Facebook As A Tool For Data Collection Computer Science Essay

Published Date: 02 Nov 2017

http://www.mymarklab.com/FMS_Logo_HQ%20copy.png

Facebook: Data Collection has become easier? - A study

Group 8 | Section B

Saurav Barua

F 131

Shekhar Chadhary

F 133

Sumit Kumar

F 140

Sourabh Kukreja

F 138

Sumanta Mandal

F 139

Shalinee Suman

F 132

Sneha Dugar

F 137

Siddharth Yadav

F 136

Abstract

The project deals with the analysis of the Facebook as a tool for data collection. The Facebook has more than one billion users and generates one petabyte of data each day. The basic objective of the project is to analyse how Facebook, which has millions of data updated every day, has made Data Collection easier for marketing research companies. Through secondary research, we have explored all the possible data collection method and its effectiveness and also the security measures used to prevent data theft. However, there were gaps in the literature survey which we tried to fill up through our study on the subject. Through Primary research we have tried to get consumer insights on the effectiveness and authenticity of the data collected using various research methodology. Conclusive Research methodology has been used in this project. Primary research has been done in the form of survey, Facebook account observation etc. and objective is established using hypothesis testing. In the end we presented a conclusion based on our analysis.

Contents

Introduction

Launched in 2004 the social networking site Facebook has become a phenomenon with more than 1 billion users by September 2012, more than half of them using Facebook on a mobile device. This social networking service allows users to register on the website and create online profile with pictures and contact information. User can communicate with friends and other users through private or public messages and a chat feature. Additionally, users may join common-interest user groups, organized by workplace, school or college, or other characteristics, and categorize their friends into lists such as "People from Work" or "Close Friends".

 Along with its global appeal Facebook also accommodates tonnes of data (5 petabytes per day) visible by the facts that there are 3.2 billion likes and comments per day, 300 million photos uploaded per day, with an average user spending 8 hours 18 minutes per month. These data are highly valuable to many companies as it helps find the right target and they can customise their product accordingly and increase their sale respectively. 

This project is a research and analysis of viability of Facebook as a method of data collection, its ease and reliability. The above hypothesis is the agenda of the research. The research design used is conclusive (in Descriptive design segment). Literature review forms part of secondary research followed by primary research consisting of Survey and observation. The study employs descriptive research design methodology with secondary data analysis, survey and observation techniques.

Literature Review and Objective of Study

Literature Review

Social Media and Traditional Research Methodologies: Stronger Together than Apart - By SDL

Kara Clark, Market Research Manager, SDL (2012) has discussed about filling the research gaps of traditional research methodologies with the help of social media research in her whitepaper titled – "Social Media and Traditional Research Methodologies: Stronger Together than Apart".

The author has compared the traditional research (focus groups, in depth interviews, surveys) with the social media research.

Benefit of Social Media over Traditional Research Methods

Shorter lead time

Economical

Unbiased opinion

Comprehensive

Drawbacks of Social Media over Traditional Research Methods

Not all consumers covered

Difficult to identify responder

No NDA (Non-disclosure agreement)

The author has concluded that social media has clear advantage over traditional methods as it creates better insights at minimum cost. However, we also need to be aware of the potential disadvantages of the social media research methods.

Reliable Online Social Network Data Collection

Fehmi Ben Abdesslem, Iain Parris, and Tristan Henderson have discussed in this paper that how online social network has become an attractive source of data collection as large volume of information are shared through them. They have discussed how different ways are used by research companies for collecting data. Some of the methods used by them and key findings are listed below:-

Data Collection Method

Key Findings

Passive Data Collection

Data is collected from applications etc. without disturbing the user but the data’s meaning depend on the interpretation of the evaluator

Private Content Collection

Researcher try to analyse user’s personal data, their content sharing behaviour

Self-Reported Data Collection

Questionnaires are floated and users are asked to fill it

Real Social Interaction

The user is asked on a real time basis about their experience

Challenges Faced

They have also discussed about the challenges marketing research companies’ face while collecting user data. Some of them are:-

Private Information restriction – Users generally doesn’t share their private information and hence data that they get is incomplete

Inaccuracy of self-reported data- Users might forget the real experience while taking part in the research and might report inaccurate facts

Infrastructure Required

They also discussed about the infrastructure used for conducting the research which are:-

Mobile Phone

Server

Facebook Application

How to Get Exceptional Consumer Insights and Market Research Using Facebook Data – By MicroStrategy

In September 2012, AnuragTandon and Mark LaRow presented a whitepaper which talks about organizations using the Facebook data in the most effective way to get consumer insights. The paper also compared the traditional data gathering data by researchers and the data gathered from analysis of the Facebook.

At present the Facebook has over 1.01 billion users who share considerable amount of data on the website. Some of the common information shared by users includes name, hometown, location, check-ins, likes and friends. For market research firm, this is the exact kind of data that they are looking for. Just because of the sheer size of Facebook, experts believe that Facebook is the best social networking website from where researchers can extract data. This kind of demographic information is extremely useful for the marketing departments of organizations.

Buying data from vendors

Building database with own survey. The data decays at 2% per month due to the long time required to build the database

Self-Maintaining Database

Information is updated automatically by the users

Typical Data Collection

Data Collection from Facebook

The most useful data comes from the "likes" of the user. The user "likes" various brands, organizations, places, celebrities, products, events, books, movies, TV shows and concerts.

How the data can be gathered and how it is useful for marketing team

Using "likes" and customer information such as gender, age, location, relationship status, education, job, religion, we can makes inferences such as - "Married Hindu women in India, living in metro cities, between the ages of 25-35 like xyz brand soap".

So market researches can use various filters to gain knowledge about specific users on the FB.

Prying Data out of a Social Network

In the paper entitled ‘Prying Data out of a Social Network’, Joseph Bonneau and Jonathan Anderson have discussed challenges faced by social networks in preventing adversaries from collecting huge amounts of user data from their site and methods use to do that.

Following are the major findings of the paper:

Protection against data crawling has not kept pace as Facebook and other social networks evolve into global networks.

Personal data and social graph data can be extracted from social networks, regardless of a user’s privacy settings.

Very fewer accounts are needed to view majority of the network.

There is a lack of user understanding of the limited amount of privacy which can actually be enforced by today’s sites.

Social networks should limit the number of mechanisms to access user data, combat phishing aggressively, reduce reliance on sub-network membership for access control, and eliminate friend-of-friend data sharing.

The authors concluded that though Facebook itself collects user information and has even partnered with Datalogix for that, it has not been able to prevent external APIs from collecting its users’ data.

How to use Facebook in your market research - Jordi Casteleyn, André Mottart and Kris Rutten, Ghent University

The paper states the fact that although everyone knows that Facebook is increasingly being regarded as a huge source of information, no one has come up with a specific technique that adequately helps in retrieving the information in a useful manner. The paper suggests one such technique, dramatism which was developed by Kenneth Burke.

To validate the use of the theory, the writers suggest that an average Facebook user does not necessarily fill his profile information with true facts but rather with how he would like to be perceived by others. In other words people are marketing themselves on Facebook. Also the act of people on Facebook joining various sets the stage for dramatism.

The writer then goes on to describe the actual theory of dramatism, which is based on two sources of guilt in a human communication. The first one being that ‘we necessarily need to define ourselves and others in terms of what we are not’. This is done on Facebook with people showing their interests and favourite music and similar acts. The second being that humans always structure their lives in hierarchies and this triggers off competition between the lower and higher ones. According to Burke, the ultimate goal of public speaking is to rid ourselves of this guilt by either blaming ourselves or blaming someone else.

The dramatism technique consists of five elements: agent, act, agency, scene, and purpose. Using these five elements, various ratios such as scene-act ratio, agent-purpose ratio can be used to interpret human actions.

Using this method, the writer analyses a Facebook group I bet I can find 1,000,000 people who dislike Heineken, in which the members of the group are the agents, the act dislikes Heineken, the agency is how Heineken is being disliked, the scene is the Facebook wall and the purpose is to denounce Heineken.

Upon using the ratios on the above Facebook the page, the writers discovered that in the agents-agency ratio, the agency (how Heineken is being disliked) was being determined by the agents. The agents influencing the conversation take it from generally describing the bad taste of Heineken to rejecting Heineken because it is a Dutch beer. This shows that the scene is being slowly changed by the agents. Upon further investigation, it was found that most of these agents (who reject it because it is a Dutch beer) are from Belgium (which has a folklorish rivalry with its neighbouring country, the Netherlands).

They conclude by stating that this research shows that Burke’s theory can be used to analyse Facebook wall posts and also pointing out where more market research is needed.

Gaps in the Review of Literature

The literatures reviewed by us include the benefits of social media research over traditional research methodologies. It mentions how "likes", content sharing and user profiles can be helpful in market research; however, it misses out completely on the point that how user’s activeness on Facebook is affecting the useful data generated by users.

Hence, we have tried to fill the gaps in review of literature though our study

Objective of Study

Objectives of the study are as follows

To understand and analyse how Facebook has made Data Collection easier for marketing research companies

To examine relationship between user activeness on the Facebook with the amount of use useful data generated

Research Methodology

Research Design

Since in this research we always had a hypothesis, our research agenda was to test the validity of the given hypothesis. We went for Conclusive Research. Within conclusive research we started with Descriptive Research.

In the given marketing research problem, we first quantitatively analysed the secondary data available on various websites and encyclopaedias, also reviewed 5 eminent literatures to understand hypothesis.

In the second phase of research we went for primary research. A survey was designed. Our first task was to decide upon our target population. So we divided the target population into two parts.

Ones who regularly conduct marketing research

Ones who take part in the research (Facebook Users)

We designed two questionnaires, to find the effect of Facebook on different facets of marketing research, both in the demand and supply side of data. Hence we went for multiple cross sectional design. Data was to be collected from two different sample populations. But due to paucity of time and resources we could not conduct the research among researchers but we conducted an extensive research on the second set of population.

The third phase of this research was to observe how data on Facebook is used by marketers to promote their products. We observed a couple of Facebook accounts of users with different age group, profession and interest and saw the difference in products displayed on their page. By this we got the impression how users personal data is used for targeted marketing.

Questionnaire Design

As stated earlier, before designing the questionnaire we divided the target population into two parts.

Ones who regularly conduct marketing research

Ones who take part in the research (Facebook Users)

The required information from both the sets was different. In case of the researchers we wanted to find out the ease of research along with other financial and non-financial benefits. While in case of respondents we wanted to know their readiness to supply a pool of data.

Questionnaire 1:

For the researchers we divided the research process into 3 broad stages and then tried to analyse their convenience in all the 3 stages.

Pre collection activities

Collection

Present Findings

Pre collection activities: Some of the pre collection activities are defining the goal, reaching the target sample, deciding upon the research methodology. Here we tried to find out critical questions like, if Facebook was well equipped to cater all kinds of marketing research goal, is it possible to reach all possible sets of population through Facebook, the experience of conducting a research on FB etc.

Collection: We tried to compare various facets of data collection like number of respondents, readiness to respond, reliability of information, economic benefits etc. in the traditional research and the one collected from Facebook.

Present Findings: It is often believed that a huge chunk of data can be collected through social networking sites but we wanted to understand how easy it is for the researcher to sort and analyse that data.

Questionnaire 2:

In the questionnaire designed for general users we had some filter questions to distinguish the Facebook users from non-users, it also gave us an idea about the frequency of use. Our aim was to understand how easily a user contributes for generation of data in Facebook. There are 3 ways of data collection from Facebook

Personal information of the user

Likes and comments on various products/communities etc.

Filling up of surveys

We tried to find out the amount of personal information a user provides on Facebook and how reliable it is. How frequently a user interacts with different groups or communities on Facebook and is his likes and comments genuine. Ease of filling up surveys and providing information for user etc.

Sources and techniques of data collection:

In our research methodology 3 techniques of data collection were proposed

Online survey

Personal Interviews

Facebook Account Observation

An online survey was conducted among the Facebook users. Various web mail sites and social networking sites including Facebook were used for circulating the survey.

We wanted to conduct personal interviews of some professional researchers in the industry but due to paucity of time that was not done.

We also observed some Facebook accounts to study how personal information given on Facebook is used by marketers to promote various products

Sampling Technique

Till now we have conducted 2 stages of this research.

In the first stage we conducted a survey which was circulated on Facebook and other web email sites. The population present in these sites were regular users of social networking sites. So in this way we went for judgemental sampling under non probabilistic sampling.

The second stage of research was Facebook Account Observation for which we used simple random sampling.

Sample Size and graphs

All the participants who took part in the survey had a FB account

More than 90% user who have FB account visit the website daily

Over 89% users accept that the data provided by them on the FB are genuine

Most of the FB users not like post comment on product/company/celebrity pages

Most FB users spend between 10-30 mins on Facebook during each visit

Over 90% user prefer online survey or pen-paper survey

Most of the FB users allow apps to gather their informationHypothesis Framing

We want to understand if the amount of useful data generated by the Facebook users depends on the activeness of the user on the website. Hence, more active a Facebook user is, the more easily organizations can gather data from the Facebook. To understand it we formed the hypothesis

Null hypothesis: Users participate in the product/company/celebrity communities on the Facebook by suggesting over product development/changes does not depend on their activeness on Facebook

Alternative hypothesis: Users participate in the product/company/celebrity communities on the Facebook by suggesting over product development/changes depends on their activeness on Facebook

Software Used for Statistics

Microsoft Excel 2010 has been used for analysis of the data generated from Primary Research (survey form)

Surveymonkey website was used to run a survey and gather user data.

Analysis and Findings

The test involves comparing the expected cumulative distribution function under the null hypothesis being true with that of observed cumulative distribution function. If we designate Fo(X) as the expected cumulative distribution function and Sn(X) as the observed cumulative distribution function, Kolmogorov-Smirnov D is calculated as D = Max |Fo(X)-Sn(X)| (D is the absolute difference between the expected cumulative proportion and the observed cumulative proportion). Please note that n is the sample size. The following table shows the necessary calculations.

Kolmogorov-Smirnov test is more powerful than the chi-square test for the following reasons. 

It is easier for people to compute

It does not have the problem of min freq in each cell as the chi-square test

For very small samples,  the chi-square test cannot be used, but the Kolmogorov-Smirnov test can be used

When samples are small and there are categories to be combined for proper use, the chi-square test is definitely less powerful than the Kolmogorov-Smirnov test.

Table 1

Null hypothesis

Ho: Users participate in the product/company/celebrity communities on the Facebook by suggesting over product development/changes does not depend on their activeness on FB

Alternative hypothesis

Users participate in the product/company/celebrity communities on the Facebook by suggesting over product development/changes depends on their activeness on FB

Computed D = Max |Fo(X)-Sn(X)| = 0.328571. The critical D value for a level of significance of 5% is given by 1.36/n^1/2 where n=84.

The critical D=0.1484 is lesser than the calculated D, hence reject the null hypothesis.

Table 2

Null hypothesis

Ho: Users participate in the product/company/celebrity communities on the Facebook by Comment on discussions/posts by other users does not depend on their activeness on FB

Alternative hypothesis

Users participate in the product/company/celebrity communities on the Facebook by Comment on discussions/posts by other users depends on their activeness on FB

Computed D = Max |Fo(X)-Sn(X)| =0.188095

The critical D=0.1484 is lesser than the calculated D, hence reject the null hypothesis.

Table 3

Null hypothesis

Ho: Users participate in the product/company/celebrity communities on the Facebook by Complaints about bad services/ product failures does not depend on their activeness on FB

Alternative hypothesis

Users participate in the product/company/celebrity communities on the Facebook by Complaints about bad services/ product failures depends on their activeness on FB

Computed D = Max |Fo(X)-Sn(X)| =0.245238

The critical D=0.1484 is lesser than the calculated D, hence reject the null hypothesis.

Table 4

Null hypothesis

Ho: Users participate in the product/company/celebrity communities on the Facebook by Discussion over products before it is launched does not depend on their activeness on FB

Alternative hypothesis

Users participate in the product/company/celebrity communities on the Facebook by Discussion over products before it is launched depends on their activeness on FB

Computed D = Max |Fo(X)-Sn(X)| =0.257143

The critical D=0.1484 is lesser than the calculated D, hence reject the null hypothesis.

Table 4

Null hypothesis

Ho: Users participate in the product/company/celebrity communities on the Facebook by filling survey forms does not depend on their activeness on FB

Alternative hypothesis

Users participate in the product/company/celebrity communities on the Facebook by filling survey forms depends on their activeness on FB

Computed D = Max |Fo(X)-Sn(X)| =0.257143

The critical D=0.1484 is lesser than the calculated D, hence reject the null hypothesis.

Findings

After the KS test and analysis we found that users participation in the product/company /celebrity communities on the Facebook by

Suggestions over product development/changes

Comment on discussions/posts by other users

Complaints about bad services/ product failures

Discussion over products before it is launched

Fill survey forms depends on their activeness on Facebook. Hence the more the activity of the user the more is the participation and hence, more easier it is for any organization to collect data from the Facebook

Hence, according to our analysis we reject the null hypothesis. We can concluded that more active a Facebook user is, the more easily organizations can gather data from the Facebook

Conclusion

To understand if data collection has become easier with the Facebook we had planned to cover 2 sets of population using primary research

Facebook users

Market Researchers

However, due to lack of time and the limited scope of the study, we could not cover the second portion i.e. opinion of market researchers. We only relied on secondary research data to understand how data gathering from the Facebook has become easier for market researchers.

So, in future study we can include the responses of market researchers through primary research to present a more comprehensive view of the topic.