With the growth and emergence of online education, more and more people are using videos on various platforms to learn better. Designing relevant videos is essential to Instructional Designers and educators all over the world. This is where data-driven decisions will facilitate a more effective teaching and learning process by deciding which factors help create an effective video, beginning with its length.
2. ANALYSIS OF THE IMPACT OF VIDEO
LENGTH ON STUDENT PERFORMANCE
• 26 students were assessed on a topic ‘Measures of Central
Tendency”, specifically – Mean, Median and Mode.
• They were tested on the topic and each student viewed a
YouTube video, followed by a similar test. Then, each student
viewed a YouTube short(less than 60 seconds) on the same topic
and was tested again.
• The same set of students were pretested and each student was
first shown a YouTube short and tested again. Then each was
shown a longer video on the same topic followed by another test.
• The following were the variables-
• Age
• Video Length
• Number of likes
• Number of comments
• Pretest Score
• First Score
• Final Score
3. THE PROBLEM STATEMENT
• For Student Sample 1 – Pretest Score -> First Score
(Longer Video) -> Final Score (YouTube Short)(Max Score =
5)
• For Student Sample 2 – Pretest Score -> First Score
(YouTube Short) -> Final Score (Longer Video) (Max Score =
5)
• Average Video Length was calculated for both the sample
sizes.
• The problem – Is video length a significant factor that
affects learning ? … more specifically, is final score
dependent on average video length?
4. DATA COLLECTION
• The students of both samples were tested using Google Forms and the
resulting data was mapped in MS Excel.
• Using the .csv file , the data was analysed in R to comprehend the correlation
between the variables.
• Link to quiz 1 for student sample 1 -Mean, Median and Mode - 1 - Google
Forms – Dataset MMM.csv
• Link to quiz 2 for student sample 2 - Mean, Median and Mode - 2 - Google
Forms – Dataset MMM2.csv
• Similar videos were shared with the students of both the samples.
5. LINEAR REGRESSION
• The data was cleaned, there were no null values.
• Correlational coefficients were computed and corrplots and scatterplots were analysed.
SAMPLE 1 SAMPLE 2
11. KEY OBSERVATIONS -1
• In both samples, the correlation between average video length and final score is not very
significant. However, it was observed that the first score(after viewing the video
irrespective of length) has higher correlation with the final score.
• We can conclude that an indirect relationship exists between average video length and
student performance , basis the correlation between first score and final score.
• The correlation between average video length and student performance is positive in the
first sample and negative in the second sample. So shorter videos were contributing to
better performance and vice versa. However, the relationship is not very significant.
(Correlation coefficients near 0)
• A paired t-test was further decided upon to find out if the means of the final score were
similar or not in both student samples.
12. KEY OBSERVATIONS - 2
• But before that, the samples were checked for
normality using the Shapiro-Wilk normality test.
The p-value was low for all variables except for
average video length of the second student sample.
• Histograms and boxplots were also utilized to test
for normality.
• The Wilcoxon Signed Rank test was used for testing
the two samples as data were not normal.
13. KEY OBSERVATIONS - 3
• As the p-value is less than
0.05, we can reject the null
hypothesis.
• Hence, we can conclude that
there is no statistically
significant impact of video
length on student
performance as the mean
final score of the two student
samples is not zero.
14. CONCLUSION
• Microlearning is a common trend in the learning space of
today. However, video length is not a very significant factor
in determining student success.
• Reduced extraneous load and enhanced intrinsic load help
in facilitating concept retention.
• Effective instructional design enabled students to score
well irrespective of the video length. Hence, video length is
not as significant to facilitating student performance.
• However, the sample size is low – 26. Larger datasets
would help in further analysis.
• The content quality and relevance of the video is more
important than its duration.
15. FURTHERMORE….
• We may predict the final score on
the basis of the first score (sample
1 correlation coefficient 0.627)
using a linear regression model,
as first score is correlated to final
score.
• Findings – The model has a low p-
value so it is statistically
significant. We proceed with
predicting the final score based on
the first score.
16. PREDICTION OF THE FINAL SCORE
2.0 2.5 3.0 3.5 4.0 4.5 5.0
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Regression for Final Score on First Score
Final Score
FirstScore
17. PREDICTION OF THE FINAL SCORE
• Predicting the Final Score on the basis
of First Score may help instructional
designers to use relevant creative
videos to assess performance and
return to the storyboard in case of
anomalies.
• The Mean-Squared Error (MSE) and
Squared Root of MSE are greater than
1 indicating significant predictions
that may highlight utility while
designing online educational content.