Math exploration

Be active on Instagram but study

Introduction
Social media such as Instagram is one of the latest online avenues for social
interactions, sharing stories and pictures of everyday life. It is noticed that today’s
young people spend rather a lot of time on social media, so less time remains for other
activities and first of all for studying, as well as reading, watching TV or playing
outdoors. The problem involves time management, so it seems desirable to understand
the actual relationship between the time spent by students on updating their feed on
Instagram and students’ academic performance.

Statement of task
The project is intended to study to what extent the academic grade of IB
students is affected by the time spent on Instagram. It is natural to assume that in
general, a student who spends more time on Instagram shows poorer performance.
Hence, the alternative hypothesis is
H1: time spent on Instagram and academic grade are dependent,
from which automatically follows the null hypothesis:
H0: time spent on Instagram and academic grade are independent.

Data collection
A random sample consisting of 26 IB students from 10-12 grades was selected
and each student was asked to record how many minutes during a day on average they
spend on Instagram and their average academic grade. The results are summarized in
the table

Time spent on Average Time spent on Average

# Instagram during academic # Instagram during academic
a day / minutes grade a day / minutes grade
1 43 5 14 86 5
2 40 7 15 180 5
3 120 5 16 15 6
4 45 6 17 10 7
5 30 7 18 25 6
6 0 7 19 25 6
7 0 7 20 0 7
8 0 7 21 5 6
9 90 5 22 22 7
10 60 6 23 38 5
11 55 7 24 35 6
12 48 7 25 35 6
13 31 6 26 15 6
Data processing
In given investigation, both introduced variables are quantitative since their
values are obtained through measurement (time spent) and estimation (average grade).
Farther, the time spent is considered as the independent variable and is denoted x,
while the average grade is treated as the dependent variable and is denoted y.
To depict the relation between variables the data from the above table is
plotted

8
Relationship between average grade and
7
time spent

5
Average grades

y = -0.0125x + 6.6774
4 R² = 0.4209

0
0 50 100 150 200
Time spent / minutes per day

The scatter diagram shows a moderate negative relationship between variables. So we

can say that in general the more minutes are spent on Instagram the less is grade. To
reveal the strength of relationship however more detailed analysis is needed.

The correlation coefficient allows us to estimate the strength of the

relationship between variables and is defined as the covariance divided by the product
of standard deviations. So to calculate the correlation coefficient we can use the
formula
∑(𝑥 − 𝑥̅ )(𝑦 − 𝑦̅)
𝑟=
√∑(𝑥 − 𝑥̅ )2 √∑(𝑦 − 𝑦̅)2

The table below contains the results of performed computations

 𝑥 𝑦 𝑥 − 𝑥̅ 𝑦 − 𝑦̅ (𝑥 − 𝑥̅ )2 (𝑦 − 𝑦̅)2 (𝑥 − 𝑥̅ )(𝑦 − 𝑦̅)
1 43 5 1.56 -1.16 2.4336 1.3456 -1.8096
2 40 7 -1.44 0.84 2.0736 0.7056 -1.2096
3 120 5 78.56 -1.16 6171.674 1.3456 -91.1296
4 45 6 3.56 -0.16 12.6736 0.0256 -0.5696
5 30 7 -11.44 0.84 130.8736 0.7056 -9.6096
6 0 7 -41.44 0.84 1717.274 0.7056 -34.8096
7 0 7 -41.44 0.84 1717.274 0.7056 -34.8096
8 0 7 -41.44 0.84 1717.274 0.7056 -34.8096
9 90 5 48.56 -1.16 2358.074 1.3456 -56.3296
10 60 6 18.56 -0.16 344.4736 0.0256 -2.9696
11 55 7 13.56 0.84 183.8736 0.7056 11.3904
12 48 7 6.56 0.84 43.0336 0.7056 5.5104
13 31 6 -10.44 -0.16 108.9936 0.0256 1.6704
14 86 5 44.56 -1.16 1985.594 1.3456 -51.6896
15 180 5 138.56 -1.16 19198.87 1.3456 -160.73
16 15 6 -26.44 -0.16 699.0736 0.0256 4.2304
17 10 7 -31.44 0.84 988.4736 0.7056 -26.4096
18 25 6 -16.44 -0.16 270.2736 0.0256 2.6304
19 25 6 -16.44 -0.16 270.2736 0.0256 2.6304
20 0 7 -41.44 0.84 1717.274 0.7056 -34.8096
21 5 6 -36.44 -0.16 1327.874 0.0256 5.8304
22 22 7 -19.44 0.84 377.9136 0.7056 -16.3296
23 38 5 -3.44 -1.16 11.8336 1.3456 3.9904
24 35 6 -7.44 -0.16 55.3536 0.0256 1.1904
25 35 6 -7.44 -0.16 55.3536 0.0256 1.1904
total 1036 154 41468.16 15.36 -517.76
mean 𝑥̅ = 41.44 𝑦̅ = 6.16

The 𝑥̅ and 𝑦̅ represent the mean values of the independent and dependent variables
respectively:
∑𝑥 1036 ∑𝑦 154
𝑥̅ = 𝑛 𝑖 = 25 = 41.44, 𝑦̅ = 𝑛 𝑖 = 25 = 6.16.
Using the computed values of the totals
∑(𝑥 − 𝑥̅ ) ∙ (𝑦 − 𝑦̅) = −517.76,
∑(𝑥 − 𝑥̅ )2 = 41467.16 and ∑(𝑦 − 𝑦̅)2 = 15.36
for the correlation coefficient, we find
−517.76
𝑟 = 41468.16×15.36 = -0.648.
√
and the coefficient of determination is
𝑟 2 = (−0.648)2 = 0.420.
As we expected the variables show a moderate negative association.
Moreover, the obtained value of the coefficient of determination 𝑟 2 = 0.420 indicates
that about 42% in the variation of the academic grade is caused by the variation in the
variation in the time devoted to the Instagram.
The regression analysis of data
Now, let us apply the regression analysis in order to examine the influence
of the time spent on Instagram (independent variable) on the academic performance
(dependent variable). To determine the degree of linearity between the two random
variables 𝑋 and 𝑌 given 𝑛 ordered pairs (𝑥1 , 𝑦1 ), (𝑥2 , 𝑦2 ), …, (𝑥𝑛 , 𝑦𝑛 ) and to find the
least squares regression line we need to determine the following quantities:
∑(𝑥−𝑥̅ )(𝑦−𝑦̅)
covariance 𝑠𝑥𝑦 = ,
𝑛
(𝑥−𝑥̅ )2
standard deviation of independent variable 𝑠𝑥 = √ .
𝑛
Once again using the values of totals from the last raw of the above table we find
−517.76 41468.16
𝑠𝑥𝑦 = = −20.71 and 𝑠𝑥 = √ = 40.72.
25 25
Now we can determine the equation of the least squares regression line
𝑠𝑥𝑦
𝑦 − 𝑦̅ = 𝑠2 (𝑥 − 𝑥̅ ).
𝑥
Substituting the obtained values for covariance and standard deviation we get
−20.71
𝑦 − 6.16 = 40.722 (𝑥 − 41.44)
or after simplification
𝑦 = −0.0125𝑥 + 6.678 .
The statistic program like Excel also allows to precisely calculate the equation
of the regression line. Excel even draws the regression line between the data points
and displays its equation on the scatter diagram. As we see the equation provided by
Excel is the same as we obtained. The gradient of the regression line -00125
indicates that every additional minute spent on Instagram decreases the
academic grade by 0.0125 points.

𝝌𝟐 test
Finally, let us determine whether the degree of involvement on Instagram
and academic performance are dependent factors. The inferential technique
appropriate to treat interval data is the chi-square test of the contingency table. The
test compares the observed set of frequencies to an expected distribution.
To test whether the time spent on Instagram and grades are independent, we
first consider only the sum values of the contingency table. Next, we start to calculate
the values we would expect to obtain if the variables were independent. We can also
complete an expected frequency table. This displays the values we would expect to
obtain if the variables were independent.
𝜒 2 test examines the difference between the observed values we obtained from
our sample and the expected values we have calculated.

Contingency table of observed frequencies, 𝑓𝑜

Grade =7 Grade < 7 sum

Time ≥30 4 11 15
 Time <30 6 5 11

sum 10 16 26

The above table is used to find the expected set of frequencies using the
relation
(𝑟𝑜𝑤 𝑡𝑜𝑡𝑎𝑙)×(𝑐𝑜𝑙𝑢𝑚𝑛 𝑡𝑜𝑡𝑎𝑙)
𝑓𝑒 = .
𝑔𝑟𝑜𝑠𝑠 𝑡𝑜𝑡𝑎𝑙
For example, the expected number of students who spend more than 30
minutes and received grade 7 is
15×10
𝑓𝑒 = 26 = 5.762307

Contingency table of expected frequencies, 𝑓𝑒

Grade =7 Grade < 7 sum
Time ≥30 5.76231 9.230769 15

Time <30 4.230769 5.76231 11

sum 10 16 25

The test statistic used to measure the similarity of the observed and expected
frequencies is
(𝑓𝑜 −𝑓𝑒 )2
𝜒 2 = ∑𝑘𝑖=1 .
𝑓𝑒

The table contains the performed computations

(𝑓𝑜 − 𝑓𝑖 )2
𝑓𝑜 𝑓𝑒 𝑓𝑜 − 𝑓𝑖 (𝑓𝑜 − 𝑓𝑖 )2
𝑓𝑒

4 5.729 -1.729 3.12936 0.542444

11 9.23 1.77 3.1329 0.339426
6 4.23 1.77 3.1329 0.740638
5 6.76 -1.76 3.0976 0.458225
𝜒2 = 3.280733

We found that the test statistics is equal to 𝜒 2 = 3.28. To judge the size of the obtained
value we need to specify the rejection region. The rejection region is defined by
inequality
𝜒 2 > 𝜒𝛼,𝜐
2
,
where 𝛼 is the significance level and 𝜐 shows degrees of freedom of chi-square
distribution. For the task of given investigation it is quite suitable to infer at 5 %
significance level, so 𝛼 = 0.05. Contingency table contains two rows and two columns
so degrees of freedom in our case is 𝜐 = (2 – 1) × (2 – 1) = 1. From the tables of 𝜒 2
2
distribution we find 𝜒0.05,1 = 2.84.
Since we decided to employ a 5% significance level, the rejection region is
defined by the inequality
2
𝜒 2 > 𝜒0.05,1 = 2.84

Conclusions
We obtained that the test statistic 𝜒 2 = 3.28 is beyond the critical value of 2.84
and therefore we reject the null hypothesis. This means that the differences between
the observed and expected frequencies are large enough to support the alternative
hypothesis. So, we conclude that the time spent on Instagram affects the grade of IB
student making student’s academic performance poorer. The obtained result in
general confirms our assumption that large amounts of times devoted to Instagram do
not contribute to improving performance.

Reference:

G Keller, 2011, Statistics for Management and Economics, 9th edition,

CENGAGE learning

Mal Coad, Glen Whiffen, Robert Haese & el. Mathematics for the
international student, Mathematical Studies SL, Haese and Harris Publications

Ben Foley, 2018, What is regression analysis and why should I use it?
[online], Available at: https://www.surveygizmo.com/resources/blog/regression-
analysis/ (accessed 2 March 2019)