benlehrburger/college-text-corpus · Datasets at Hugging Face

text stringlengths 0 2.12k
Participants
I solicited survey responses from 66 undergraduate Dartmouth students. I set up a table at the
bottom of the Berry Library staircase alongside the Novak Café for a total of four hours between
1:30 and 3:30 PM on Tuesday, November 9 and Wednesday, November 10. I approached
students to take a survey and ensured that each participant was a student that I had never met
before to mitigate selection bias. Two survey responses had to be discarded because of
incomplete information. Accordingly, the compiled results consist of data from 64 students.

Materials
I created a survey to get data on the variables in question and control for any confounding
variables. The survey included five questions. The first question inquired about a participant’s
social network structure:

How would you classify your friends?
a. Most of my friends are friends with each other
b. Most of my friends are in a few different groups
c. Most of my friends are independent from one another

Each letter response represents one of the social network structures in McCabe’s model. Option
A describes a tight-knitter, option B points to a compartmentalizer, and option C is indicative of
a sampler. Although McCabe’s model only has empirical support in an offline study, the

literature suggests that offline and online social network structures are analogous to one another
(Dunbar et al., 2015). If the literature is correct and McCabe’s model can be applied to online
social network structures, then I expect to see the number of online relationships that students
have vary across options A, B, and C.

The second question solicited the number of Instagram followers a participant has. According to
a study of social media use among teens, Instagram is the most popular social media with 72% of
teen social media users being active on the Instagram platform (Anderson et al., 2018). For this
reason, Instagram was my platform of choice for gauging digital social networks. The number of
Instagram followers that a participant has is representative of the size of their online social
network. Followers, as opposed to the number of users that the participant is following, are
willingly a part of the participant’s social network and are therefore a good indicator of the scale
of a participant’s online social network.

The third question asked for a participant’s GPA. Responses from this question were used as a
measure of academic performance. The fourth and fifth questions recorded a participant’s major
area of study and class year of graduation, respectively. These last two questions were added to
control for the influence of major and class year on experimental results.

Procedure
Experimental Design – I loaded the survey on a tablet during the data collection timeframes
noted above and approached unfamiliar students one at a time. If a student consented to being
part of the study, I handed them the tablet and asked them to fill out the survey. I noted that data was anonymous and gave them privacy while they recorded their responses. I elected to conduct
the survey digitally and self-report style to minimize the time cost of data collection and
maximize the number of willing participants. If a student did not have an Instagram account or a
GPA, I still asked them to record their response and discarded their data later on. I budgeted four
hours for data collection and stopped accepting responses afterwards.

Statistical Analysis – I compiled the data into a spreadsheet and categorized it according to
reported social network structure. No participants identified as samplers; they did not record that
their friends are independent from one another. Since the data only comprised two social
network structures, these variables were independent, they had equal variances, and the data
came from two randomly sampled normal populations, I decided to use a two-sample t-test to
determine the significance of varying Instagram followers and GPA across network structures

The responses of 64 students were collected and analyzed. 41 participants identified as tight-
knitters, 23 participants described themselves as compartmentalizers, and 0 participants

classified themselves as samplers. Statistical analyses returned evidence that GPA and the
number of Instagram followers that students have vary significantly across social network
structures.

H1 (Instagram Followers) – The 23 participants who described themselves as
compartmentalizers (M = 1460.43, SD = 564.66) compared to the 41 participants who classified
themselves as tight-knitters (M = 1249.88, SD = 364.43) demonstrated significantly larger online
social networks, t(62) = 1.81, p < 0.05. On average, compartmentalizers had more Instagram
followers than tight-knitters. Thus, we reject H1’s null hypothesis and conclude that we have
evidence that McCabe’s social network structure model is applicable to online social networks.

Students with different social network structures exhibit variance in academic
performance and the size of their online social network. Specifically, compartmentalizers have
significantly larger online social networks and higher GPAs than tight-knitters. Participants’
major area of study and graduation year were not found to have any confounding influence on
the results. Evidence from these studies supports two conclusions:

C1: McCabe’s social network model involves meaningful structural distinctions
that are realized in online social networks

C2: the structure of a student’s offline or online social network may play a role
in shaping their academic performance

These conclusions have far-reaching consequences. C1 is encouraging for
extending applications of McCabe’s model from just offline social networks to both
offline and online social networks. It also illustrates that college students tend to
translate structural features of their in-person social networks to their digital social
networks. C2, in conjunction with C1, indicates that who a student knows and how they
know them, whether it be in-person or through social media, might be a determinant of
the grades they receive. Since we have evidence of the structural parallels between
offline and online social networks, we can reason that those structures are more or less
similar and that the overarching structure could influence academic performance.
In particular, it appears that a compartmentalized social network structure is
more academically advantageous than a tight-knit social network structure. On average,
compartmentalizers have more online relationships and higher GPAs than tight-knitters.
I hypothesize that compartmentalizers have more friends with more diverse academic
interests – and therefore more academic support – than tight-knitters because they draw
friendships from different domains. Tight-knitters seek out friendships in a closed
network; if some resource does not exist in that network, say the answer to a math
problem or notes from biology class, then the tight-knitter has no way of getting the help

Participants

I solicited survey responses from 66 undergraduate Dartmouth students. I set up a table at the

bottom of the Berry Library staircase alongside the Novak Café for a total of four hours between

1:30 and 3:30 PM on Tuesday, November 9 and Wednesday, November 10. I approached

students to take a survey and ensured that each participant was a student that I had never met

before to mitigate selection bias. Two survey responses had to be discarded because of

incomplete information. Accordingly, the compiled results consist of data from 64 students.

Materials

I created a survey to get data on the variables in question and control for any confounding

variables. The survey included five questions. The first question inquired about a participant’s

social network structure:

How would you classify your friends?

a. Most of my friends are friends with each other

b. Most of my friends are in a few different groups

c. Most of my friends are independent from one another

Each letter response represents one of the social network structures in McCabe’s model. Option

A describes a tight-knitter, option B points to a compartmentalizer, and option C is indicative of

a sampler. Although McCabe’s model only has empirical support in an offline study, the

literature suggests that offline and online social network structures are analogous to one another

(Dunbar et al., 2015). If the literature is correct and McCabe’s model can be applied to online

social network structures, then I expect to see the number of online relationships that students

have vary across options A, B, and C.

The second question solicited the number of Instagram followers a participant has. According to

a study of social media use among teens, Instagram is the most popular social media with 72% of

teen social media users being active on the Instagram platform (Anderson et al., 2018). For this

reason, Instagram was my platform of choice for gauging digital social networks. The number of

Instagram followers that a participant has is representative of the size of their online social

network. Followers, as opposed to the number of users that the participant is following, are

willingly a part of the participant’s social network and are therefore a good indicator of the scale

of a participant’s online social network.

The third question asked for a participant’s GPA. Responses from this question were used as a

measure of academic performance. The fourth and fifth questions recorded a participant’s major

area of study and class year of graduation, respectively. These last two questions were added to

control for the influence of major and class year on experimental results.

Procedure

Experimental Design – I loaded the survey on a tablet during the data collection timeframes

noted above and approached unfamiliar students one at a time. If a student consented to being

part of the study, I handed them the tablet and asked them to fill out the survey. I noted that data was anonymous and gave them privacy while they recorded their responses. I elected to conduct

the survey digitally and self-report style to minimize the time cost of data collection and

maximize the number of willing participants. If a student did not have an Instagram account or a

GPA, I still asked them to record their response and discarded their data later on. I budgeted four

hours for data collection and stopped accepting responses afterwards.

Statistical Analysis – I compiled the data into a spreadsheet and categorized it according to

reported social network structure. No participants identified as samplers; they did not record that

their friends are independent from one another. Since the data only comprised two social

network structures, these variables were independent, they had equal variances, and the data

came from two randomly sampled normal populations, I decided to use a two-sample t-test to

determine the significance of varying Instagram followers and GPA across network structures

The responses of 64 students were collected and analyzed. 41 participants identified as tight-

knitters, 23 participants described themselves as compartmentalizers, and 0 participants

classified themselves as samplers. Statistical analyses returned evidence that GPA and the

number of Instagram followers that students have vary significantly across social network

structures.

H1 (Instagram Followers) – The 23 participants who described themselves as

compartmentalizers (M = 1460.43, SD = 564.66) compared to the 41 participants who classified

themselves as tight-knitters (M = 1249.88, SD = 364.43) demonstrated significantly larger online

social networks, t(62) = 1.81, p < 0.05. On average, compartmentalizers had more Instagram

followers than tight-knitters. Thus, we reject H1’s null hypothesis and conclude that we have

evidence that McCabe’s social network structure model is applicable to online social networks.

Students with different social network structures exhibit variance in academic

performance and the size of their online social network. Specifically, compartmentalizers have

significantly larger online social networks and higher GPAs than tight-knitters. Participants’

major area of study and graduation year were not found to have any confounding influence on

the results. Evidence from these studies supports two conclusions:

C1: McCabe’s social network model involves meaningful structural distinctions

that are realized in online social networks

C2: the structure of a student’s offline or online social network may play a role

in shaping their academic performance

These conclusions have far-reaching consequences. C1 is encouraging for

extending applications of McCabe’s model from just offline social networks to both

offline and online social networks. It also illustrates that college students tend to

translate structural features of their in-person social networks to their digital social

networks. C2, in conjunction with C1, indicates that who a student knows and how they

know them, whether it be in-person or through social media, might be a determinant of

the grades they receive. Since we have evidence of the structural parallels between

offline and online social networks, we can reason that those structures are more or less

similar and that the overarching structure could influence academic performance.

In particular, it appears that a compartmentalized social network structure is

more academically advantageous than a tight-knit social network structure. On average,

compartmentalizers have more online relationships and higher GPAs than tight-knitters.

I hypothesize that compartmentalizers have more friends with more diverse academic

interests – and therefore more academic support – than tight-knitters because they draw

friendships from different domains. Tight-knitters seek out friendships in a closed

network; if some resource does not exist in that network, say the answer to a math

problem or notes from biology class, then the tight-knitter has no way of getting the help