Hlavní práce68933_krir02.pdf, 1.9 MB Stáhnout

(1)

University of Economics, Prague Faculty of Economics

Field of Study: Economics

Blocked Social Media and Academic Performance

BACHELOR THESIS

Autor: Robert Kříž

Thesis supervisor: PhDr. Lubomír Cingl Ph.D.

Year: 2019

(2)

University of Economics, Prague Department of Economics

Faculty of Economics Academic year: 2018/2019

BACHELOR THESIS TOPIC

Author of thesis:

Robert Kříž

Study programme: Economics and Economic Administration

Field of study: Economics

Topic:

Blocked Social Media and Academic Performance

Guides to writing a thesis:

1. The thesis is concerned with the effect of using online social media use on learning outcomes. The objective of the thesis is to analyze whether voluntary imposition of a commitment device (in this case website blockers that limit the time a user can access selected web-pages per day) leads to a higher learning efficiency of the students of VŠE.

The effect can be both positive as the social media may play a role as a source of distraction, but also negative since students these days often share the study material on social media.

2. Social media use is a favorite free-time activity of an increasing number of online users, especially the youth: 97 % of Czech youth aged between 16 – 24 are active on social media [1]. While social media use relates to many positive effects such as easier sharing and collecting information, better communication and increased creativity, it is also connected to many negative effects such as the loss of ability to concentrate on one task [2]. The interface of most social media sites is designed to be distracting and addictive. Limited will-power of its users may cause that even though they would prefer to engage in studying they would get often distracted by social media and would fail to accomplish what they would set to do.

To fight the distracting effects of social media, several programs have been developed that limit or completely block the access to websites chosen by the user. The programs seem to be frequently in use, but to date there has not been a comprehensive evaluation of their efficacy in the scientific literature. A proper analysis of the costs and benefits is needed in this area so students and the general population can make better-informed decisions.

The research question is based on previous findings that social-media use leads to

multi-tasking which then leads to loss of cognitive control [3]. Cognitive control is crucial for focusing attention on goal-relevant information [4] Youth who frequently use several media simultaneously may become accustomed to processing information from several sources simultaneously [5].

3. The data for this thesis will be collected through an experiment. Students in several large classes will be asked to fill out a questionnaire and a randomly chosen half of them will be asked to install a program that will limit the social media and other websites they select. We hope for a minimum number of students to be 30 in the treatment and 30 control group. The ultimate outcome will be the students’ performance in midterm examination. By conducting an experiment the thesis will be able to analyze the immediate effects of the voluntary imposition of the commitment device. Observational data will not suffice because there is a strong self-selection of people who use it which then creates endogeneity bias. Both groups of students will be also asked to self-report (1) the time they spent studying, (2) how well they met their self-set goals, (3) if they felt better focused while studying and (4) whether they were seeking for any substitutes for social media. This data will be at the end contrasted with actual grades scored in the midterm tests to test whether moderation of social media use leads to higher learning efficiency and motivation.

(3)

Length of thesis: 45 Selected bibliography:

1. DellaVigna, S., & Malmendier, U. (2006). Paying Not to Go to the Gym. American Economic Review, 96(3), 674–819.

1 Czech Statistical Office https://www.czso.cz/csu/czso/vyuzivani-informacnich-a-komunikacnich- technologii-v-domacnostech-a-mezi-jednotlivci

2 Lau, W. W. F. (2017). Computers in Human Behavior Effects of social media usage and social media multitasking on the academic performance of university students. Computers in Human Behavior, 68, 286–291. https://doi.org/10.1016/j.chb.2016.11.043

3 Van Der Schuur, W. A., Baumgartner, S. E., Sumter, S. R., & Valkenburg, P. M. (2015). The consequences of media multitasking for youth: A review. Computers in Human Behavior, 53, 204–215.

https://doi.org/10.1016/j.chb.2015.06.035

4 Miller, E. K., & Cohen, J. D. (2001). An integrative theory of prefrontal cortex function. Annual Review of Neuroscience, 24(1), 167–202. http://dx.doi.org/10.1146/annurev.neuro.24.1.167

5 Yang, X. J., Zhang, J., Zhao, W. H., Lv, D., Ma, C. F., & Li, Y. (2016). A Novel Evaluation Method on the Precision of Linear Motor Feed System in High-Speed Machine Tools. Materials Science Forum, 836–837(37), 220–227. https://doiorg/10.4028/www.scientific.net/msf.836-837.220

Bachelor thesis topic submission date: February 2019 Deadline for submission of Bachelor thesis:May 2019

Robert Kříž Solver

prof. Ing. Robert Holman, CSc.

Head of department

PhDr. Lubomír Cingl, Ph.D.

Thesis supervisor

prof. Ing. Zdeněk Chytil, CSc.

Dean NF VŠE

(4)

I declare that this thesis was composed by myself, that the work contained herein is my own except where explicitly stated otherwise in the text, and that this work has not been submitted for any other degree or professional qualification.

In Prague . . . . Author

(5)

Abstract

In the case of inconsistent preferences, people disobey their plans and regret it later. To prevent self-control failures, they can use commitment devices. One type of commitment device is website blocker, and it is designed to help people stay focused and minimize procrastination on distractive online sites. Because social media can be potentially distractive and deteriorate the academic performance of students, field experiment at the University of Economics in Prague is conducted to estimate the impact of using website blockers on test scores. Students of four courses (ntreatnent = 46, ncontrol = 47), were encouraged to download website blockers five days prior to midterm tests. Possibly due to a small take-up rate in the treatment group (15.21 %), no statistically significant evidence of commitment device improving grades is found. However, when conducting G*power post-hoc sample size analysis, it is estimated that with a total sample size of n = 272 (actual sample size is 93), the impact would be significant (ceteris paribus). 81,8 % commitment device users self-reported that it is useful. In the pre-survey study, subjects also admitted being distracted by social-media more often than not while a third reported being distracted very often. Conclusion of the practical part is verification of the theory of inconsistent preference, which is the key concept of the theoretical part.

Keywords: Self-control, Voluntary imposition of commitment device, website blocker, field experiment, academic performance

JEL classification: D010, D900

(6)

Abstrakt

V případě nekonzistentních preferencí, lidé nenaplňují své původní plány, a pak toho litují. Aby zabránili problémům se sebekontrolou, mohou si pomoci závazkem (tzv. commitment device). Příkladem takových závazků jsou například aplikace - website blockery, které jsou navržené, aby pomáhaly lidem se soustředěním a minimalizovaly prokrastinaci v online prostředí. Zejména sociální sítě jsou potencionálně rozptylující, a mohou tak neg- ativně ovlivnit akademický prospěch studentů. Tato práce se zabývá měřením výsledků průběžných testů na Vysoké škole ekonomické a následným analyzováním rozdílů mezi studenty, kteří byly v rámci field experimentu pět dní před průběžnými testy vybídnuti k nainstalování těchto aplikací (n_treatnent = 46), a kontrolní skupinou (n_control = 47).

Možným vlivem nízké míry využití pomůcek, nebyl u zkoumané skupiny nalezen signifik- antní efekt. Nicméně, post-hoc analýza v programu G*Power odhalila, že při navýšení celkového počtu pozorování na n = 272 (ze skutečných n = 93) by byl rozdíl mezi skóre zkoumané a kontrolní skupiny signifikantní. Použití pomůcek považuje 81,8% uživatelů za prospěšné. Dále, studenti v rámci dotazníkového šetření v průměru uvedli, že je sociální sítě ruší více než ”občas” a třetina uvedla, že je ruší minimálně ”velmi často”. Zároveň 48% subjektů přiznává velmi častou prokrastinaci. Jedním ze závěrů praktické části práce je tedy potvrzení ekonomické teorie nekonzistentních preferencí, která je klíčovým pozn- atkem části teoretické.

Klíčová slova: sebe-kontrola, commitment device, website blocker, field experiment, aka- demický prospěch

JEL klasifikace: D010, D900

(7)

Rád bych na tomto místě poděkoval PhDr. Lubomírovi Cinglovi Ph.D. za odborné vedení práce, velmi cenné rady, náměty a vstřícnost během celé doby spolupráce. Především si vážím jeho trpělivosti odpovídat na mé nesčetné dotazy. Také děkuji Ing. Tomášovi Miklánkovi, M.A., Ph.D. za poskytnuté rady k řešení analýzy výsledků. Rovněž děkuji své rodině a přátelům za neocenitelnou podporu během studia.

(8)

1 Introduction

Social media (SM) is a worldwide phenomenon. Of all Internet users, 98% have used a SM network in the past month with an average of 2 hours 22 minutes per day spent on social networks (GlobalWebIndex, 2018). The largest network Facebook, still grows with almost 2,5 billion users as of 2019 (GlobalWebIndex, 2018). The success of social media may be attributed to its potency to help people to maintain contact with the large online network and to the unlimited access to user-generated content. However, another possible reason for extensive use is its addictive qualities. Moreover, social media can be the go-to place for procrastination.

In the experiment of Rosen (Rosen, Carrier, and Cheever, 2013), it took students 6 minutes on average to check distracting websites after starting study sessions. Also, those students that accessed Facebook when studying had lower GPAs (Grade Point Average). In another research (Lavoie and Pychyl, 2001), 50.7 % of the subjects self- reported frequent procrastination while 47% of all procrastination happened online. In the US, the average worker gets interrupted every 10.5 minutes by external notification coming from social-media (Marotta and Acquisti, 2017).

Studies on social media are relevant to economists and policymakers in a similar way as studies on reducing overeating, undersaving, procrastination, and other self-defeating behaviours that feel good now but generate larger delayed costs. When a student, for example, makes a decision to study tomorrow and when tomorrow comes procrastinates on Facebook instead, he or she experiences a preference reversal, which is a key concept in behavioural economics. Just as with overeating or undersaving, this procrastinating behavior becomes a problem when individuals acknowledge that they would rather eat less, save more and be more productive. Because human decision-making across time is a domain of economics, many theories of intertemporal choice have been proposed to explain preference-reversals. The economic theory, therefore, offers suitable tools to explain social-media procrastination. In a sense, this topic is believed to be important to economics because of potential societal costs, but economics, in turn, offers theory, which is important for the topic itself.

Generally, people often fail to stick to their plans. They expect exercising regularly and so buy a season gym pass. They also want to start saving money for retirement with

(11)

1 Introduction

monthly savings deposits. Alternatively, they want to be more productive when studying or at work. However, a few weeks after their New Year’s resolution, their gym attendance drops, their savings diminish, and they are caught again in the vast online content. Neo classical economics assumes that consumer preferences are time-consistent. Therefore, dynamically incosistent behavior such as procrastination cannot be explained with neoclassical paradigm. Economists have therefore formalized the self-control phenomenon by proposing that most consumers have time-inconsistent preferences and discount future value hyperbolically(Loewenstein and Prelec, 1992).

There is one practical solution that helps people overcome their self-control dilemmas:

commitment device. To ensure that they will follow their intended course of action, they can voluntarily self-impose some form of commitment such as: betting a colleague from work that they will lose 10 pounds over the course of next three months, depositing money to Christmas club savings account or downloading website blocker.

The existence of commitment devices suggests that people acknowledge their self-control problems. One of the commmitment devices is a program that prevents visiting webpages where people procrastinate, so-called website blocker. Today, there is a dozen of website blocker apps such as Freedom or Stay Focused that help people stay productive.

The primary objective of this thesis is to analyze whether website blockers help students achieve better grades, put in more effort, feel more focused and spend less time on SM.

These hypotheses are tested by conducting a field experiment at the University of Eco- nomics in Prague where students of four courses were encouraged to download website blockers five prior to their midterm exams. The data of the academic performance and other outcome variables were collected in a post-survey study, which took place during midterm test-taking sessions.

The theoretical part (Chapter 2) of this thesis is structured into three main sections such that firstly, the economic theory behind self-control is provided. Secondly, that the practical solution in the form of a commitment device is presented. And thirdly, the contemporary research on the relationship between SM and academic performance is summarized.

The first section includes analysis of contrasting paradigms of neo-classical economics and contemporary behavioural economics by presenting an overview of the evolution of inter-temporal choice models in economics. Key concepts such as preference reversals (SS LL rewards), hyperbolic-discounting nad naif-sofisticate customer division are employed.

The second section initially analyzes commitment devices in general, which is followed by describing one in particular - website blocker. The last section defines SM and later on employs theoretical knowledge from the first section to show why it is subject to self-control issues and therefore, why this particular research is relevant to economics.

(12)

The practical part is divided into two parts, where the first one explains the methodology, and the second one presents the results. The methodology (Chapter3) first focuses on the subject of the research and states the main hypothesis regarding website blocker and academic performance. Also, three other hypotheses are stated. Then, key tools and concepts for collecting and analyzing data are mentioned, followed by a chronological description fo the experimental procedure. The present author provided four models, and their results are in Chapter 4.

(13)

2 Theoretical Background

2.1 Self-Control

Self-control refers to the voluntary regulation of conflicting thoughts, feelings, and actions in accordance with long-term goals. In a self-control dilemma, the individual wants to do something that is immediately rewarding and, in addition, wants to do something else that has more enduring personal value (Duckworth, Milkman, and David Laibson, 2018).

In a sense, a person with perfect self-control studies for a midterm test in advance, does not watch funny videos amidst of doing homework and meets deadlines with ease.

The phenomenon of self-control is a determinant of behaviour in many areas interesting to economists and policymakers. Self-control failures contribute to issues with retirement savings (Richard H. Thaler, 2007), obesity (VanEpps, Julie S. Downs, and Loewenstein, 2016) and overall well-being of people (Wiese et al., 2017). People with greater self-control are healthier and also wealthier (Moffitt et al., 2011). Not surprisingly, the percentage of self-control scholarly articles has multiplied five times over the past two decades (Duck- worth, Milkman, and David Laibson, 2018). What is puzzling about self-control is that people experience self-control failure even though they recognize it as a personal issue (Norcross and Vangarelli, 1988). People are aware of this and feel like they can handle it, but eventually, do not (Augenblick and Rabin, 2018). Many self-control lapses occur despite the full knowledge of negative consequences (Julie S Downs, Loewenstein, and Wisdom, 2009).

Self-control is often discussed to be one of the important variables that determine the outcome of social media use of students (Lee-Won, Herzog, and Park, 2015). In their case, self-control helps them to overcome the temptation to check what their friends are up to and to allocate more time to homework, which yields more value in the long-run.

Yet, it is often the case with students that they postpone their intended course of action even though they regret it afterwards. Such a phenomenon is attributed to self-control failure. One possible approach to this is to think of a student or any person in general as an agent with multiple selves, so the conflict of interests within some time horizon can happen. At any point in time, when students make decisions whether to study or to open Facebook, every such decision can be represented as a separate self that may or may not

(14)

2.1 Self-Control be in conflict with other selves in terms of preferences. When a student decides to do task A tomorrow, but when tomorrow comes does B instead, it is called a preference reversal.

To explain preference reversals, which challenge the neoclassical models, it is needed to have a look at economic theories of intertemporal choice.

2.1.1 Time Discounting Models

The substantial evidence of self-control issues, in part mentioned in the previous section, has been a threat to the neoclassical paradigm in recent decades. The primary problem lays in the theoretical framework of intertemporal choice, in which NE assumes that agents make consistent choices across time.

In classical economics, the intertemporal choice was first summarized by Samuelson’s Discounted Utility Model (Samuelson, 1937). Samuelson put together ideas of other economists and created a simple but elegant model. In it, present consumption is valued more than future consumption, which is discounted by exponential factorδ. It is assumed that an agent in timet, agrees with his future “self” in time t+ 1 upon what is the utility of consumption at any given time t+n. Thus, with no more information being available to the agent, his preferences are going to be consistent throughout all time periods. In such a case, agents are dynamically consistent.

P resentV alue=V alue₀ ∗δ^Delay (2.1)

Discounting future value by an exponential factor is the only type of discounting that makes choices of agents in time consistent. The first economist to point this out was Strotz. Although he did not come up with alternative function himself, he suggested that in reality, the discount functions of people are non-exponential (Strotz, 1955). What particularly caught his attention was the demand of people for commitment. By explaining commitment as additional costs that individuals voluntarily undertake just so they follow their intended course of action, Strotz concluded that such additional costs imply that people are aware of their inconsistencies and are willing to pay to prevent it. George Ainslie (Ainslie, 1975) first proposed that actual discount function might be hyperbolic.

His proposed function was based on experiments by psychologist Richard Herrnstein where he states that subjects tend to sample two concurrent streams of reward in proportion to the mean rates, amounts, and immediacies of those rewards (Herrnstein, 1961). He named the two streams as Smaller Sooner rewards (SS) and Larger Later rewards (LL).

In Figure 2.1, two hyperbolic discounting curves can be seen. One of them ends at the point of SS reward, indicating that reward comes sooner in time but is smaller in value.

(15)

2 Theoretical Background

Figure 2.1: SS and LL rewards (Ainslie, 2012)

The other discounting curve represents action with greater reward but happens later. If an agent made a decision before the curves cross at time t, then he would choose the larger later reward. If such a decision was on the right from time t, that is, close to the small sooner reward, then the agent would choose the smaller sooner reward. The essence of hyperbolic discount function as pinpointed by Ainslie is that they get steeper as the reward is close but also contain bigger “tail“ than in the case of exponential function which, with their constant discounting rate, would never meet and thus the conflicting decision making would not have arisen. The formula below is a form of standard hyperbolical function with dynamically inconsistent preferences (Ainslie, 2012).

P resentV alue= V alue₀

1 + (k∗Delay) (2.2)

The present value here is equal to immediate value divided by one plus the multiplication of degree of impatience k and delay.

Many past and contemporary experiments back-up the non exponentiality of human and nonhuman discounting. In one Richard Thaler’s experiment (Thaler, 1981), he asked his subjects to specify what amount of money would make them indifferent in one month, one year and ten years compared to $15 now. The median responses were $20/$50/$100.

Using these responses, Thaler modelled average discount functions for each time period with an annual discount rate of 345% for a one-month horizon, 120% for a one-year horizon and only 19% for a ten-year horizon. Ainslie (Ainslie, 1992) then supported economists with evidence from human and nonhuman experiments concluding that the human and animal discount functions are approximately hyperbolic.

However, because the shape of hyperbolic functions are not compatible with economic computations assuming consistent preferences (Koopmans, 1960), David Liabson came

(16)

2.1 Self-Control

up with quasi-hyperbolic function to synthesise both approaches (D. Laibson, 1997).

P resentV alue = V alue₀∗β∗δ^Delay (2.3) V alue0 is the immediate value, β is the depreciation rate equal either to one or to the interval between zero and one. The symbol δ is a depreciation rate is between zero and one.

Figure 2.2: Quasi hyperbolic function(Ainslie, 2012)

As we can see in the Figure 2.2, now the decision maker is consistent until theβspike which happens shortly before SS reward. The sense in which this function is quasi-hyperbolic is given by its two parts, where the first one (β < 0) is discounted exponentially, and the second one is discounted hyperbolically (β = 1). Richard Thaler made a distinction between these two parts, calling the first self a "planner" and the second self a "doer"

(Thaler, 1981). The planner is a typical case of the economic man in the classical economic theory, which makes plans rationally, but relies on the doer to execute these plans. The planner will ask the doer to allocate a good amount of work for studying, but as the SS reward gets just close enough to arouse the doer’s appetite for watching YouTube, the doer will disobey to execute the original plan. McClure argued that the shift between the doer and the planner happens right around the time the doer’s appetite gets aroused (McClure et al., 2007).

Self-control is a skill, a kind of human capital, that helps people to overcome their impul- sive behaviour. An example of a self–control success would be a person that acknowledges the small utility of SS reward and stays on the path of LL reward.

(17)

2.1.2 Naive and Sophisticated customers

Another area of focus in the field of self-control is the classification of consumers into groups based on their ability to recognize their true levels of self-control. Such classification helps us to understand the importance of experience and expectations of consumers that are building stones of their consumer behaviour. According to O’Donoghue and Ra- bin (Donoghue and Rabin, 1999) we have three major groups. Firstly, the time-consistent consumers are those that behave consistently across time. Such consumers discount time exponentially and thus behave according to the neoclassical paradigm. The other two groups, naïfs and sophisticates, favour SS rewards more than LL rewards.

The groups vary in their ability to accurately self-report their tendencies. Sophisticates are aware of their self-control failures and can better account for future deviations from the planned consumption path. On the other hand, naifs tend to be overconfident and expect no deviations from the desired consumption path. The actual behaviour of naifs nevertheless, is more like that of sophisticates with one major difference: sophisticates use commitment devices such as website-blockers as a way of coping with future instability of preferences. Thus, sophisticates may achieve better long-term results provided that all other aspects of human capital are the same.

Wong (Wong, 2008) demonstrated that students’ levels of sophistication and naïveté, regarding exam studying, predicted their academic class performance. But when naifs are asked to make precommitments just like it is done in the practical part of this thesis, they may actually achieve better results because unaware of their self-control problems they will pre-commit in such ways that are similar to their ideal plans (Mandel et al., 2017). Paradoxically, false assumptions about one’s self-control can make commitment devices more useful (at least in the short run).

2.2 Commitment Devices

Commitment devices are tools that force individuals to behave as initially planned. They are imposed voluntarily in period 1 and then, in turn, enforce the desired behaviour in subsequent periods. Commitment devices vary in enforcement strength. In (Bryan, Karlan, and Nelson, 2010), the authors distinguish soft and hard commitment devices.

Soft commitment devices may enforce behaviour only hypothetically; that is, there is an only small economic cost associated with self-control failure.

An example of a soft commitment device is separate savings account labelled “holiday with kids”. The account owners decide to save some money for a holiday and simply creates a separate saving account resembling “money jar”, commitment device of the past when

(18)

2.2 Commitment Devices frequently used for depositing money in a bank was not that common. If he or she fails to deposit money on to the savings account, there is no real economic cost associated with failing other than the time spent creating the account which turned out to be a useless decision in this case. On the other hand, hard commitment impose real costs. In such a case, if the monthly deposit was not made, the account owner would receive no interest.

Hard commitment devices (for obvious reasons) enforce the desired behaviour with bigger strength. Yet, soft commitment may also deliver desired results simply because they can make humans more reflective and thus restrain them from habitual behaviour.

A classic example of a commitment device is Christmas club which is nothing more than a financial account to which payments are deposited regularly throughout the year and are withdrawn from it during Christmas. Owners of such accounts are aware of their ten- dency to spend more money on SS rewards throughout the year than they would desire.

When Christmas comes, they have not sufficient funds to please their families. In period 1 (Christmas) they realize they prefer pleasing their families. Also, they observe their failures in the past to stick to given financial habits. Therefore, at the beginning of the next year, they pre-commit to automatic monthly transfers to the Christmas club accounts.

Such behaviour is in accordance with sophisticated customer described in Section 2.1.2.

What is a crucial aspect of a commitment device is that it imposes additional costs on those who implement it on themselves. For instance, some people might use money jars with labels such as “food”, “electricity” or “gas” to pre-commit to certain consumer behaviour even though they are foregoing interest which they would otherwise receive if deposited on a financial account. If people behave as sophisticates they will compare their planned behaviour with expected behaviour. If the difference in terms of utility between LL and SS reward is greater than the cost of a commitment, then the sophisticates will use a commitment device. In other words, we observe behaviour that is not in accordance with the neoclassical paradigm of rational choice theory. Time consistent individuals know their consumption path and take appropriate actions. It is only in the case of hyperbolic discounting as showed by Strotz, that imposing additional costs is rational.

2.2.1 Website Blockers

Similarly in a more recent example, the existence of commitment devices built within internet browsers called “website blockers”, shows that some people feel that they use social media and online entertainment too much and voluntarily restrict their own access to problematic sites. As downloadable add-ons in popular browsers, blockers like “Freedom”

“Stay Focused” let people pre-commit to a fixed amount of time daily to be used on chosen websites. Users can either go into “work mode” for some time they want to be focused and have access only to useful content online without any distraction or set a maximum

(19)

limit to distracting sites daily. Turning off such commitment devices often requires some effort so that users are discouraged from checking Facebook, for example. Users have to perform some action before they can turn off the website blockers, which is another form of self-imposed costs. In some cases, these website blockers are paid. In other words, those users voluntarily give up part of their income to decrease unwanted consumption of social media and entertainment. There are various types of website blocker, and they work both on smartphones and computers.

As mentioned in the previous paragraph, website blockers are sometimes (e.g. Stay Fo- cused) designed to make it difficult for the user to turn it off. Nevertheless, they still fall into the category of soft commitment devices which were discussed in Section 2.2.

Even though on the one hand, spending too much time on social media for fun may cause students, for instance, to reach the time limit and thus restrain their access to student study groups, on the other, they still can disable website blocker in their browsers and thus suffer minimal economic costs. Most popular operating systems (OS) do not allow website blockers to gain control over electronic devices. That way, there is no option for a hard commitment which some people might find useful.

Previous research on website blockers, academic performance and productivity A recent study on website blockers and academic performance by Patterson shows a positive relationship between these two (Patterson, 2018). The author was conducting an experiment where he was looking at completion rates of a massive international online course and found out that users who used website blockers had a significantly higher probability of course completion. Relative to students in control group, a student in the treatment group spent 24% more time working on the course, received course grades that were 0.29 standard deviations higher, and were 40% more likely to complete the course.

The evidence of the positive impact of previously mentioned website blocker "Freedom"

was found in the field experiment of (Marotta and Acquisti, 2017). Here the authors tested the effect of Freedom on the productivity of office workers. Interestingly, they found no significant impact of "Freedom" on the "endogenous" treatment group where the subject could independently set the limit for distracting websites. However, the exogenous group that had pre-determined blocked access to potentially distracting sites performed significantly better.

2.3 Social Media and Academic Performance

The present author first gives a brief overview of Social Media (SM). Then recognizes three major features of SM that may cause self-control problems in students while using

(20)

2.3 Social Media and Academic Performance

SS-LL analysis.

Social Media Overview

Social media is a group of Internet-based applications which allow the exchange of user- generated content (Kaplan and Haenlein, 2010) which is such content that is (1) accessible to public or a closed network, (2) original and not merely replication of other content and lastly (3) is created outside of professional practices (OECD, 2007). In a more economical approach, one could describe social media as a market with information. Every user has its own network of users to whom the information they share is of value and vice versa.

Social media became an online market for User-generated Content with very low entry price. Social media is a global phenomenon that is being adopted more increasingly by all age groups. As of 2018 report (GlobalWebIndex, 2018), 98% of all Internet users have used a social media network in the past month with an average of 2 hours 22 minutes per day spent on social networks.

While the success of social media can be partially attributed to its immense usefulness, it also should be taken into consideration that the competitive market of social-media networks pushes their development towards attracting more and more attention at what- ever cost. In the US, the average worker gets interrupted every 10.5 minutes by external notification coming from social-media (Marotta and Acquisti, 2017). Similarly, students who often rely on the easy accessibility of academic resources on Facebook and Youtube may get often distracted during class or home studies.

From one point of view, social media is of great value to students. They can access third- party study material as well as the recommended study resources from the teachers and do not have to make costly visits to the library. Moreover, they can share tips, experience and student-made material between each other in their study groups (Lau, 2017). The downside of social media is that they are not designed for academic purposes such as Facebook or Youtube is that once students log on the websites, they are instantly exposed to unrelated content that may be distractive and is intended to catch their attention. In some cases, social media can be a go-to place for procrastination, and while it may be only a facet of the procrastinating behaviour, it can also cause it. In Rosen, Carrier and Cheeve, students reported that Facebook makes them lose track of time and that they postpone their intended course of action, e.g. preparing for exams (Rosen, Carrier, and Cheever, 2013). In the research of Lavoie, J. A. A., and Pychyl, T. A., 50.7 % of the subjects self-reported frequent procrastination while 47% of all procrastination happened online (Lavoie and Pychyl, 2001).

The relationship between grades and social media has been a frequent subject of research lately. Jacobsen and Forste obtained a valuable 3-day log data of first-year college students

(21)

and found a negative correlation between time spent on social media and GPA (Jacobsen and Forste, 2011). For every additional hour spent on social media in one day, the grades were worse by 0.06. Similarly, a study done at Ohio State University shows that students with fewer friends on Facebook have better grades (Kalpidou, Costin, and Morris, 2011).

Curiously, the negative relationship was found only with first-year students suggesting that more experienced students have better self-control characteristics. Lee-Won, Herzog,

& Park suggest that low self-control may be a central driver of unintended Facebook use (Lee-Won, Herzog, and Park, 2015).

Social-Media as a SS reward

In Section 2.1.1 we could see the SS LL models of intertemporal-choice explaining preference reversals. In this part, I would like to demostrate 3 features of social media that make it into SS reward that interefers with student’s LL rewards (school).

The first and most important feature in this analysis is the user-generated content that in its essence may be a source of pleasure for students. There is no doubt that the success of SM sites such as Facebook or Instagram is primarly built on its specific content that is being indivually algorhytmically distributed to user’s feed and notification centers. What makes SM unique is that the content that is there very often strongly relates to user themselves. The content firstly relates to their sense of self (likes, comments on their own content), secondly to their sense of being a member of community (friend posts, group posts) and thirdly to their interests outside their community. According to neuro-science, social interactions - laughing faces, positive recognition by our peers, messages from loved ones - activate dopaminergic reward pathways (Krach, 2010) Smartphones have provided us with a virtually unlimited supply of social stimuli, both positive and negative. Every notification, whether it’s a text message, a “like” on Instagram, or a Facebook notification, has the potential to be a positive social stimulus and dopamine influx.

Drawn from the content of social-media being considered as a reward itself, the second feature we are about to discuss is the imminence of SM. With electronic devices such as computers, tablets and smartphones, SM is always literally in students’ pockets. More- over, every major SM is free and to create an account only takes minutes. When regarded as a source of pleasure it is only one step from here to regard as possible SS reward.

Users, when being notified about some new content related to them, only have to take their phones from their pockets and unlock their screens. To many people this has become a habit as 73% of people claim to experience anxiety after not having their phone with them (Aarestad and Eide, 2017). Because the reward here is so easily accessible it bears resemblance to the SS reward and within the context of students it might as well be, that the “doer” frequently checks SM even though the “planner” initially intended to study.

(22)

2.3 Social Media and Academic Performance For simplicity, the model of Ainslie shows only 2 possible outcomes: The LL reward and SS reward. In the case of SM and the imminence of the rewards associated with it, we should instead imagine one curve for LL reward being constantly crossed by continous stream of SS reward curves. In reality, if a student gets seduced by one or many of these SS rewards (e.g. procrastinating on Facebook), he or she can still pursue the LL reward (writing a bachelor’s thesis) but it gets diminished overall, as the quality of it decreases given that less time was allocated for it.

(23)

3 Methodology

Chapter methodology consists of three sections. Section 3.1 Research Questions explains what is the subject of research by presenting 4 research questions and hypothesis. Then, Section 3.2 Applied Methods and Concepts is concerned with how the research is conducted by briefly describing the approach taken to collect and analyze data. Finally, Section 3.3 presents the actual procedure in chronological fashion.

3.1 Research Questions

1. Do website blockers help students increase their academic performance?

outcome variable: test score

Based on the previous research and theoretical reasoning in behavioral economics provided in the Chapter 2, it is assumed that website blockers might increase productivity of students and through that also their test scores. Contemporary electronic devices upon which most of people are dependent to get by in social and career related networks put enormous effort on capturing their attention resulting in various possible symptoms: lack of focus, anxiety, depression or addiction. Website Blockers aim to increase productivity and overall well-being. The sole existence of commitment devices in form of website blockers and the demand for them in contemporary world, give validity to the initial problem. The aim now is to see if the particular use of website blockers later presented in the methodology of this thesis might help.

H₀ :Commitment device will not increase the test scores of those using it 2. Do website blockers help students put in more effort?

outcome variable: effort

Effort as in “hours dedicated to preparation” can also potentially be influenced by website blockers. However, the effect can work both ways. Procrastinating student that is not using website blockers might self-report more hours spent studying than someone who is using them. Procrastinating student might be less efficient or just be more optimistic.

Those using commitment might in turn, more accurately tell for how long they have been studying given that website blockers have the option to run a “focus session” for some

(24)

3.2 Applied Methods and Concepts

period of time.

H₀ : Commitment device users will self- report thesame time dedicated to midterm preparation.

3. Do website blockers help students feel more focused?

outcome variable: focus

The way students focus is assumed to have impact on their academic performance and well-being. Although website blockers may simply motivate students for better performance, they way they focus is believed to be a major channel of better results.

H₀ : Commitment device users will not self- report f eeling more f ocused during midterm preparation.

4. Do website blockers decrease time spent on social media?

outcome variable: social media

It is also expected that students in the commitment group will self-report less time on social-media. Although the access to social-media is essential for majority of students who learn from student-made course material, the overall average time spent on SM is hypothesized to decrease for those in commitment group, as it is expected that website blockers will eliminate some of that habitual “unnecessary swiping”.

H₀ : Commitmentdevice users will not self- report less time spent on social media and entertainment.

3.2 Applied Methods and Concepts

3.2.1 Field Experiment

In this research, data for evulating the validity of hypothesis mentioned in Section 3.1 are collected through field experiment. Field experiment is a fusion of 2 other type of experiments: natural and laboratory. It is similar to natural experiment as it takes place in natural environment where subjects live their regular lifes. In case of a natural experiment, researchers are only strict spectators of phenomena. When conducting field experiment, they however, impose some form of treatment on subjects and then observe possible changes in behavior. In laboratory experiments however, the setting is artificial as it takes place in a prepared environment. Similarly, like in a field experiment, there is an intereference with natural conditions although in case of laboratory experiments the extent of intereference is bigger.

Each of the types of experiments have its strengths and weaknesses and they all are

(25)

3 Methodology

suited for different type of research (Harrison and List, 2004). Natural experiments have the potential to bring authentic real-life results. But researchers often have their hand tied.

3.2.2 Soft Commitment Device

The distinction between soft and hard commitment devices was already explained in Sec- tion 2.2. The specific use of this method in this research is determined by the convenience of it which was hoped to make the treatment more accessible to subjects and thus increase the take-up rate. The alternative, hard commitment, would have to impose real economic costs on subjects. Such costs could be restrained access to blocked sites with no option to disobey. In the real world however, website blockers never have this sort of power over electronic devices. Therefore, not only it would be difficult to asses this type of treatment, but also, results would not have much practical use apart possibly from measuring the efficacy of parental control applications which help parents control the internet activity of their children. For this reasons, it is believed by the author that website blockers in form of soft commitment devices are the better option for this research.

3.2.3 Custom Settings

In this experiment, subjects could choose their own settings as well as website blocker of their liking. The alternative to this type of treatment is pre-determined set of rules that subjects have to obey. In the case of this research, the aim was to make the add-ons as accessible as possible so that the take-up rate was as high as possible. This way the subjects were provided with options to set their own time limit, their own distracting websites, the active hours and active days. Subject could also choose to use website blockers either on smartphone, computers or both.

3.2.4 Randomized Control Trial

This research employs Randomized Control Trial (RCT) which is a type of scientific experiment based on measuring differences between output values of two or more randomly allocated groups. It is needed to first obtain demographic and other relevant information about subjects which are then divided into treatment and control groups so that their characteristics are balanced across groups. This way, when done properly, we can see a pure causal effect of treatment on outcome variable (if there is any). By randomizing subjects into the two groups we thus avoid selection bias. RTCs are a norm in medical research but their use remains controversial in evaluating social programs (Imbens and

(26)

3.2 Applied Methods and Concepts

Angrist, 1994).

3.2.5 Intention-To-Treat Analysis

Intention-to-treat (ITT) analysis measures treatment effects and aims to cure problems of noncompliance and missing outcomes occuring in Randomized Control Trials (RCTs).

After assigning subjects to treatment, RCT method generally excludes those that did not participate fully or at all in the experiment (noncompliants). In contrast, in ITT the treatment group contains all of those assigned to treatment, thus including those who actually participated as well as those who ignored the request to participate. While ITT includes observations that are lost in RCT, the actual effect is more conservative. The supporters of this approach claim that ITT gives realistic estimates of treatment outcomes because in reality, there is noncompliance and the ITT effect thus avoids overoptimistic estimates found in RCT. ITT analysis shows rather the actual efficacy of certain treatment than the potential efficacy like RTC(Heritier, Gebski, and C Keech, 2003).

ITT is allowed to use the econometric method of OLS, given that both treatment and control are randomized. A model employing ITT analysis is the same as in the case of RTC where only the treatment sample differs. Such model looks like this:

Y_i = α+β₁T_i+β₂χ_i+ (3.1)

where Yi indicates a type of outcome for every i,α is an autonomous unexplained effect, β1is the estimate ofT, a type of treatment (e.g. website blocker),χi is a vector of control variables,β₂ is a vector of control estimates andis the error term. This particular model is used in this research (see subsection 4.2.1).

3.2.6 Average Treatment Effect On The Treated

Average Treatment Effect On The Treated (ATT) is an analysis that substracts noncompliants from the treatment sample in contrast with ITT. ATT delivers Local Average Treatment Effect (LATE) which shows the average efficacy of geniune treatment. It is an alternative to ATT which in cases where the take-up rate is small might deliver very conservative or no effect at all. It employs 2SLS econometric method using an instrument variable to explain the treatment participation. A criterium for using this method is existence of a valid instrumental variable that explains dependent outcome variable channeled only through independent treatment variable. In this research, ATT is used to estimate the average treatment effect onNew Users which is a group of subjects that had not been using website blocker prior to this experiment but had started using it during it.

(27)

3 Methodology

In the first stage , we need to explain New Users through instrumental variableTreatment (Z) and control variables in vector χ_i:

X = α+δ₁Z+β₂χ_i+µ (3.2)

We then save the fitted values from the previous stage as ˆX and use them in the second stage where we return explaining course outcomes through compliants:

Y = α+β1Xˆ +β2χi+ε (3.3)

3.3 Experimental Procedure

3.3.1 Overview

The data for the practical part of this thesis were collected through a field experiment taking place at University of Economics in Prague, Czech Republic. Students of four courses (Introduction to Game Theory, Economics of Life, Economics and Psychology and Institutional Economics) were reached through their university e-mail adresses and asked to fill in a survey concerning social-media and academic performance. Students were randomized and assigned to treatment and control groups based on the questionnaire unknowing that it is a baseline survey for upcoming social-media experiment. Then, the subjects in the treatment group were mailed a request to install website blockers either or both on their smarthphones and computers 5 days prior to their midterm exams. Finally, both subjects of the baseline survey and rest of the students were asked to fill in endline surveys which were attached to printed midterm tests. In the endline survey, students were asked to self-report on their website blocker use as well as subjective experiences.

Both students that filled in the baseline survey and those that did not were asked to do this during the test taking. The collected data was then used to analyse the impact of website blockers on midterm scores, effort, subjective levels of feeling focused and the time spent on social-media and entertainment 5 days prior to the test taking.

In addition to the midterm scores, final test scores were also added to the dataset for analyzing long run effects of commitment device.

3.3.2 Measurement of individual characteristics

In the pre-survey study (baseline survey), demographic and self-control data was collected to ensure balance across control and treatment groups. It was previously showed that effort and previous GPA (grade point average) have impact on current grades . Similarly,

(28)

3.3 Experimental Procedure

Figure 3.1: Experimentatal Procedure: Overview

the university experience (number of succesfuly passed semesters) can have a positive impact (Ballester, 2012). In another study, an international online course outcome was influenced by gender and nationality (Coldwell-Neilson et al., 2012).

The baseline survey consisted of 18 questions related to demographics, self-control and other variables that could possibly influence the scores in midterm tests. Students of 4 courses ( „EoL“ - Economics Of Life, „IGT“ - Introduction To Game Theory, „EP“ - Economics and Psychology and „IE“ - Insitutional Economics) were reached through e- mail (see Figure 3.2) with the help of teachers and asked to fill in online surveys powered by Google Forms.

In order to attain a structure of subjects as random as possible, there was no other exogennous interaction with them other than 2 e-mail messages from Mr. Cingl. The first message was a request to fill in a survey for a research on the effect of social media on academic performance. There was no mention of any experiment just yet. Students had around 3 days to fill out the surveys with the exception of those in IE, which had 5 days. However, there was no increase in participation rate relative to other courses. The second message was a reminder for students to fill out the survey before deadline. There was no incentive provided to students for filling out the surveys.

(29)

3 Methodology

Figure 3.2: Email to students

3.3.3 Randomization

Subjects were randomized to generate balance across treatment and control groups. An- swers from the baseline survey that were expected to be relevant in the experiment were gender, GPA, hours spent on social-media and online entertainment, levels of distraction caused by social-media and online entertainment, expected hours to be spent studying for the midterm, expected score and prior website blocker use. The subjects were randomized in the classes they were taking, so that each of the treatment groups could be reached for the experiment later on seperately as every midterm test took place at different time during the semester. A special attention was paid to possible duplicities so that students who study more of these classes at once are not assigned to treatment in one of them and to control in the other one. Subjects who filled out the baseline survey after deadline (that is at a point when treatment groups was already asked to participate in the experiment), were automatically added to control groups.

After the end of the experimentation period, those subjects that did not make it to the final tests were ruled out of the sample. After that, the new sample was tested again for homogenity with Two-Tailed Mann Whitney U-Test. See Table 3.1 for the sample comparison. The numbers in treatment and control columns represent the average value (occurence for dummy variables). The fourth column shows p-values of Mann Whitney U-Test. If p > 0.05, then the two groups are not significantly different from each other.

No variables differed at the 5% level, therefore the randomization was succesful.

(30)

Table 3.1: Randomization of Treatment and Control Groups Treatment Control P-Value

Female 0.276 0.326 0.681

Expected Score 72.195 69.638 0.841 Expected Effort 7.883 8.837 0.465 Distracted by SM* 3.297 3.173 0.477 Social Media Hours 14.324 14.642 0.068

GPA 2.341 2.192 0.726

Old Users 0.130 0.208 0.496

Observations 46 47 -

* How often do you get distracted by social media

3.3.4 Treatment Assignment

The whole group of subjects who filled out the baseline survey was split into treatment (n = 52) and control groups (n = 60). Those participants assigned to treatment group were then asked through email to participate in the website blocker experiment. They were sent instructions to download website blockers. The subjects were not incentivized for taking part in this experiment so we were dependent solely on their good will. The subjects were not monitored in any way so that they are not discouraged to participate in the experiment. Although it would be handy to have data about the log times of subjects to accurately evaluate the efficiency of website blockers, it would be a great obstacle to actually persuade subjects to participate in conditions that would not respect their privacy also provided that they do it for free.

Subjects were not incentivized because it was hypothetized by the author that such method can produce biased results for the reason that, in reality, no one is probably going to reward (at least in these times) people for using website blockers. The voluntary imposition of commitment device (in the case of this research) must be driven by instrinsic incentives.

3.3.5 Experimentation Period

The experimentation period began right after treatment assignment when subjects received e-mails with instructions to download website blockers. The periods differed from course to course as the dates of midterm tests differed for each one of them. This fact was taken into account and each of the treatment groups were reached 4 days prior to the test taking. The timetable of the experimentation period is depicted in Figure 3.3.

(31)

3 Methodology

Figure 3.3: Experimentation Period

3.3.6 Measurment of outcomes

Outcomes for research questions 2,3 and 4 concerning student effort, focus and social media use were measured during midterm test taking session on self-report basis. The endline surveys were attached to the actual tests. Outcomes for research question 1 regarding midterm test scores were acquired thanks to the teachers that were grading the tests. They wrote scores on the surveys and returned them to the present author. In the case of the final test scores, they were sent by the teachers via e-mail correspondence after final exam period.

Students were asked 6 questions (see Figure 3.4). Both subjects from treatment and control groups as well as other students that did not participate in the baseline survey, were now surveyed.

Of 112 respondees of the baseline survey 93 were actually present at the midterms and 89 filled out the endline survey. Of 260 students present at midterms, 92% filled out the endline survey (n = 239). After grading the tests, the teachers returned the endline surveys to the present author including test scores and names so that treatment and control groups can be linked to their baseline surveys answers.

Based on the answers, the subjects were assigned to sub-groups which where then analyzed in the following part of this research. Here are all the groups:

general groups

(32)

Figure 3.4: Endline Survey Treatment

Subjects that were assigned to treatment. The actual treatment participation was 23%

resulting in 12 users using website blockers although 5 of them were using website blocker prior the experiment so we can only speak of 7 subjects that installed website blockers because of the assignment.

Control

Subjects assigned to control group to ensure the validity of the results of the treatment given that the participants of this study might have some unobserved characteristics that other students have not.

Other Students

This group contains students that did not fill in the baseline survey but filled in the endline survey while taking the exam.

commitment device groups New Users

Group of subjects that reported in the baseline survey that they were not using website blockers but reported that they were using them before the exams. They are also called

(33)

3 Methodology

compliants. In the treatment group there were 7 of them but interestingly, there were 10 of them in the control group. This fact could be due to the spillover effect. Subjects in the control group which did not know about the experiment may have heard from their classmates in the treatment group about the website blocker experiment. Interested, they may have installed them on their own. It is nevertheless, a peculiar fact, that the control group contains more of new users than the treatment group.

Old Users

Those treatments and controls that reported they were already using comittment devices in form of website blockers. There was 14 of them (n_control= 9, n_treatment= 5).

Other Users

Suprisingly the largest group of students. There were 43 of those who were using website blockers on their own and did not participate in the baseline survey. This fact implies the relevance of this form of commitment devices for college students, although once again, it is not clear whether the spillover effect might be causing this. Curiously, 28 of total 74 who reported they were using website blocker during the exam period, had chinese names.

While chinese students accounted for “only” 22.7% of all students, 37% are in fact in this category.

This imbalance was discovered randomly by the author when looking at the data. The test scores of those who were using website blockers were strikingly smaller so the present author was looking for possible explanation by looking at students last names. No question regarding nationality was included in the endline survey because it was designed to be short so that the students find the time during the test taking to fill it in. That is why, the author who suspected some imbalance, traced the origins of the students’ last names, and based on the frequent occurences hypothezes, that possibly these students did not understand the question properly. Research by suggests that chinese exchange students, particularly, have lower academic performance due to their below average english skills (Li, Chen, and Duanmu, 2009). Alternatively they may have been mistaking these commitment devices with chinese government blockers which restrain access to many western social-media websites such as Twitter or Facebook (MacKinnon, 2011). Maybe, it is just popular among the chinese.

outcomes

Let us now have a look at the research outcomes:

Average Test Score

The average score for every group where 30 was maximum. In the case of one course (Economics and Psychology), the test scores had to be recalculated because the maximum for the course was 25. The average scores were modestly different for every subject and to

(34)

insure fixed effects, dummy variable for each subject is included in the course variables.

Average Course Effort

The effort of students was simply measured in hours that they self-reported to be dedi- cating for midterm exam preparation.

Average Social Media

Self reported amount of hours spent on social-media and online entertainment websites.

Average Focus

Using a scale from 1 (the best) to 10 (the worst), subjects were asked to evaluate how focused they felt during studying.

Experience

Subejcts that reported they were using website blockers, were also asked if they found them useful.

3.3.7 Analysis of measured outcomes

To examine the impact of commitment device on course outcomes, the present author employs 4 models using econometric software called R-Studio. Firstly, using ordinary least squares, the Intention to Treat Effect (ITT) is estimated, where the treatment group contains both those who received the treatment and those who did not. Secondly, again using OLS, the correlation between new users of commitment devices and the course outcome is estimated. Thirdly, correlation between all commitment users and course outcomes is estimated in a similar fashion like in previous model. And fourthly, using two stage least squares (2SLS) the Average Treatment Effect on the Treated (ATT) is estimated.

In addition to the models above, long run effects were also estimated using the same models by explaining final test scores. This was not originally planned but because the present author postponed the deadline of this thesis until august it was possible to analyze test scores of the final tests. However this time, no post-survey was attached to the tests because it had not occured to the author at the time (amidst studying for his own finals) that the experiment could be further extended. In summary, data for self-report outcome variables are non-existent.

3.3.8 Model Compononents

In this sub section, a summary of all variables included in the models is displayed in tables. Course Variables

(35)

3 Methodology

To ensure fixed effect, 4 dummy variables for each course was included. It is expected that some subjects might require less effort. Control variables are categorized into 3 groups:

Demographic, Self-Control and Course Variables.

Table 3.2: Model Components: Output Variables Output Variables Measure Range Description midterm test score points <0,30> ^-

effort hours <0,∞> Hours spent studying (self-report)

focus grade <1,10> Subjective perception of feeling focused (self-report), 1 - best, 10 - worst

social media hours <0,∞> Hours spent on social media (self-report)

final test score points <0,50> -

(36)

Table3.3:ModelComponents:ControlVariables DemographicVariablesMeasureRangeDescription genderdummy<0,1>- universityexperienceyears<0,∞>Yearsspentatuniversity. chinadummydummy<0,1>Dummyvariablesincludingchinesestudents. gpagrade<1,5>Gradeperaveragefromallsubjectsduringallsemesterstakenatuniversity workhours<0,∞>- expectedscore%<0,100>- expectedefforthours<0,∞>- creditscredits<0,∞>numberofcreditstakenduringcurrentsemester CourseVariables mandatorydummy<0,1>somestudentsmayputinlesseffortifthesubjectisisnotmandatory igtdummy<0,1>IntroductiontoGameTheory eoldummy<0,1>EconomicsofLife ecopsychodummy<0,1>EconomicsandPsychology iedummy<0,1>InstitutionalEconomics ControlVariables botherpoints<1,3>“Doesitbotheryoutowastetimeonsocialmedia?”.1-Notatall,3-Alot distractionpoints<1,5>“Howoftendoyougetdistractedbysocialmedia?”1-Never,5-Always social-mediahours<0,∞>“Howmanyhoursperweekdoyouspendusingplatformsmentionedabove?” funversusproductivitypoints<1,5>“Howoftendoesfungetinwayofyourproductivity?”1-Never,5-Always

Hlavní práce68933_krir02.pdf, 1.9 MB Stáhnout

University of Economics, Prague Faculty of Economics

Field of Study: Economics

Blocked Social Media and Academic Performance

BACHELOR THESIS

Autor: Robert Kříž

Thesis supervisor: PhDr. Lubomír Cingl Ph.D.

Year: 2019

BACHELOR THESIS TOPIC

Robert Kříž

Blocked Social Media and Academic Performance

Contents

Appendix A: Pictures II

1 Introduction

2 Theoretical Background

2.1 Self-Control

2.1.1 Time Discounting Models

2.1.2 Naive and Sophisticated customers

2.2 Commitment Devices

2.2.1 Website Blockers

2.3 Social Media and Academic Performance

3 Methodology

3.1 Research Questions

3.2 Applied Methods and Concepts

3.2.1 Field Experiment

3.2.2 Soft Commitment Device

3.2.3 Custom Settings

3.2.4 Randomized Control Trial

3.2.5 Intention-To-Treat Analysis

3.2.6 Average Treatment Effect On The Treated

3.3 Experimental Procedure

3.3.1 Overview

3.3.2 Measurement of individual characteristics

3.3.3 Randomization

3.3.4 Treatment Assignment

3.3.5 Experimentation Period

3.3.6 Measurment of outcomes

3.3.7 Analysis of measured outcomes

3.3.8 Model Compononents