistics education in many ways. They can be used as an illustration component in lectures, as a
computation tool, as an electronic interactive textbook, as a research tool, and as a medium for thinking.
Many presentations at the ICOTS and IASE meetings have emphasized the benefits of computers in
statistics education. In the early 1980s, Taylor (1980) added a third category "Tutee" to the well-known
categories of "Tool" and "Tutor.” Tutee refered to the capacity of the computer to emphasize
metacognition and to stimulate reflection on the learning processes. Now not only the use of computers to
conduct exploratory data analysis (EDA) is emphasized, but also the ability of computers to provide
multiple representations for stochastic information, particularly the graphical abilities of computers. Thus,
computers provide both an exploratory aspect and a representational aspect (Shaughnessy, 1992). The
Role of Technology, Spain 1996: Gilberte Schuyten and Hannelore Dekeyser

G. SCHUYTEN & H. DEKEYSER
208
requirements for an ideal software tool for supporting learning and doing statistics are given by Biehler
(1993).
What about computer software in applied statistics courses for humanities and social sciences students?
Not only do we want our students to learn statistics, but also we want them, as potential future users, to learn
how to use a professional statistical package. In Flemish universities and at the Department of Psychology
and Educational sciences, SPSS is generally accepted as a special purpose tool for doing statistics. The
reasons for this are twofold: (1) students will use SPSS later in their careers; this will be true for those
conducting research as well as for those in other professions where more and more data analysis is needed;
and (2) technology is used as an amplifier; that is, it provides students more practice in less time, because it
takes the burden of computing away from them.
STATISTICAL SYSTEM FOR DATA ANALYSIS AS A
COGNITIVE TECHNOLOGY FOR STATISTICS EDUCATION
The use of a statistical system for data analysis not only affects student skills in handling the program
but also their competence in statistics. It acts as a concrete tool and as "a representation and an object to
think with" (Biehler, 1993, p.179). Following Pea (1987), "a cognitive technology is any medium that helps
transcend the limitations of the mind (e.g., attention to goals, short-term memory span) in thinking,
learning, and problem-solving activities" (p. 91). The ability to make external the intermediate products of
thinking offers the possibility to analyze, to reflect on, and to discuss the thinking process.
In order to assess existing software and to guide assignments for computer workshops, the taxonomy of
functions of educational technologies in mathematics education, as proposed by Pea, are also useful in
statistics education. He distinguishes purpose functions, which may affect whether students choose to think
mathematically, from process functions, which may support the component mental activities of
mathematical thinking. Pea (1987) sees three purpose functions that could help strengthen intrinsic
motivation: (1) the student has ownership of the problem, (2) learning events should be viewed as
opportunities for acquiring new understanding and thus promote self-worth, and (3) knowledge for action,
which means that knowledge should empower students to understand or do something better than they
could prior to its acquisition. The five process functions dealing with the cognitive support provided to
thinking include: (1) tools for developing conceptual fluency, (2) tools for mathematical exploration, (3)
tools for integrating different mathematical representations, (4) tools for learning how to learn, and (5)
tools for learning problem-solving methods. In assessing software for students enrolled in applied statistics
courses, the purpose functions are of special interest, because those students are heterogeneous in terms of
their mathematical background, most are not oriented towards the positivist paradigm emphasizing
empirical quantifiable observations (Schuyten, 1991), and most lack confidence in mathematical thinking.
Small project work on data concerning themselves could affect the students’ ownership; that is, the success
in running a program could affect their self-worth and empower them.
As for the process functions, the first, third, and fourth are particularly interesting in assessing
professional software. In considering the first process function (conceptual fluency), the software program
frees the students from conducting tedious calculations and from the possibility of making calculation
errors, and frees up the mental resources of the students so that they can explore procedures and guide their
learning. In terms of the third process function, the value of linking multiple representations for
understanding was reported in the second study. However, a supplementary problem concerns the question:

16. COMPUTER-BASED AND COMPUTER-AIDED LEARNING OF APPLIED STATISTICS
209
"Is a statistical concept as represented in the software equivalent with that in an ordinary textbook?" This
will be discussed below using the concept of a data matrix. For the fourth process function, the program
traces students' problem solving and promotes reflective learning, the program monitors and assesses, the
program helps students to control their activities, and the program helps the students learn how to learn.
COGNITIONS ASSOCIATED WITH DATA MATRIX
In his notation-representation foundations, Kaput posits a world of mental operations, which is always
hypothetical, and a world of physical operations, which is frequently observable; these two worlds interact
(Kaput, 1992). These physical operations deal with notation systems that are sets of rules, which are
considered abstract until we decide to instantiate it using the material world (i.e., the physical medium)
(Kaput, 1992). What are the cognitions associated with a data matrix as represented in traditional textbooks
and as represented in professional software? How are these cognitions integrated? The classical notation of
the object "data matrix" consisting of rows and columns in applied statistics courses is mainly used to
display information. It is presented as a tool to organize data and as such it is connected with the object
"tabulation." Later, by studying association, the object "contingency table" is introduced. Next, by studying
differences between means of groups, the ideas of "dependent and dependent groups" become relevant. The
cognitions associated with data matrix, tabulation, cross-tabulation, and dependent and independent groups
need to be integrated. The notation systems used to express these concepts interfere, and students become
confused and often ask, "Aren't they just rows and columns? What is the difference?"
Let us illustrate this using a simple data matrix. The data matrix in Table 1 consists of 6 units of analysis
(cases) and 4 variables: sex, a question with a simple yes or no answer, pretest results, and posttest results.
Except for the first column, which contains the identification codes, the body of the table consists of
measurements. For the concept of assocation, crosstables are presented. In crosstables, the categories are the
row and column headings, and the body of the table contains the frequencies of each combination of
categories for the two studied variables. The crosstabulation shown in Table 2 still consists of rows and
columns, but their meaning is different from those in Table 1 (the data matrix), and the body of the tables
are different.
Hypothesis testing using two means (t-test) deals with two different situations: (1) the groups of
measurements are independent (Table 3; i.e., "Is there a difference between the mean of the three men and
the mean of the three women on pretest scores?") and (2) the groups of measurements are dependent
(Table 4; i.e., "Do the means of pretest and posttest scores of the six cases differ?"). For the independent
situation, pen-and-paper exercises are mostly presented as two columns of data and the rows are not
relevant; in the dependent situation, rows are relevant and the two columns are taken out of the data matrix.
The foregoing illustrates why many students have problems reorganizing the data represented in a data
matrix into representations such as crosstabs or groups of measurements or vice versa. Many students lack
flexibility in changing the view and cannot cope with different representations of the same data. However,
what is the added value of the electronic data matrix? Why should the notation of a data matrix be more
powerful in an electronic medium than in a pen-and-paper medium? Is the electronic data matrix more
flexible in changing the view?

210
Table 1: The data matrix
ID Sex Pretest Posttest Question
1 1 6 6 1
2 2 4 7 0
3 1 8 7 1
4 1 5 9 0
5 2 6 7 1
6 2 7 6 1
Table 2: Contingency table of question with sex
Sex
Question 1 2
0 1 1
1 2 2
Table 3: Difference of means on pretest scores between men and women
Sex = 1 Sex = 2
6 4
8 6
5 7
Table 4: Difference of means between pretest and posttest scores
Pretest Posttest
6 6
4 7
8 7
5 9
6 7
7 6
EMPIRICAL RESEARCH
Research design
Little is known about the educational impact of cognitive technology. Empirical testing of the
exploratory new instructional curricula that embody the various functions Pea described is necessary (Pea,
1987). Case studies, effect studies of software, and evaluation of new curricular materials in natural settings
of teaching are needed.
The applied statistics courses at the Faculty of Psychology and Educational Sciences consist of four
components: lectures, pen-and-paper exercices, an independent learning introduction module to SPSS, and
SPSS exercices. In this natural setting of a university course, it is not possible to isolate the effect of the

211
software program. In order to explore the added value of computer software, a survey was conducted
concerning student perception of the impact of the software on learning and doing statistics.
The statistics curricula
The applied statistics course is a two-year course. The first-year course deals with descriptive statistics,
association, probability distributions, hypothesis testing, and estimation. The second-year course deals with
effect size, testing and estimating measures of association, nonparametric tests, and parametric tests (t-test
and one-way analysis of variance). In the first year, students get started with SPSS. The practical exam takes
place at the beginning of the second year. This practical exam focuses on minimal competence needed to
use SPSS (e.g., constructing a SPSS command using the SPSS menu, knowing how to use the glossary,
saving a textfile in the editor Review, and so forth).
In the independent learning guide for SPSS, every step of the process of analyzing data with SPSS is
based on a questionnaire that is filled out by the students at the start of the academic year. It contains
questions on motivation for enrollment and attitude towards statistics, research, and computers. Every
analysis is performed on these data and is discussed in terms of research question, hypotheses (if relevant),
and content-linked interpretation. The first year independent learning guide contains an introduction to the
computer program SPSS/STUDENTWARE and SPSS/PC+ (release 5.0), the SPSS editor REVIEW, the SPSS
Menu, the output file SPSS.LIS, data preparation (units of analysis, variables, values, construction, and entry
of the data matrix), an introduction to elementary descriptive statistics (frequency tables and statistics,
barchart, and histogram), elementary data transformation (selection, recoding), bivariate analyses (crosstabs,
regression analysis, Pearson's correlation), and plotting. The second year package deals with saving the data
matrix and data definition into a SPSS system file, and using nonparametric tests and analysis of variance.
The students
In October 1994, 550 students were enrolled in the first course: In October 1995, 388 of those 550
students were enrolled in the second course. The practical exam, administered in October 1995, was taken
by 375 of those 388 students. The questionnaire was filled out by 193 of those 388 students attending a
lecture in April 1996.
The questionnaire
Students were questioned about their perception concerning the impact of SPSS on their attitude toward
statistics and quantitative research, on self confidence, on their meta-cognitive skills, on how SPSS enhanced
their understanding of statistical concepts, and on what they perceived as the added value of SPSS as a tool
for "doing statistics." The impact of SPSS on attitude was measured by judgments made on a 5-point scale,
where 1 = Don't agree at all and 5 = Do agree totally. Students rated their perception of the positive impact
of working with SPSS on their attitude towards statistics, on their self-confidence in doing statistics, on their
interest in quantitative scientific research, and on the metacognitive skills needed to analyze the research
question, to select a procedure, and to guide and control their thinking process. The impact of SPSS on
enhancing understanding of statistical concepts was measured for 30 statistical concepts using three
response categories--promotes understanding, doesn't affect understanding, and hinders understanding.

212
From the list of 30 statistical concepts, 10 concepts were considered by the authors as irrelevant for
assessing the impact of SPSS on enhancing understanding of statistical concepts taking into account the
instructional curricula. However, these 10 concepts were included to provide information concerning the
validity of the instrument. A concept was considered irrelevant if the SPSS activities did not focus on the
concept (e.g., standard deviation, critical region, and so forth) or if the SPSS activities were not related at all
to the concept (e.g., effect size, expectation, and so forth). The list was divided into two sublists--Form A
and Form B. Each student gave his/her opinion about 15 concepts, 5 of which were supposed to be
irrelevant. The variable "tool for learning statistics" was created from the 15 concepts. This variable
indicates how frequent the “promotes understanding” category was crossed in the list of 15 concepts (scale
0-15). Students were also asked to give their global opinion about the impact of SPSS on enhancing their
understanding of statistical concepts. The perceived added value for "doing statistics" was measured by 14
items to which the students could answer yes, no, or don't know. Based on these items, the variable "tool for
doing statistics" was created. Students were also asked about their most positive and negative experiences in
working with SPSS and their fluency in handling the software.
Background information
The results of the practical exam (which is graded on a scale of 0-5) administered in October 1995 are
available as well as data concerning the motivation and attitude toward statistics collected from the students
in October 1994. Because the practical exam focuses on the minimal competence needed to work with
SPSS, a score of less than 3 means that the minimum competency requirements have not been met. Only
11% of the students tested did not manage to reach that level. Based on this score, two groups of students
were formed: the high competency group with scores of more than 4 (n = 99) and the low competency
group with scores of less than 4 (n = 40). Students with a score of 4 (n = 54) were excluded from the
analysis in which high and low comptency groups were compared.
Results
The impact of SPSS on attitude toward statistics and on metacognitive skills
The results of the questionnaire administered in October 1994 indicate that the motivation of the
students to enroll in courses at the Faculty of Psychology and Educational Sciences is based more on
"helping people" [mean (M) = 3.9; standard deviation (SD) = .94; on a scale of 1-5] than on "doing
scientific research" (M = 2.6; SD = 1.32; on a scale of 1-5). The impact of SPSS on attitude toward statistics
(Table 5) was minimal (M = 2.9). The least positive impact was on stimulating interest in quantitative
scientific research (M = 2.4); that is, 54% don't agree and only 17% felt SPSS was beneficial. There was
also a small effect on metacognitive skills (see items Analyse a Question, Select a Procedure, and Control of
Thinking Process in Table 5). The highest percentage of agreement was for "select a procedure" (33%), but
this item also has a high non-agreement percentage of 39%. The most positive impact wa on attitude toward
statistics with 32% non-agreement and 28% agreement.
Analysis of variance showed significant differences between the high and low competency groups for
"statistics" (p = .012), "selfconfidence" (p = .026), and "control of thinking process" (p = .011). The means
for the high competency group were higher.

213
Table 5: Questionnaire results
Note: Judgments were made on 5-point scores where 1 = Don’t agree at all and 5 = Do totally agree; Categories 1 and 2 were
collapsed into “Don’t agree” and categories 4 and 5 were collapsed into “Agree”
Item Impact
Don’t Agree -- Agree Mean SD
Statistics 32% 40% 28% 2.9 1.0
Self-Confidence 46% 35% 19% 2.6 1.0
Quantitative Research 54% 29% 17% 2.4 1.1
Analyse a Question 37% 35% 28% 2.8 1.0
Select a Procedure 39% 28% 33% 2.9 1.1
Control of Thinking Process 39% 35% 27% 2.8 1.0
Fluency in SPSS
Judgments were made on a 5-point scale concerning the students’ perception of their fluency with SPSS.
Only 23% of the students feel they can manage SPSS; 35% feel they cannot. Correlations with the
affection/conation items were all significant, and the highest was with statistics (r = .41). The correlation
between perception of fluency and the practical exam was not significant (r = .14)
Impact on understanding statistical concepts
Students were asked to indicate whether working with SPSS had an effect on the understanding of 15
statistical concepts. The categories presented were promotes understanding, doesn't affect understanding,
and hinders understanding. In Table 6, concepts are ordered following descending positive impact on
understanding, measured by frequency of the category promotes understanding. The last column indicates
the authors’ opinions concerning the relevance of the possible impact, taking into account the curricular
materials.
The positive impact on the concept of data matrix is clear; this concept is followed by frequency table,
histogram, variable, and frequency distribution. Also, an understanding of the research question is
stimulated by SPSS (51% of the students). The concepts dependent and independent variable are clarified
for 46% and 42% of the students. These findings about the impact of the software cannot be isolated from
the impact of the curricular materials associated with the software. The impact on the concept of analysis of
variance is very low, which can be explained by the fact that SPSS-work associated with this topic was done
after the survey. These findings also indicate that more SPSS-work has to be integrated into the curricular
materials concerning dependent and independent groups. The item concerning the global impact on
understanding was answered in a positive way: 59% agree that SPSS enhanced understanding, only 2% felt
that SPSS hindered understanding. The result for the variable "tool for learning statistics" is also rather
positive (M = 5.43, SD = 2.82), taking into account that 5 of the 15 concepts could be considered irrelevant
for the impact of SPSS. The impact of the software on understanding statistics is not affected by students
minimum competence skills in handling the software. The means of the low competency and high
competency groups were not significantly different.

214
Table 6: Positive impact of SPSS on understanding statistical concepts
(r = relevant concept, i = irrelevant concept)
Table 6A % r/i Table 6B % r/i
Data Matrix 94 r Histogram 73 r
Frequency Table 73 r Frequency Distribution 67 r
Variable 69 r Research Question 51 r
Unit of Analysis 56 r Dependent Variable 46 r
Independent Variable 42 r Measurement 40 r
Mean 37 r Variable Concept 38 r
Independent Groups 32 r Dependent Groups 34 r
Correlation Coefficient 32 r Linear Regression 34 r
Level of Measurement 30 r Median 30 r
Standard Deviation 23 i Probability Distribution 30 i
Critical Region 18 i Significant Result 27 r
Standard Error 16 i Variance 20 i
Decision Rule in Hypothesis Testing 14 i Critical Value 18 i
Analysis of Variance 13 r Expectation 17 i
Effect Size 3 i Standard Score 15 i
Impact on doing statistics
Students were asked to indicate if in their opinion SPSS-work was positive or negative for "doing
statistics." The 14 items in Table 7 are ordered following descending positive impact on doing statistics. The
top two include "Creating a data matrix" and "Saves you from calculations." The typical software
characteristics "Error messages concerning thinking process" and "Menu-help for procedures" are
appreciated by 2/3 of the students. Other process functions supporting mental activities such as "Error
messages concerning typing," "Translating the research question into SPSS-language," and "Error messages
concerning file-management" have a rather high number of students (20%, 18%, and 17%, respectively)
expressing a negative impact on doing statistics. The underlying reason for this may be that these functions
are positive when working with SPSS, but are irrelevant when doing statistics without a computer.
Depending on how a student interprets it, he/she may answer yes or no. This could also explain the different
appreciation for "Analysing research question" (74% positive) and "Translating the research question into
SPSS-language" (52% positive). The item "Saves you from calculations," which refers to the first process
function proposed by Pea (1987; i.e., tools for developing conceptual fluency), was appreciated by 4/5 of
the students. The item "Own data," which refers to the first purpose function proposed by Pea (i.e.,
ownership), was appreciated by 73% of the students. The summing up of positive reactions produces the
variable "tool for doing statistics" (on a scale of 0-14). Student opinion concerning the impact of SPSS on
doing statistics was positive (M = 9.2, SD = 2.4, mode = 10). There was no effect of minimum competence
on "tool for doing statistics" (p = .091).
Relation between affective variables and cognitive variables
Table 8 shows that the impact of SPSS on understanding statistics is correlated with the three
metacognitive items, attitude, and selfconfidence, but it is not related with stimulating quantitative research
nor with impact of SPSS on doing statistics. The impact of SPSS on doing statistics is only related to

215
attitude towards statistics (Item 1) and self-confidence (Item 2), not to metacognitive aspects (Items 4, 5, and
6). Although the students found several aspects of SPSS positive for doing statistics, this is not related to a
change in attitude towards quantitative research (Item 3).
Table 7: Impact of SPSS on doing statistics
Items % Positive % Negative
Creating a data matrix 83 4
Saves you from calculations 81 5
Analysing research question 74 3
Own data 73 3
You notice when “it doesn’t work” 69 11
Stimulates to look for errors 69 8
Error messages concerning thinking process 68 6
Menu-help for procedures 68 4
Interpretation of an output 68 10
Error messages concerning typing 62 20
Linking up theory and practice 61 10
Translating the research question into SPSS-language 52 18
Working together in the computerlab 47 12
Error messages concerning file-management 44 17
Table 8: Intercorrelations Between affection, conation, and cognition (n = 191)
Variable 1 2 3 4 5 6 7 8
1. Statistics -- .66** .42** .41** .40** .46** .23** .19**
2. Self confidence -- .50** .39** .38** .39** .19** .19**
3. Quantitative research -- .28** .21** .37** -.00 .13
4. Analyse question -- .69** .64** .28** .13
5. Select procedure -- .65** .31** .17
6. Thinking process -- .29** .17
7. Impact understanding -- .15
8. Impact doing --
Note: 2-Tailed Significance * = .01 **= .001
Discussion
The information provided by the students concerning the impact of the statistical software package
should be treated carefully, because (1) software cannot be isolated from curricular materials; (2) student
characteristics could clarify why some benefit and others do not; and (3) the questionnaire used here needs
to be refined and open-ended items that probe for how and why explanations should be added. Tentative
conclusions are that the software has a positive effect on understanding and doing statistics. It seems to
support the component mental activities of statistical thinking. The results concerning the idea that the
software enhances intrinsic motivation were not encouraging. Working with their own data helps the
students, but their selfconfidence was not enhanced and their attitude toward quantitative research was not
influenced in a positive way. It is rather promising that approximately 30% of the students express a
favorable change in attitude toward statistics and in metacognitive skills. Cognitions associated with the data
matrix are stimulated by the software package. The data matrix ranks first for understanding statistics and

216
for doing statistics. In order to answer the questions stated above in Sections 2 and 3 (i.e., "Is a statistical
concept as represented in the software equivalent with that in an ordinary textbook?" and "Is the electronic
data matrix more flexible in changing the view?") more research and other research strategies are needed.
REFERENCES
Biehler, R. (1993). Cognitive technologies for statistics education: Relating the perspective of tools for learning and of
tools for doing statistics. In L. Brunelli & G. Cicchitelli (Eds.), International Association for Statistical Education,
Proceedings of the First Scientific Meeting (pp. 173-190). University of Perugia.
Kaput, J. J. (1992). Technology and mathematics education. In D.A.Grouws (Ed.), Handbook of research on mathematics
teaching and learning (pp. 515-556). New York: Macmillan.
Pea, R. D. (1987). Cognitive technologies for mathematics education. In A. H. Schoenfeld (Ed.), Cognitive science and
mathematics education (pp. 89-122). Hillsdale: Erlbaum.
Schuyten, G. (1991). Statistical thinking in psychology and education. In D. Vere-Jones (Ed.), Proceedings of the Third
International Conference on Teaching Statistics (Vol. 2, pp. 486-489). Voorburg, The Netherlands: International
Statistical Institute.
Schuyten, G., Dekeyser, H., & Goeminne, K. (1994). Manipulating the teaching-learning environment in a first year
statistics course. In Proceedings of the Fourth International Conference on Teaching Statistics (Vol. 1, p. 290).
Marrakech: ICOTS.
Schuyten, G., Goeminne, K., & Dekeyser, H. (1995). Student preference for verbal, graphic and symbolic representation of
information. In Proceedings of the 50th Session Tome LVI Book 4 (p. 1997). Beijing: International Statistical
Institute.
Shaughnessy, J. M. (1992). Research in probability and statistics: Reflections and directions. In D. A. Grouws (Ed.),
Handbook of research on mathematics teaching and learning (pp. 465-494). New York: Macmillan.
Taylor, R. (1980). The computer in the school: Tutor, tool, tutee. New York: Teacher's College Press.

16. Computer Software In Statistics Education Student Views On The Impact Of A Computer Package On Affection And Cognition

More Related Content

Similar to 16. Computer Software In Statistics Education Student Views On The Impact Of A Computer Package On Affection And Cognition (20)

More from Cheryl Brown (20)

Recently uploaded (20)

16. Computer Software In Statistics Education Student Views On The Impact Of A Computer Package On Affection And Cognition