Analyzing Data Across Demographic
Groups To understand how various
demographic factors affect the
utilization and perception of AI in
educational settings, we categorize
survey responses and conduct
relevant statistical analyses.
Demographic Factors: Age: Given
that most participants are aged 18-
24, we can differentiate between
undergraduate and graduate
students to assess their usage and
attitudes towards AI. For instance:
Are undergraduate students more
inclined to utilize AI tools for tasks
like note-taking and problem-solving
than their graduate counterparts? Do
younger students (ages 18-24)
interact with AI tools more often than
older students (ages 25 and above)?
Parametric Test: If the data follows a
normal distribution, an independent
t-test can be employed to compare
the means of the two groups (e.g.,
undergraduates vs. graduates). Non-
parametric Test: In cases where the
data does not meet normality
assumptions, the Mann-Whitney U
test would be a more suitable option
for group comparisons. Gender: The
respondent distribution (62.1% male,
33.3% female, and 3.4% non-binary)
can provide insights into any gender-
related differences in the use or trust
in AI tools. For example: Do males
utilize AI tools more often than
females? Do females exhibit greater
trust in AI-generated responses
compared to males? Parametric Test:
An independent t-test is appropriate
for comparing the mean usage
frequency of AI tools between male
and female students, assuming the
data is normally distributed. Non-
parametric Test: If normality is not
satisfied, the Mann-Whitney U test
can be used to analyze differences in
usage or trust levels between
genders. Academic Status
(Undergraduate, Graduate, Doctoral
Students): This variable enables us to
evaluate the influence of AI tools
across different academic tiers. For
example: Do undergraduate students
report a higher frequency of AI tool
usage for memory recall compared to
graduate or doctoral students? How
do the impacts of AI on critical
thinking and decision-making differ
across academic levels?
Parametric Test: A One-Way ANOVA
is suitable for comparing the average
AI usage among three or more
groups (undergraduate, graduate,
and doctoral students), provided the
data follows a normal distribution.
Non-parametric Test: If the data does
not follow a normal distribution, the
Kruskal-Wallis test serves as a good
alternative for comparing the
medians of the three groups. Fields
of Study (Business, Science &
Technology, Medicine & Healthcare):
Students from various fields may
utilize AI tools differently based on
their academic focus. For instance:
Do business students rely on AI tools
for decision-making more than those
studying healthcare or
science/technology? Is there a
noticeable difference in how creative
students from different disciplines
feel when using AI tools? Parametric
Test: Once again, a One-Way ANOVA
would be appropriate for assessing
the average AI usage or its effects
across the three fields, assuming the
data is normally distributed. Non-
parametric Test: The Kruskal-Wallis
test would be the method of choice if
the data does not conform to
normality. 2. Parametric Tests These
tests are effective when the data
meets the criteria of normality
(following a bell-shaped curve) and
homogeneity of variance (similar
data spread across groups). They are
generally used to compare means of
continuous data. Independent t-test:
This test evaluates the means of two
independent groups. It’s beneficial
for comparing responses between
two groups, such as: The average
frequency of AI usage between male
and female students. The average
effect of AI on critical thinking skills
between undergraduates and
graduates. One-Way ANOVA: This
test assesses the means across three
or more groups. It’s useful for
comparing more than two groups:
The influence of AI on problem-
solving efficiency among students
from different academic fields (e.g.,
business, science, medicine). The
impact of AI on memory recall among
undergraduates, graduates, and
doctoral students. Paired t-test (if
applicable): If the study involves pre-
and post-comparisons, a paired t-test
would be appropriate.
3. Non-Parametric Tests Non-
parametric tests are employed when
the assumptions of normality are not
met, such as in cases of skewed data
or ordinal measurements. These
tests do not rely on a specific
distribution and are particularly
effective for comparing medians or
overall distributions. Mann-Whitney U
Test: This test assesses the medians
of two independent groups and
serves as an alternative to the
independent t-test when the data
does not follow a normal distribution.
Example: - Analyze the differences in
how males and females perceive the
influence of AI on creativity. -
Examine the frequency of AI usage
among undergraduate and graduate
students when the data does not
conform to a normal distribution.
Kruskal-Wallis Test: This test is the
non-parametric counterpart to the
One-Way ANOVA and is used to
compare the medians of three or
more groups. It is beneficial for: -
Assessing the effect of AI on critical
thinking among students in business,
science, and healthcare disciplines. -
Investigating the influence of AI on
memory recall across various
academic levels (undergraduate,
graduate, doctoral). Wilcoxon Signed-
Rank Test: Designed for paired data
(such as pre- and post-survey
responses from the same group), this
test evaluates differences in ranks
and serves as the non-parametric
equivalent of the paired t-test.
Example: - Investigate the impact of
AI tools on academic decision-making
before and after the incorporation of
AI into the curriculum. 4. Practical
Example: If you aim to compare the
effect of AI on problem-solving
efficiency between undergraduate
and graduate students, the following
approaches can be taken: Parametric
Approach: Assuming the data is
normally distributed, an independent
t-test could be utilized to compare
the means of AI's effect on problem-
solving efficiency between the two
groups. Non-Parametric Approach: In
cases where the data does not follow
a normal distribution, the Mann-
Whitney U test could be applied to
compare the medians of the two
groups regarding the same metric.
5. Visual Representation of
Comparisons: Prior to conducting any
statistical analyses, it is beneficial to
visualize the data to gain insights
into its distribution. Common
visualization techniques include: Box
plots or histograms, which are useful
for evaluating the data's distribution
and assessing normality. Bar charts
or scatter plots, which facilitate the
comparison of differences in AI usage
and its effects on students.
Conclusion: To effectively analyze
the survey data, it is crucial to
choose the appropriate statistical
test that aligns with the
characteristics of the data.
Parametric tests, such as the
independent t-test and One-Way
ANOVA, are appropriate for
comparing means when the data
adheres to a normal distribution.
Conversely, non-parametric tests,
including the Mann-Whitney U test
and Kruskal-Wallis test, are
applicable when the data is skewed,
ordinal, or does not satisfy normality
criteria. These statistical methods will
assist in determining whether
significant differences exist in AI
usage
and its effects across various
demographic categories, such as
age, gender, academic level, and
field of study. By selecting the
correct statistical approach, we can
enhance our understanding of the
perceptions and utilization of AI tools
within academic environments.
