Use of Excel Laboratories in Undergraduate Elementary Statistics
Madison Pilkerton
Faculty Advisor: Jana Asher
BACKGROUND
Undergraduate students enrolled in elementary statistics
courses would rather not be, and these students leave with
little to no retained knowledge. Statistics courses should
engage the student in a meaningful way to create longlasting knowledge. The 2016 report by the American
Statistical Association titled “Guidelines for Assessment and
Instruction in Statistics Education” (GAISE 2016) states that
“students should be given numerous opportunities to analyze
data with the best available technology (p. 11).” In addition,
students should learn statistics as an “investigative process
of problem solving and decision-making (p. 13).” Finally,
GAISE encourages an active learning approach, involving
“students in doing things and thinking about the things they
are doing (p. 18).” Little research exists on the effect of
using laboratory activities in elementary statistics courses
that include the three practices above.

RESULTS

−
−
−
−

SUMMARY STATISTICS FOR ALL STUDENTS
Statistical
Concepts

n

Mean

Variance

16.56

Standard
Deviation

20.02

Standard
Error

4.47

Median

0.26

Range

Post-assessment sum
of scores

291

16

Pre-assessment sum
of scores

313

16.41

15.97

4.00

0.23

16

20

Post – pre

242

-0.32

19.84

4.45

0.29

0

21

Post-assessment sum
of scores

223

2.76

34.56

5.88

0.40

2

27

Pre-assessment sum
of scores

229

7.36

28.87

5.37

0.37

8

23

Post – pre

213

-4.61

34.23

5.85

0.40

-4

33

Pairs of students completed 7 self-paced computer-based
laboratory activities throughout a 15-week semester. Each
laboratory activity replaced one 50-minute lecture-based
instruction period. These laboratory activities were designed to
allow students to explore real datasets, engage in active learning,
and think conceptually about statistical concepts as they analyze
and interpret the data.
Four faculty members at Slippery Rock University teaching
elementary statistics courses were divided into two blocks.
These blocks were based on their teaching methods, teaching
style, gender, and personality.
– One professor in each block was randomly selected to
use the laboratory activities and the other introduced
Minitab through lecture but did not include in-class
practice with Minitab.
– Students in the control group used Minitab for
homework assignments.
Students in the courses were given the CAOS-4 pre-assessment
and post-assessment. Students were additionally given a pre and
post assessment on attitudes towards statistics, based on Araki
(1995).
– Students were encouraged to answer all questions in
the pre-assessment.

86

16.62

13.13

Laboratory group: post - pre

69

0.20

17.52

4.19

0.50

1

Control group: post-assessment

49

15.94

22.14

4.71

0.67

16

24

Control group: pre-assessment

68

16.25

17.95

4.24

0.51

16

16

Control group: post - pre

44

-1.05

22.09

4.70

0.71

-0.5

20

Two-sample t-test

3.62

0.39

t-stat: 1.47

17

18
19

p-value: 0.07

Here we see that the lab group started off higher on average on the
pre-assessment than the control group. However, the mean decreased
for the control group but increased for the lab group on the postassessment. We also noticed that the mean and median differences
are negative for the control group but not the lab group. The
difference between the laboratory group and the control group is
borderline significant, suggesting the laboratories might have a
significant impact on retention of statistical concepts.

RESULTS FOR ATTITUDES TOWARDS STATISTICS

n

Mean

Variance

Standard
Deviation

Standard
Error

Median

62

3.39

29.32

5.42

0.69

3.50

22

Laboratory group: pre-assessment

62

8.06

28.26

5.32

0.68

9.00

21

Laboratory group: post - pre

62

-4.68

31.63

5.62

0.71

-4.00

25

Control group: post-assessment

42

0.00

26

Control group: pre-assessment

42

6.74

34.83

5.90

0.91

7.50

23

Control group: post - pre

42

-4.98

44.71

6.69

1.03

-4.00

26

Two-sample t-test

43.89

t-stat: 0.24

6.63

1.02

students could randomly select correct answers with high

Variance

Standard
Deviation

1.03

2.03

Laboratory group: pre-assessment

62

1.03

1.77

Laboratory group: post - pre

62

0

2.16

Control group: post-assessment

42

3.12

2.20

Control group: pre-assessment

42

3.64

Control group: post - pre

42

-0.52

Two-sample t-test

Standard
Error

probability. Even so the laboratory group appears to have a

1.42

Median

higher average score than the control group.

Range

0.18

1

7

1.33

0.17

1

7

1.47

0.19

0

8

1.48

0.23

4

6

1.06

1.03

0.16

4

6

2.26

1.50

0.23

0

9

t-stat: 1.76

p-value: 0.81

Here we see that the lab group started off higher on average on the
pre-assessment than the control group, and they ended higher than
average on the post-assessment than the control group. The
difference between the post and pre-assessment is similar for both
groups and there is no statistical significance.

− Results for attitudes towards statistics are similar among the
two groups. Unfortunately, attitudes towards statistics appear
to become more negative after taking a statistics course.

LIMITATIONS AND IMPROVEMENTS

p-value: 0.08

Here we see that the lab group started and ended the same, while the control
group scored higher on the pre-attitudinal assessment than on the post-attitudinal
assessment. The difference between the lab and the control group is borderline
significant. There might be better attitudes among the laboratory group than the
control group.

− Comparisons between the lab/control groups and the servicelearning/honors groups might be affected by confounding

RESULTS FOR ACCOMPLISHMENT QUESTIONS ACROSS ALL GROUPS

effects such as professor effects.
− The labs and assessments were run during the Spring 2020

n

Mean

Variance

Standard
Deviation

Standard
Error

Median

Range

Laboratory group: post-pre

62

0.00

2.16

1.47

0.19

0

8

Control group: post-pre

42

-0.52

2.26

1.50

0.23

0

9

Service-Learning group: post – pre

16

0.06

5.13

2.26

0.57

0

10

Honors group: post - pre

22

-1.18

1.68

1.30

0.28

-1

4

ANOVA

F-stat: 3.66

p-value: 0.0142

completed.

with an altered pre-assessment.
− Students do not answer the question if they do not
know the answer.
− Students should attempt to answer the question if they
have learned the concept previously.

RESULTS FOR PROFESSIONAL QUESTIONS ACROSS ALL GROUPS
Variance

Standard
Deviation

Standard
Error

Median

− This should allow a clearer difference to see if the labs

Range

Laboratory group: post-pre

62

-1.82

6.31

2.51

0.32

-2

12

Control group: post-pre

42

-2.02

8.27

2.88

0.44

-2

12

Service-Learning group: post-pre

16

0.19

Honors group: post-pre

22

-1.73

2.30

1.52

0.38

4.40

2.10

0.45

F-test: 3.34

0

6

-2

9

p-value: 0.0211

Here we see that there is a significant difference between the groups and their use
of statistics in their careers.
t-stat

− Originally, there were 10 laboratory activities. Five

− A potential improvement would be running this test again

− Statistics should be a required part of my professional training.
− Statistical skills will make me more employable.
− I am interested in being able to communicate statistical information on to
others.
− Statistics is useful to the typical professional
− I will have application for statistics in my profession.

Mean

online instead of in-class.

end of the course, seven laboratory activities were

PROFESSIONAL ATTITUDINAL SURVEY QUESTIONS

n

semester. Due to COVID-19, the semester had to be finished

laboratories were completed before COVID-19. At the

There seems to be evidence that the lab group and service-learning group have
better professional attitudes than the control group and honors group.

Range

Laboratory group: post-assessment

1.76

Mean
62

ANOVA
n

− Results for statistical concepts is muddied by the fact that

RESULTS FOR ACCOMPLISHMENT ATTITUDE QUESTIONS
Laboratory group: post-assessment

Attitudes
towards
Statistics

Laboratory group: pre-assessment

DISCUSSION

26

It was found that students were doing just as well on average
on the pre-assessment as they were on the post-assessment.
We believe this is from encouraging students to answer all
questions on the CAOS-4. Some of the questions on the
assessment only had two multiple-choice choices, so students
had a 50-50 chance of getting the assessment question correct.
GOALS
We can find an expected value if they were guessing. But we
To improve learning outcomes in statistics education at
cannot adjust by the expected value because we do not know
Slippery Rock University and in the academic discipline of
if the students came to the course with any previous statistics
statistics in general.
knowledge. Variability on the post-assessment increased. We
– To explore the effectiveness of the use of
believe this is from at least some of the students no longer
Microsoft Excel as a statistical software package in guessing on the questions. It was also noted that student
elementary statistics to improve student
attitude towards statistics was higher going into the course
understanding across all learning outcomes.
than when leaving the course on average.
– To explore the use of in-class laboratories as a
SUMMARY FOR STATISTICAL CONCEPTS
high-impact learning practice, through active,
n
Mean Varianc
Standard
Standard
Median Range
collaborative learning in elementary statistics
e
Deviation
Error
Laboratory group: post-assessment
89
17.07
18.09
4.25
0.45
17
19
classes.

METHODS

ACCOMPLISHMENT ATTITUDINAL SURVEY QUESTIONS
I plan/tried to complete all of my statistics assignments.
I plan/tried to work hard in my statistics course.
I plan/tried to study hard for every statistics test.
I plan/tried to attend every statistics class session.

p-value

Service Learning vs. Laboratory: t-test

-4.06

0.0002

Service Learning vs. Control: t-test

-3.79

0.0004

Service Learning vs. Honors: t-test

3.27

0.0024

Here we see that the difference between each learning group and service-learning
is extremely significant. Meaning that the service-learning group feels that
statistics was more relevant to their career than the other groups feel.

are leading to better learning.

REFERENCES AND ACKNOWLEDGEMENTS
− Araki, L. (1995). An exploratory study of student attitudes
toward statistics and their retention of statistical concepts.
− Carver, R., Everson, M., Gabrosek, J., Horton, N., Lock, R.,
Mocko, M., ... & Wood, B. (2016). Guidelines for assessment
and instruction in statistics education (GAISE) college report
2016.
− Ben-Zvi, D., & Garfield, J. B. (2007). Assessing students’
conceptual understanding after a first course in statistics.
International Statistical Review, 75, 372-396.
− The author thanks Danielle Dumbeck, Amanda Goodrick,
Woosuk Kim, James Porter, and Dil Singhabahu for their
participation in the experiment.