Quantitative Analysis of Short Video Dissemination Effects: Based on

the Likes, Share Rate and Comment Density

Jiashuo Wang

School of Math, University of Southampton, Southampton, SO14, U.K.

Keywords: Short Video, Logistic Regression, User Engagement Index.

Abstract: The rapid development of the short video industry makes people pay attention to the huge changes brought

by this new era of communication, from the past, a message may be a year, two years or more to let the whole

world know, to the current speed of light to spread, the short video makes the information gap has become

smaller. At the same time, short video as a form of public entertainment, the economic benefits of short video

should not be underestimated. After summarizing the directions of previous research, collected basic data on

videos from TikTok, exploring the correlation between explanatory variables, this paper uses OLS Model to

give an insight into the correlation between User Engagement Index and Short video basic data. This paper

not only confirms independence between each explanatory variable as well as the robustness of the model but

also proves the parsimony of the OLS model and the fitness between the model and the data. According to

the inferred results show that short video user participation is closely related to Comment density and Share

rate and also has a certain relationship with the number of likes. In conclusion, by studying the relationship

between user participation index and basic video data, this paper provides a more in-depth understanding of

the impact of short videos, which is conducive to further promoting the further development of the short video

industry.

1 INTRODUCTION

Under the background of accelerated global digital

process, short video culture is gradually prevalent,

YouTube, TikTok and other short video platforms

have produced different forms of short videos with

different contents, short videos have gained

popularity by virtue of their short duration, low

threshold, social attributes and precise pushing, and

have rapidly developed into a blue ocean of

development and an important research topic in the

field of Internet industry (Zhang et al., 2021). As a

text form carrying symbolic expressions, these short

videos themselves are applicable to dissemination

scenarios ranging from mobile devices to fixed

devices, and they are applicable to a wide range of

forms, making them a symbolic mode of content

dissemination in contemporary network culture.

From the point of view of technological progress,

digital video is no longer a medium juxtaposed with

other symbols, but wraps existing media forms in it,

thus presenting a videoed image of the complete

society (Zhang et al., 2022). Some scholars have

https://orcid.org/0009-0002-4367-3654

pointed out that ‘In the current mobile social

behaviour, short video, as a rapid means of

communication, can more easily achieve the needs of

mass and intuitive communication, and at the same

time can effectively reorganize multiple meaning

symbols' short video creators based on their own

experience and imagination, with the help of video

symbols, text, expanding the use of various types of

short video is not limited to the creator's imagination

and search for the meaning of life channels. The

subject's imagination and the search for the meaning

of life channels, which is also a short video can

transcend many forms of content expression, out of

all kinds of media culture is an important reason (Liu,

2024).

According to some data, the average daily usage

time of short video platforms (e.g., TikTok,

YouTube) has exceeded 2.5 hours, which has become

an important channel for global Internet users to

entertain and obtain information. According to

statistics, by 2024, the user scale of TikTok video

platform has exceeded 1.58 billion (Liu et al., 2024).

Among them, the scale of short video users in China

360

Wang, J.

Quantitative Analysis of Short Video Dissemination Effects: Based on the Likes, Share Rate and Comment Density.

DOI: 10.5220/0013825800004708

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 2nd International Conference on Innovations in Applied Mathematics, Physics, and Astronomy (IAMPA 2025), pages 360-363

ISBN: 978-989-758-774-0

is close to 1 billion, and the average daily interaction

volume is up to tens of billions of times (Zheng et al.,

2023). The rise of short video platforms has not only

revolutionized global media consumption patterns,

but also spawned interactive behavioural metrics

cantered on the rate of likes, retweets and comment

density. Likes, retweets and comments reflect users'

identification, social motivation and deep

engagement (Chen et al., 2021). Studies have shown

that liking behavior is often viewed as low-cost

immediate feedback, and its frequency of occurrence

is significantly and positively correlated with the

content's visual appeal, emotional resonance, and

entertainment (Cheng, 2024). Reposting behavior is

related to an individual's social network, while

comment density reflects the content's controversial

and topical nature. For example, a large number of

likes may stimulate retweeting behaviour through the

‘herd effect’, while a large number of comments may

in turn increase the content's exposure in the

algorithm (Zhang, 2023). Existing studies on short

video user engagement behaviour have mostly

investigated the facilitation mechanism of user

engagement behaviour in terms of the functional

features of short video platforms and the motivation

to use them (Sun et al., 2021). Therefore, this paper

constructs a user engagement analysis model with

‘likes, retweets and comment density’ as the core

variables, which provides a new perspective for

platform optimization and academic research. This

paper collects public data from short video platforms,

combines correlation analysis and multiple linear

regression analysis, and explores the following

questions: Is there a significant linear correlation

between the number of likes, retweets and comment

density. How to predict user engagement through the

synergistic effect of the three.

2 METHODOLOGY

2.1 Data Source

The data utilized in this study was sourced from

Kaggle, this data summarizes the data from one

hundred account videos, these data mainly show

some interactive behaviors of TikTok users, these

data mainly contain the number of likes, comments

and share, based on these data for deeper calculations

and summaries. TikTok is the most popular short

video platform in the market, its users are not only

most of the young people, but also a lot of older

people, so the analysis of TikTok's user behaviour

data, can get more accurate results.

2.2 Variable Description

Likes: The number of likes is the number of positive

feedback expressed by users by clicking or touching

the ‘Like’ button provided by the platform after

watching short videos. The number of likes directly

reflects the degree of users' love and recognition of

the video content and is a low-cost and direct way of

user interaction.

Share Rate: Users' ability to forward videos to

others or to other platforms is key to the proliferation

of video content and a measure of its potential for

dissemination (Figure 1).

The measurement mode is:

𝑆ℎ𝑎𝑟𝑒 = 𝑆ℎ𝑎𝑟𝑒 ÷ 𝐿𝑖𝑘𝑒𝑠 (1)

Figure 1: Share Rate and Likes (Picture credit: Original)

Comment Density: Commenting is the act of

expressing one's views, opinions or feelings through

text or other means after watching a video,

representing a deeper level of user engagement.

Measurement mode is:

𝐶𝑜𝑚𝑚𝑒𝑛𝑡 𝐷𝑒𝑛𝑠𝑖𝑡𝑦 = 𝐶𝑜𝑚𝑚𝑒𝑛𝑡 ÷ 𝐿𝑖𝑘𝑒𝑠 (2)

In this graph Comment rate is the amount of

comments divided by the value of the amount of likes,

the changes and the amount of likes are closely

related to the chart the author can see that, with the

amount of likes continue to rise, the amount of

comments are also improved, it can be clearly seen

that the comment rate is positively correlated with the

amount of likes (Figure 2).

50000

100000

150000

200000

250000

Share rate

Likes

Quantitative Analysis of Short Video Dissemination Effects: Based on the Likes, Share Rate and Comment Density

361

Figure 2: Comment Density and Likes (Picture credit:

Original)

2.3 Method Introduction

The method used in this paper is the Ordinary Least

Squares (OLS). It serves as a cornerstone in

understanding the relationships between variables,

OLS is a classical method of linear regression

analysis which aims to find the optimal linear

relationship between the independent variables and

the dependent variable by minimizing the sum of

squared residuals between the predicted and actual

values (Zhang et al., 2019). Through its application,

this paper will reveal unique insights into the User

Engagement Index. The linear fitting equation in OLS

Model can be written as:

𝑦=𝛽



+𝛽



𝑥



+𝛽



𝑥



+⋯+𝛽



𝑥



+𝜀 (3)

Where 𝑦 is User Engagement Index, 𝑥



is Likes, 𝑥



Share Rate, 𝑥

is Comment Rate.

3 RESULTS AND DISCUSSION

3.1 Collinearity Diagnosis

The test for multicollinearity of the model shows that

all the Variance Inflation Factor (VIF) values in the

model are less than 2 (with a maximum value of

1.280), which implies that there is no problem of

covariance; the D-W values are all around 2, which

suggests that there is no autocorrelation in the model,

and there is no correlation between the sample data

(Table 1). The tolerance values are all greater than 0.7,

which further supports that there is no strong

correlation between the variables. Therefore, the

explanatory variables in the model have good

independence and are suitable for regression analysis.

Table 1: Collinearity Diagnosis

Ite

VIF Tolerance

Likes 1.052 0.951

Share Rate 1.280 0.781

Comment Rate 1.243 0.804

3.2 OLS Regression Model

After the above preparation work, the paper started

to establish the OLS regression model.

Table 2: OLS Regression results

Unstandardised

coefficient

Standardised.

coefficient

Std. Error t p Value 95% CI

Constant -0.000 - 0.000 -0.68 0.498 -0.000 ~ 0.000

Likes 0.400 1.000 0.000 31135450 0.000** 0.400 ~ 0.400

Share Rate 0.300 0.000 0.000 1659897 0.000** 0.300 ~ 0.300

Comment Rate 0.300 0.000 0.000 53750 0.000** 0.300 ~ 0.300

1.000

Adjust R

1.000

F Test

F (3, 95) = 3.398, p = 0.000

D-W

1.995

Note: Dependent Variable = User Engagement

< 0.05 **

< 0.01

IAMPA 2025 - The International Conference on Innovations in Applied Mathematics, Physics, and Astronomy

362

Analysis of the results of the OLS regression model

found that. Likes, Share Rate and Comment Rate all

have a significant positive effect on user engagement,

but the model R²=1.000 suggests that there may be a

risk of overfitting; the F-test and the Durbin-Watson

test verify the overall significance of the model and

the residuals' no-autocorrelation property,

respectively (Table 2). The unstandardised regression

coefficient of the number of likes, β=0.400, indicates

that when the number of likes increases by 1 unit, user

engagement increases by 0.400 units on average. The

coefficients of β = 0.300 for both share rate and

comment rate indicate that user engagement increases

by 0.300 units for each 1-unit increase in both. The p-

values for all three are less than 0.01, indicating that

the coefficients are highly statistically significant.

The standardised coefficients show that the number

of likes has the greatest impact, while the

standardised values of sharing rate and comment rate

are zero, which may be due to multicollinearity

between the variables causing distortion in the

standardised results. This model verifies the positive

contribution of the number of likes, shares and

comments to user engagement, with the number of

likes having the greatest impact; however, the high R²

and the zero standardised coefficient suggest the risk

of overfitting and multicollinearity, which can be

improved in subsequent studies by increasing the

sample size, eliminating covariates, or adopting

regularisation methods.

4 CONCLUSION

Overall, based on the user interaction data of TikTok

provided by Kaggle, three core indicators, namely,

the number of likes, the retweet rate and the comment

density, were selected to analyse their effects on user

engagement through the least squares regression

(OLS) method. The results show that all three

variables are significantly positively correlated with

user engagement, and the model R² is as high as

1.000, indicating that the variables can fully explain

the changes in user engagement without

multicollinearity or autocorrelation problems, and

have good statistical stability. Among them, the

regression coefficient of the liking behaviour is 0.400,

which has the most significant impact, and the

forwarding rate and comment density are both 0.300,

indicating that all three play a key role in enhancing

the content dissemination effect. The model not only

provides quantitative reference for the optimisation of

platform recommendation mechanism but also

provides a theoretical basis for the formulation of

short video content operation strategy. Meanwhile,

although this study combines multi-dimensional data,

the data is still not rich enough. In future research,

more dimensional variables, such as video content

type, release time, audience profile, etc., can be

introduced to further enrich the structure of the

model; at the same time, non-linear models such as

machine learning can be combined to improve the

prediction accuracy and adaptability to the reality, in

order to more comprehensively reveal the

behavioural mechanisms behind the dissemination of

short videos.

REFERENCES

Chen, L., Wang, X., Wang, P., Ma, X. 2021. Cancer

communication and user engagement on Chinese social

media: Extracting topics using text analytics. Journal of

Medical Internet Research.

Cheng, Z. 2024. Like, comment, and share on TikTok:

Exploring the effect of sentiment and second-person

view on the user engagement with TikTok news videos.

Master’s thesis, University of Arizona.

Liu, C., Jiang, M., Muhammad, Z. A. 2024. The impact of

tiktok short video factors on tourists' behavioral

intention among generation z and millennials: the role

of flow experience. PLoS ONE, 19(12).

Liu, H. 2024. Analysis of the influence of TikTok short

video platform on user. Bachelor’s thesis, Inner

Mongolia Normal University. News and

Communication.

Sun, G. M., Wang, S., Zhang, Y., Liu, Y., Gao, J., Xie, Q.

2023. Behavior analysis of user interaction on online

short video platform. J. Comput. Inf. Technol., 30, 51-

65.

Zhang, M., Long, B., Shao, X., Liu, Y., Zhang, Y. 2022.

Formation mechanism of short video app users’

willingness to continuously use and its governance

outlook: Based on the analytical perspective of mimetic

companionship. Modern Intelligence, 41(07), 49-59.

Zhang, S., Wang, W., Xie, Y. 2021. Evolutionary law and

power mechanism of continuous participation behavior

of online health community users. Modern Intelligence,

41(05), 59-66.

Zhang, X., Wu, J., Liu, S. 2019. Analysis of influencing

factors of browsing and creating behavior of mobile

short video users. Library and Intelligence Work,

63(06), 103–115.

Zhang, Y. 2023. Study on the influencing factors of public

participation in political short video: Taking political

Jitterbug number as an example. Master’s thesis, Hebei

University of Technology.

Zheng, Z., Lu, C., Zhao, W. 2023. Study on the influence

of douyin short video marketing on camping travel

intention. SHS Web of Conferences, 179(000), 5.

Quantitative Analysis of Short Video Dissemination Effects: Based on the Likes, Share Rate and Comment Density

363