the role of geo-demographic big data for assessing the effectiveness of crowd-funded software...

7/23/2019 The Role of Geo-Demographic Big Data for Assessing the Effectiveness of Crowd-Funded Software Projects: A Cas…

http://slidepdf.com/reader/full/the-role-of-geo-demographic-big-data-for-assessing-the-effectiveness-of-crowd-funded 1/27

94

Chapter 4

The Role of Geo-Demographic Big Data

for Assessing the

Effectiveness of Crowd-Funded Software Projects:

A Case Example of “QPress”

Jonathan BishopCentre for Research into Online Communities and E-Learning Systems, UK

ABSTRACT

The current phenomenon of Big Data – the use of datasets that are too big for tra-

ditional business analysis tools used in industry – is driving a shift in how social

and economic problems are understood and analysed. This chapter explores the role

Big Data can play in analysing the effectiveness of crowd-funding projects, using

the data from such a project, which aimed to fund the development of a software

lug-in called ‘QPress’. Data analysed included the website metrics of impressions,

clicks and average position, which were found to be significantly connected withgeographical factors using an ANOVA. These were combined with other country

data to perform t-tests in order to form a geo-demographic understanding of those

who are displayed advertisements inviting participation in crowd-funding. The

chapter concludes that there are a number of interacting variables and that for Big

Data studies to be effective, their amalgamation with other data sources, including

linked data, is essential to providing an overall picture of the social phenomenon

being studied.

DOI: 10.4018/978-1-4666-8465-2.ch004

Copyright ©2015, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.




95

INTRODUCTION

In the current digital age, we have seen an unprecedented global recession that could

be seen to have challenged the willingness of persons to take risk in innovation (Etz-kowitz, 2013), but this is not always the case (Singh, 2011). One approach that has

been suggested as an appropriate means to help overcome such financial shortfalls

is crowd-funding. Put simply, crowd-funding is the procurement of financial capital

from those who want to benefit from a particular innovation (Kshirsagar & Ahuja,

2015; Ordanini et al., 2011) . The question that is often asked is how to assess the

effectiveness of crowd-funding projects and also how they should be benchmarked.

This chapter argued that an important part of this process is the use of what has

become called ‘Big Data.’ Big Data is still a maturing and evolving discipline and

Big data databases and files have already scaled beyond the capacities and capabili-ties of commercial database management systems (Kaisler et al., 2014) . Big data

is defined as datasets whose size is beyond the ability of typical database software

tools to capture, store, manage, and analyse, where the primary characteristics are

‘volume, velocity, and variety’ (Malgonde & Bhattacherjee, 2014; Zhang et al., 2014) .

It has been argued that geography might provide a useful lens through which to

understand big data as a social phenomenon in its own right in addition to provid-

ing answers to the complexity of social and spatial processes (Graham & Shelton,

2013) . Even so, it has been argued that the aggregation of social media as big data

is not necessarily social science data, even in the fields of human geography andgeographic information science (Wilson, 2014). This chapter shows how using

geo-demographic analyses with Big Data can improve the effectiveness of crowd-

funded projects.

BACKGROUND

This chapter is in essence looking at effective means for assessing the impact of a

crowd-funded campaign supported by advertising. It is argued that geo-demographicfactors play a significant role in the effectiveness of crowd-funding projects, particu-

larly those supported by advertising. It is further argued that Big Data can be used

to identify trends that go beyond the usual metrics for advertising campaigns – such

as impressions, clicks and average position – while at the same time supporting the

use of such measures.




96

Big Data

According to the New York Times, many think Big Data is synonymous with “Big

Brother,” in the form of mega-corporations collecting masses of surveillance informa-tion on their customers or potential customers. However, as this chapter advocates,

it can also be of use to smaller entities, such as crowd-funded projects. It has been

estimated that Google alone contributed 54 billion dollars to the US economy in 2009

as a result of Big Data, but there is still no clear consensus on what it is (Labrinidis

& Jagadish, 2012) . Even so, Big Data is something that each business will have to

adopt as a normal way to develop business strategy (Woerner & Wixom, 2015) .

Crowd Funding

It has been argued that the main objective of crowd-funding is to give entrepreneurs

a way of raising money that does not involve banks or venture capitalists, which usu-

ally involved giving the customers of the product a stake in it in one way or another

(Kshirsagar & Ahuja, 2015) . This can be seen as desirable in the current capitalist

environment, where banks charge excessive rates of interests and venture capitalists

extort shares in profits totally unrelated to the assistance they have actually given a

project. Crowd-funding has been used in fields as diverse as archaeology (Bonacchi

et al., 2015) and sustainable development (Kunkel, 2015). By its very name it is

clear that by targeting potential beneficiaries of a product or service, to ask them toget on board with the funding of something they want, then a strong customer base

is possible for products before they have even hit the shelves – or download pages

as may be more appropriate. Crowd-funding is increasingly becoming a trusted way

to provide finance to businesses and consumers alike (Laven, 2014). The existence

of specialist websites like IndieGoGo (Dushnitsky & Marom, 2013; Stern, 2013)

and tailored charitable giving websites, it can be seen the opportunities for reaching

potential investors is possible whether one is looking to be sponsored for a race or

other effort, or get a product off the ground, then crowd-funding can be a means to

do it. It is usually conducted via the Internet and involves not only the investors andinitiator, but often third-parties, such as programmers, who use the funds collect-

ing through crowd-funding initiatives to produce the project being crowd-funded

(Ferrer-Roca, 2014).

It is clear that crowd-funding has a lot of potential to replace the traditional

capitalist way of doing things, where a product’s development is driven by someone

intending to make financial profit through efficient use of human and financial capi-




97

tal, rather than it achieve benefits to stakeholders is other ways also (Harvey, 2011)

. This is because when customers become financers, the traditional model of risk

being taken on by a small number of shareholders in a company that brings in money

from selling to customers to reward that risk appears less relevant. It is known that

the cost of private sector services can be equivalent to the value of human capital

in the voluntary sector (Bishop, 2012). On that basis, whilst this study will focus

on the crowd-funding of software through receiving funds to pay programmes, it

is easily conceivable that in the future crowd-funding will involve venture human

capital as much as the micro-financing it does at present.

Online Advertising

Big data is on the one hand distinct from the Internet, and on the other it can be

seen that the Web makes it much easier to collect and share data (Cukier & Mayer-

Schoenberger, 2013) . One such situation in which Big Data can be collected is in

the case of online advertising, which produces a huge number of metrics. This has

long been the case with online advertising, where the tracking of a user’s interests

based on the websites they visit and using that to present custom adverts is long es-

tablished (Arthur et al., 2001). As can be seen from Table 1, the standardised metrics

collected from such platforms are clicks, impressions, cost and average position.

The concept of Big Data in terms of paid advertising campaigns is argued to

yield results in terms of business effectiveness, allowing for the refining and rolling

out of new and existing campaigns (Brechner, 2013). Big Data has been argued to

have a lot of potential for influencing the online advertising markets, with real-time

bidding being widely considered as an emerging frontier in computational advertis-

ing research (Yuan et al., 2014) . The effectiveness of a real-time bid is reflected in

Table 1. Key metrics collected in the serving of online advertisements

Metric Description

Clicks When an advert has been selected, whether by tapping it with a finger or stylus on a

tablet or by using a mouse pointer on a desktop it is said to have been clicked. Thenumber of clicks are the number of times an advert has been selected.

Impressions An impression is when a given advert has been made available on a given page or

screen in order to entice a potential customer to click it.

Costs The cost of an advert is often related to the number of clicks it generates, or in many

cases when the visitor to a website following clicking on an advert actually engages

with the website, such as buying a product or liking a post.

Average Position The average position of an advert reflects how far up a page or screen it is. It is

assumed the farther up a page or screen, the more likely it is to be clicked. On bidding

platforms, the person paying the highest amount on a keyword goes to the top.




98

the average position of the advert concerned, as can be seen from Table 1. The

lower the number of the advert position, the more successful the campaign has been,

as the more likely it will have been seen by potential customers. Big Data can

therefore play an important role in seeing what makes an advertisement successful.The cost of an online advertising campaign is usually measured in cost-per-

impression, cost-per-click, or cost-per-action (Huang, 2013). By exploring the overall

picture around this, Big Data can improve the effectiveness of an online advertis-

ing campaign (Huang, 2013). Equally, as the point of Big Data is to mine data in

ways that bring out new information, then it might be a dataset could suggest other

ways of determining the cost of an online advert, such as ‘cost-per-friend,’ where

advertising to the friends of someone who is already a customer, may deliver new

customers. Such systems already exist, such as to recommend people join the buddy

list of someone whom a friend has expressed positive sentiment towards or whoman enemy has been declared an enemy by (Bishop, 2011).

Geo-Demographic Analysis

Effective use of analytics in relation to Big Data is now key to the success of many

businesses, whether, scientific, engineering or government endeavours (Herodotou

et al., 2011; Hurwitz et al., 2013) but by the very nature of Big Data, choosing the

correct method for analysis is a challenge (Shim, 2012). This chapter has sought

to propose that the effectiveness of crowd-funding campaigns be assessed throughmaking use of geo-demographic factors. Table 2 sets out some of the ones that will

be used in this study.

Table 2. Geo-demographic factors that influence advertising campaign effectiveness

Factor Description

Productivity Productivity is the extent to which an area is producing work, such as goods and

services. It has been found in areas with the lowest levels of trolling, productivity

is lower also.

Education Level Education level, often measured in terms of NVQ qualification level, refers to

the extent to which a person has received a formal education. Areas with the

lowest trolling often have lower NVQ levels.

Intelligence Intelligence, often measured in terms of ‘performance IQ,’ refers to the capacity

someone has to adapt to the environment around them and solve problems with

minimal mental effort. Areas with the lowest intelligence have been associated

with low levels of trolling.

Quality of Life Quality of life is the perception that one is getting out of life what one wants.

High quality of life is associated with low levels of trolling and high productivity

is associated with low quality of life.

Rooms in House A high number of rooms in a house are associated with low levels of trolling.




99

It has long been argued that labour productivity is linked to geographical factors

(Chaney & Hornbeck, 2012; Norcliffe & Mitchell, 1977) . This may appear obvious,

as some communities have more job prosperity than others. Productivity is also

linked to factors such education and academic accomplishment in terms of geo-graphic determination (Zhu et al., 2015). This can have severe impacts on prosper-

ity – and Internet participation – which may restrict the effectiveness of online

advertising if the intended market is unable to be reached. It is known in some areas

like construction that geographical factors have little impact on productivity (Odes-

ola & Idoro, 2014) . It may therefore be necessary to consider local factors when

interpreting results relating to productivity and its impact on crowd-funding. It is

likely to be that where a geographical location is deficit in one demographic factor

that they may also be deficit in others.

In terms of geographic mobility, education has been known to be an importantfactor (Olwig & Valentin, 2014) . Indeed, geo-demographic factors such as size of

household are known to be linked to education (Katircioğlu, 2014). Geographical

factors are known to be causes of barriers to education (Suhonen, 2014). Education

and geographical factors are known to be linked to socioeconomic indicators (Wil-

son‐Ching et al., 2014) . On that basis it is likely education levels in geographical

regions will affect the extent to which people participate in online advertising and

crowd-funding projects. Many crowd-funding projects that relate to software will

often have a limited user-base. It might be that the less likely a mega-corporation is

to develop a particular product or service, the more popular a crowd-funded project– doing something many people want – will be.

Intelligence is known to be linked to geographic factors (Suhonen, 2014), and

also crowd-funding is known to make the best use of finance and knowledge to

provide opportunities to multiple people from diverse backgrounds (Shiller, 2013).

Indeed, crowd-funding has become a popular means for people to improve the qual-

ity of their life though pooling financial and other resources to achieve their goals

(Feldmann et al., 2013) . Crowd-funding therefore has the potential to be a means

to increase opportunity to disadvantaged groups, including through making it pos-

sible for those who have no apparent impairment raise funds to help those who dohave difficulties, but whom may otherwise be denied equal opportunities because

they are not as profitable as mega-corporations deem they need to be in order to be

worthwhile customers.

Table 3 shows two technological factors that could have an effect on geo-de-

mographic differences in assessing crowd funding effectiveness, namely online ad

expenditure and Internet access. It has been argued that when assessing the impact

of expenditure in relation to geographical issues, these should as far as possible

reflect local market conditions (Bojke et al., 2013) . This may be the case with

ad expenditure, meaning factors such as productivity and Internet access being




100

important considerations in assessing geo-demographic issues in relating to the

advertising of crowd-funding projects. Factors such as number of rooms in a house

and Internet access available in those rooms are known to be important factors geo-

demographically (Bishop, 2014a).

A GEO-DEMOGRAPHIC ANALYSIS OF BIG DATA CONNECTEDWITH ADVERTISING OF CROWD-FUNDING PROJECTSTHROUGH THE CONTEMPORARY PRISM OF INTERNETTROLLING

Big Data as a term on the whole refers to datasets that are so complex that they

become awkward to work with using standard statistical software (Sagiroglu &

Sinanc, 2013; Snijders et al., 2012) . On that basis, testing its impact on crowd-

funded projects can be challenging, especially as many crowd-funding website donot produce analytics data. The purpose of this study is on the one hand to show that

Big Data can replicate the findings of ‘small data’ (Bishop, 2014a)when it comes to

geo-demographic datasets. On the other hand the study hopes to show that one of

the optimal ways to analyse Big Data is over a period of time based on monitored

categories of data. It is argued that methods such as Panel Data are suited to Big

Data, such as for identifying trends in behaviour over periods of time. In order to

provide evidence in support of this claim an ANOVA is used along with t-tests to

show relationships between metrics such as clicks, impressions, costs and position

with the geo-demographic data of the study to be replicated (Bishop, 2014a). As canbe seen, however, even a dataset this small has the same difficulties as the massive

datasets associated with Big Data, as the multitude of variables still produce a large,

more varied and complex structure that is akin to the difficulties associated with

storing, analysing and visualising Big Data (Sagiroglu & Sinanc, 2013) .

Table 3. Technology-based geo-demographic factors

Term Description

Online Ad expenditure The amount a country spends on online advertising, measured in US

dollars.

Internet Access The percentage of households with Internet access.



101


The Study Being Replicated

Internet trolling can be defined as the posting of a message to the Internet in order

to provoke a reaction (Bishop, 2013a; Bishop, 2013b; Bishop, 2014a; Bishop,2014b; Bishop, 2014c; Cowpertwait & Flynn, 2002; Crumlish, 1995; de-la-Peña-

Sordo et al., 2013; Hardaker, 2010; Hardaker, 2013; Jansen, 2002; Jansen &

James, 1995; McCosker, 2013; Pfaffenberger, 1995; Phillips, 2012; Walter,

2014) . One might therefore regard advertising – which attempts to provoke

certain reactions in consumers and others – to be a form of trolling. Indeed, a

premise used in this chapter is that online advertisements can be considered to be

acts of trolling users – to draw their attention to something they may not have

originally been concerned with, such as a crowd-funding project. In order to

provide evidence in support of this, the chapter will seek to replicate the findingsof a study into linking Internet trolling to geographical factors (i.e. Bishop,

2014a). This previous study sought to identify factors that could be used to predict

whether a given locality was likely to have high levels of trolling based on its

geo-demographical factors.

As can be seen from the data in Table 4, which had a CV of 3.07, it was safe to

conclude that the locality in which one lives has an effect on education outcomes,

which were likely down to the geo-demographics of the area. Wales, which had

the lowest productivity (164), which was half that of the South East of England

(320) had an education outcome of 2 compared to 3 in the case of the South East.Intel-ligence was also lower in Wales (IQ=92) compared to the South East

(IQ=105), but this may be down to biases in the measure of intelligence, which

can be linked to factors that favour more prosperous communities over more

deprived ones.

Differences between the areas in relation to intelligence meant it was suitable to

accept the claim that intelligence differs by geography, which is likely because those

who achieve higher qualifications are more likely to be well practised in the skills

Table 4. Factors affecting propensity of Internet trolling in given geographies

Variable N Wales Scotland South

East

F p-value SE

Trolling Incidents

(Adj)

- 237 365 10201 - - -

Productivity - 164 212 320 - - -

Education Level 161 2 3 3 6.63 <0.003 0.11

Intelligence 150 92 102 105 4.27 <0.003 0.62

Quality of Life 150 33.11 32.01 30.05 3.11 <0.017 0.5

Rooms in House 161 5.96 5.05 5.09 4.45 <0.014 0.01




102

that form part of intelligence testing. Perception of quality of life is also significant

and may correlate to the productivity levels as the higher productivity figures seem

to be directly proportional to lower quality of life. The number of trolling incidents

per productivity in Wales, Scotland and in the South East of England provided aclear indicator that increased productivity does not result in reduced cybercrime.

The extreme number of incidents of flame trolling in the South East suggests that

the police authorities in the region are not taking trolling seriously, and it is not

young people that to blame for the flame trolling. Indeed, one police force in that

locality is known to be soft on flame trolling.

Sussex Police refused to take action against a police officer, aged 32 from Bir-

mingham, who allegedly harassed Sussex resident, Nicola Brookes, on Facebook.

The police officer allegedly targeted Brookes directly, including hacking in to her

email, for which computer forensics of his IP address was available, but during the19 months of appeals by Brookes he allegedly had his computer reconditioned. This

clearly shows that even where there is strong geographical links showing trolling

propensity, this does not necessarily reflect itself in terms of police action. Indeed,

one might argue that the reason trolling is so high in some geographical regions is

because of police inaction.

Methodology

The study’s methodology was based on a somewhat empirical approach, where datawas collected in the form of web metrics on the basis that it could be analysed to

provide truths about those in the countries the data was recorded from. In terms of

methods the data was collected through using the metrics that come from Google

Ads, and analysed using ANOVA and t-tests on the basis of that. The dataset was

reduced to produce groups based on countries, which were reduced to 13 OECD

countries on the basis there were at least 30 observations per country and there was

a good availability of quality country-specific data from the OECD. This produced

a total number of observations of 2787. The Google Ad displayed is in Figure 1.

Figure 1. A Google Ad advertisement for ‘QPress’.




103

Results

Table 5 shows the result of the outcome of the ANOVA that was run on the dataset. It

is clear to see that for the 13 OECD countries there were three significant outcomes,namely clicks, impressions, and average position. As can be seen below, these were

used as part of a further analysis with the means for each of the countries on each

of these factors being compared on a country level.

Table 6 shows the OECD data on Internet access and online advertisement

spending merged into the same table with the Means for the factors identified in

Table 6, all listed for each of the 13 countries selected. The table is sorted with the

country with the highest amount spent on advertising at the top and the one with

the lowest advertising expenditure at the bottom. No data was supplied by the OECD

for Mexico, and as it is low on many other indicators, it did not seem inappropriateto have it at the bottom due to having a null value. With a CV of 1.786 it is clear to

see that it is safe to reject the null in the case of Clicks (F=2.271, p=<0.008), Im-

pressions (F=6.330, p=<0.001) and Average Position (F=18.745, p=<0.001). In

the case of Costs it was necessary to keep the null because the F-score of 1.229 did

not exceed the CV of 1.786, and in any case the difference was not significant

(p=<0.001). Table 6 shows all three of these factors with the Means for each coun-

try next to the Internet access rate and online ad spend for those countries to enable

to reader to easily see the relevance of these.

Comparing Countries with High and Low Country Metrics

This section seeks to find out the differences between those countries reporting the

lowest scores on various socio-economic factors and those showing the highest. The

socio-economic factor of productivity was derived from the OECD variable of the

same name. Education was calculated from the Wolfgram Alpha databank, which

ranks education level for a given country on a scale of 0 to 1. The value for each

Table 5. Analysis of variance of web metrics

Factor df Mean square F p Null

Clicks 12 0.016 2.271 0.007 Reject

Impressions 12 5437.158 6.330 0.000 Reject

Costs 12 0.009 1.229 0.256 Keep

Average Position 12 0.000 18.745 0.000 Reject




104

country in terms of education was multiplied by 5, which is often the upper level for

NVQs in the UK, which was used in the earlier study(Bishop, 2014a). The values

derived seemed comparable with the earlier study. Intelligence was calculated by

modifying the literacy rate provided by the OECD for each country. The median of

all the countries in the data set was used for the 100 baseline and the other countries

calculated in relation to that. The number of rooms was calculated using the OECD

variables of the same name.

What is noticeable from Table 7, is that the countries with the highest impressions

are generally associated with low levels of productivity. A BBC investigation found

that many of the likes on pages are bogus, and can be linked to the least prosperous

of countries. It could therefore be assumed that the high numbers of impressionscould be linked to the fact that many websites in emerging economies are set up

with bogus content for the sole purpose of raising revenue from contextualised ad-

vertising where those who click the adverts have no actual interest in what is being

advertised, but only click because doing so earns the website concerned money.

This on the one hand proves that considering geo-demographic factors in choosing

outlets for advertising, but on the other hand might suggest datasets collected from

online advertising might not be entirely reliable as it cannot be guaranteed they

represent genuinely interested parties.

Table 6. Advertising and Internet access data for 13 OECD countries (order by ad

expenditure)

Country

Online Ads

(USD 1/100 ofa million)

Internet

Access(Percent) Impressions Clicks Position Cost

United States 324.79 78.70 3.25 0.003 4.41 0.003

Japan 109.74 80.90 6.67 0.006 2.32 0.001

United Kingdom 69.98 82.50 8.15 0.016 4.06 0.018

Germany 51.06 83.90 6.44 0.000 3.78 0.000

Korea 36.12 84.00 22.34 0.027 2.00 0.017

France 28.42 81.70 6.78 0.000 3.16 0.000

Spain 13.86 69.10 12.38 0.011 2.96 0.015

Poland 7.39 67.00 11.89 0.010 2.24 0.000

Switzerland 5.78 86.80 4.89 0.000 4.01 0.000

Hungary 1.50 60.80 17.43 0.000 2.17 0.000

Portugal 0.43 57.90 14.37 0.020 2.79 0.020

Mexico 38.70 18.73 0.031 2.35 0.013




105

Productivity and Internet Access: Mexico (Lowest) v Switzerland

(Highest)

In terms of productivity Mexico was the lowest at $23.68 per person and Switzerland

was the highest with $135.83 per person. In terms of Internet access, Switzerland was

the highest with 86.8 per cent of people having access and Mexico was the lowest

with 38.7 per cent of its people having access to the Internet. As can be seen from

Table 8, there was a significant difference (p=<0.001) between Mexico (M=2.35)

and Switzerland (M=4.01) in relation to average position, and with a t-score of

3.829, it was acceptable to reject the null. There was no significant difference in

impressions (p=<0.033) between Mexico (M=18.73) and Switzerland (M=4.89),

nor with clicks (p=<0.005) between Mexico (M=0.04) and Switzerland (M=0.00).

Equally, there was no significant difference (p=<0.037) in relation to cost between

Mexico (M=0.01) and Switzerland (M=0.00). It was therefore not possible to reject

the null in these cases.

The significance of average position in relation to Internet access might be

simple to explain. The average position of the advert in Mexico was 2.35, but in

Switzerland it was 4.01, which might reflect the fact that as more people are sub-

scribed to the Internet in Switzerland (86.8%) compared to Mexico (38.7%) then

Table 7. Education and economic data (sorted by highest impressions)

Country Impressions Clicks

Productivity

(USD Per

Person) NEETs Education Intelligence

No

Rooms

Korea 22.34 0.027 41.28 8.80 4.67 105.31 2.97

Mexico 18.73 0.031 23.68 9.60 3.63 99.36 3.88

Hungary 17.43 0.000 27.29 25.90 4.33 105.31 0.38

Portugal 14.37 0.020 39.93 38.50 3.70 101.59 4.30

Spain 12.38 0.011 72.04 51.60 4.37 104.04 4.57

Poland 11.89 0.010 27.66 27.00 4.11 105.74 3.56

United

Kingdom 8.15 0.016 73.64 20.60 4.08 106.27 4.61

Canada 6.80 0.023 97.65 13.80 4.64 106.06 1.08

France 6.78 0.000 88.83 25.10 4.35 106.27 3.68

Japan 6.67 0.006 94.23 6.90 4.42 106.27 3.91

Germany 6.44 0.000 75.75 8.10 4.60 106.16 4.00

Switzerland 4.89 0.000 135.83 7.90 4.36 106.16 4.32

United States 3.25 0.003 102.50 16.10 4.70 106.16 4.65




106

the number of ads displayed will be higher, meaning many will be lower down the

list in Switzerland where there is likely to be more advertisers than in Mexico.

Productivity in Switzerland ($135.83 per capita) is significantly higher than Mex-ico ($23.68 per capita), which may also be a factor in why impressions are higher

in Mexico, where revenue from adverts will be more important. No data on the

amounts spent on online advertising were available for Mexico, but if one considers

Hungary, which has a similar productivity ($27.29) to Mexico ($23.68) as well as

having similar numbers of impressions ($17.43) to Mexico ($18.73) then their

online advertising expenditure is likely to be as low ($150m). Compared to Mexico,

Switzerland’s spending is phenomenal ($578), and the fact that costs of advertising

is higher in Mexico, might suggest that crowd-finders would make a loss by adver-

tising there.

Online Ad Expenditure: Portugal (Lowest) v United States (Highest)

The United States had the highest amount of annual online ad expenditure with

$32,479,000 dollars, and Portugal was the lowest, with $42,800 of online ad expendi-

ture. Table 9 shows that in the case of impressions there was a significant difference

(p=<0.001) between the United States (M=3.25) and Portugal (M=14.37) in terms

of impressions with a suitable t-score of -6.624. Equally there was a significant dif-

ference (p=<0.001) between Portugal (M=2.79) and the United States (M=3.25) interms of average position with a t-score of 6.533. There was no significant difference

(p<0.078) between the United States (M=0.06) and Portugal (M=0.02) in relation

to clicks as can be seen from Table 9. Equally, there was no significant difference

in terms of cost as although there was a p-value of less than 0.003, the t-score of

0.002 was not enough.

The evidence from Table 9 that the QPress advert was in position 4.41 on aver-

age in the United States, compared 2.79 in Portugal is probably evidenced by the

fact that advertising spending in the USA is nearly 800 times higher than in Portu-

gal. That is because the more advertisers then the greater the position value due to

Table 8. Comparing Mexico’s Internet access rate with Switzerland’s using Big

Data metrics

Metric Mexico (M) Switzerland (M) t-score p-value

Impressions 18.73 4.89 -1.985 0.032

Clicks 0.04 0.00 -1.389 0.004

Average Position 2.35 4.01 3.829 0.000

Cost 0.01 0.00 -1.044 0.036




107

the advert being further down the page. The fact that the number of impressions

was higher in Portugal (M=14.37) than the United States (M=3.25) might also be

reflective on this fact – the fewer advertisers that exist in a given region then themore likely an advertisement is to be shown near the top. Portugal’s productivity

39.93 and the United States’s is 102.50, which might explain why the average posi-

tion for adverts in Portugal is higher than in the USA. The fact that Portugal has

higher impressions might reflect the fact that the websites located there are more

dependent on website income, or that because the online advertising market in

Portugal is lower then there is a greater chance of adverts being explained. Indeed,

Portugal spends only $40m on online advertising compared to the USA, which

spends $32.5bn. The fewer people that are seeking to advertise in an economy, then

the more likely adverts are to be displayed in that economy’s websites, thus pushingthe impressions up.

NEETs: Spain (Highest) v Japan (Lowest)

As there was no dedicated OECD indictor for youths not in education, employ-

ment or training (NEETs), the available one of youth unemployment was used in

its place. The term NEETs will still be used for analysis purposes. The country

with the highest number of NEETs was Spain with 51.6 per cent of youths being

unemployed, and the country with the lowest was Japan with only 6.9 per cent ofyouths being registered as unemployed. As can be seen from Table 10, in the case

of clicks there was a significant difference (p=<0.001) between Spain (M=0.11)

and Japan (M=0.03) with a t-score of 2.581, it was safe to reject the null. It was

decided to keep the null due to insignificant differences in the case of impressions

(p=<0.026), average position (p=<3.165) and cost (p=<1.256).

As can be seen from Table 10, the number of clicks for Spain (M=0.11) was

significantly higher than Japan (M=0.03). Linking this to the metric of youth un-

employment, such as those not in education, employment or training, could be

conceptually challenging, even if statistically significant. Spain has a higher number

Table 9. Comparing Portugal’s online ad expenditure with the United States’ using

Big Data metrics.

Metric Portugal (M) United States (M) t-score p-value

Impressions 14.37 3.25 -6.624 0.000

Clicks 0.02 0.06 1.879 0.077


Cost 0.02 0.00 0.002 0.002




108

of NEETs in the form of 51.6 per cent youth unemployment, compared to Japan,

where youth unemployment is 6.9 per cent. It has been argued that advertising on

social media like Facebook is a cost-effective method of recruiting youth from awide population (Chu & Snider, 2013) . However, a 2012 investigation by the Brit-

ish Broadcasting Corporation found that many clicks on Facebook are done by

people with fake accounts. It could therefore be that some of these clicks are down

to trying to raise revenue for a website on which they are displayed to raise money

for that website, and this may be common in geographical regions where there is

poverty.

Even so, the results may be more related to the fact it was advertising a crowd-

funding project. Even though it has been shown that entrepreneurial activity fell in

Spain as a result of the global recession, entrepreneurship out of necessity increased(del Rio et al., 2014) . A research study into 521 Spanish undergraduate design

students, found that they demonstrate a high entrepreneurial intention (62%) and

that attitudinal factors outweighed the students’ self-perceived inability to develop

their own businesses (Ubierna et al., 2014) . It could therefore be the case that

young people are more likely to take risks with crowd funded projects where they

anticipate future returns.

Room in House, Education: Mexico (Lowest)

v United States (Highest)

In terms of rooms in house, Mexico was the lowest (M=3.88) and the United States

was the highest (M=4.65). The same was the case with education outcomes, with

the United States being highest (M=4.70) and Mexico being the lowest (M=3.63).

There was a significant difference with regards to impressions (p=<0.001) and

average position (p=<0.001). In the case of impressions it would be suitable to

reject the null as there was a t-score of -9.887 and Means of 18.73 for Mexico and

3.25 for the United States (see Table 11). In the case of average position, the sig-

nificant difference could be seen between Mexico (M=2.35) and the United States

Table 10. Comparing Spain’s percentage of NEETs with Japan’s using Big Data

metrics.

Metric Spain (M) Japan (M) t-score p-value

Impressions 12.38 6.67 1.737 0.025

Clicks 0.11 0.03 2.581 0.000


Cost 0.01 0.00 1.255 1.255




109

(M=4.41) with a t-score of 6.003, meaning it was safe to reject the null. There was

no significant difference for clicks (p=<0.210) in relation to Mexico (M=0.04) and

the United States (M=0.06), and nor was there a significant difference in relationto costs (p=<0.015) for the United States (M=0.00) and Mexico (M=0.01), which

an insignificant t-score of -1.269.

It is known that in Mexico, the number of rooms in a house is indicative of social

and economic status more widely (López-Feldman, 2014; Mora-Ruiz et al., 2014).

In Mexico, the size of the house and the number of rooms mould the number and

type of activities people can do, and the relationship with those outside of the house

is also connected with positive life experiences (Landázuri et al., 2014) . It has been

found that the highest numbers rooms equated with areas with the lowest productiv-

ity and highest levels of trolling (Bishop, 2014a). Even though not a significantdifference, the mean number of clicks were higher in the United States even though

the number of impressions were nearly six times greater in Mexico. The t-score for

impressions (i.e. -9.887) by being a negative value would suggest instead high

number of impressions is equated with low levels of education and room numbers.

This, taken with the number of clicks, would suggest that those in Mexico are less

prone to ‘feed the troll’ in the form of advertisers, which may be down to not being

able to afford the product advertised. However, as QPress is marketed at academics,

it might simply be that low education outcomes (3.65 out of 5) would lead to little

interest in QPress’s functions. The reason the average position is higher in theUnited States may be because there is more completion for advertising space in the

US, or simply that QPress is advertising for the same areas as other products and

services.

Intelligence: Mexico (Lowest) v Japan (Highest)

Even though the numbers of impressions for Mexico (M=18.73) are higher than

Japan (M=6.67), because the t-score is negative (t=-3.163) it suggests that a

higher number of impressions is linked to lower intelligence (i.e. literacy) and not

Table 11. Comparing Mexico’s rooms in house and education outcomes with the

United States’ using Big Data metrics

Metric Mexico (M) United States (M) t-score p-value

Impressions 18.73 3.25 -9.887 0.000

Clicks 0.04 0.06 0.630 0.209


Cost 0.01 0.00 -1.269 0.014




110

higher intelligence. One might expect the opposite – for searches that would call up

an advert on QPress to be linked with more academic audiences. One might there-

fore want to consider whether it was the crowd-funding aspect that draw people to

the advert. Mexico is associated with micro-financing more as a recipient than asan investor (Smith et al., 2014), but the relative poverty in the region might mean

necessity entrepreneurship would be higher. Japan is associated with several high

profile crowd-funding initiatives, and micro-financing is an accepted practice

(Ikeda & Marsumaru, 2012) .

Mexico’s productivity ($23.68 per capita) is lower than Japan’s ($93.23 per

capita), which might explain why the impressions for Mexico are nearly triple Japan’s.

Mexico needs the funding due to its low geo-demographic figures and it might be

websites are being set up for the sole purpose of extorting funds from advertisers

who have not optimised their advertising campaigns. The fact that a cost of an advertin Mexico is significantly lower than Japan, which has higher intelligence, might

suggest it is not the most optimal market for targeting crowd-funded schemes, which

want to raise funds and not expend them.

GENERALISING THE DATA

This chapter has so far explained the differences between countries in terms of the

geo-demographic data associated with them. This section aims to put this togetherto show how in the advertising for QPress findings can be used to further its devel-

opment, which might be generalised to other contexts.

Figure 2 shows the factors that are most associated with whether impressions,

clicks, average position, and cost are high or low. A low number of impressions

can be seen to be associated with countries with high education attainment and

lowest intelligence, which may appear contradictory. Conversely, a high number

of impressions were associated with low numbers of rooms in housing and low ad

expenditure for a country. This could be because low ad expenditure means that

Table 12. Comparing Mexico’s Intelligence levels with Japan’s using Big Data metrics.

Metric Mexico (M) Japan (M) t-score p-value

Impressions 18.73 6.67 -3.163 0.001

Clicks 0.04 0.03 -0.292 0.560

Average Position 2.35 2.32 -0.183 0.066

Cost 0.01 0.00 -1.760 0.000




111

one’s online ad is likely to appear higher up the page, or more frequently, becauseless people are advertising.

In terms of clicks, a high number of clicks is associated with a high number of

young people classed as NEETs. It is known that online adverts are more likely to

be clicked on by those who develop, or have already developed, strong brand aware-

ness (Dahlen & Bergendahl, 2001) . It is a fact that online ads that offer high in-

volvement products as opposed to low involvement are also more likely to be clicked

on (Dahlén et al., 2000) . The advert for QPress shown in Figure 1 by referring to

“Invest in QPress” clearly suggests an amount of involvement will come from the

activity. There is, however, known to be an amount of cynicism about the actualengagement of young people when it comes to online advertising, which is seen as

representing the commercialisation of the medium (Loader, 2007). Other reasons

may be that because young people are out of work, they are more likely to be inter-

ested in opportunities to make money, such as from investing in QPress as the online

advert displayed encouraged. The fact that few other factors were significantly as-

sociated with clicks might suggest that advertising schemes based on clicks are least

likely to be effective in achieving a return on investment. With some advertising

Figure 2. Geo-demographic factors associated with online advertising metrics.




112

initiatives being based on pay-per-interaction, such as Pages or Posts that have been

‘boosted’ on Facebook, or those banners available on Tradedoubler requiring prod-

ucts to be purchased before fees are paid to the publisher, then the pay-per-click and

pay-per-impression models will likely become obsolete, as such methods are opento abuse, where they may be viewed or clicked on by people who have no interest

whatsoever in what is being advertised.

In terms of average position, a place lower down the screen was associated with

high education and low productivity, whereas being higher up the page in terms

of average position is associated with countries with a high number of rooms per

house, high Internet access and where low ad expenditure is the norm. Average

position of an advert is an important issue in having an effective online advertising

campaign (Rutz & Bucklin, 2011). Knowing the position of a given keyword can

help one know whether to focus on it (Rutz et al., 2012) . For instance, a low averageposition could mean one has a lot of competitors, whereas a high average position

could mean that one might not be marketing to the right people.

In terms of cost, the high the cost then the lower that country’s average intelli-

gence. It is known that those who randomly surf the Internet are more likely to click

on banner adverts than those who are being more purposeful, such as information-

seekers (Li & Bukovac, 1999) . On this basis it is likely that the view that those

who do generic searches have little brand awareness (Rutz & Bucklin, 2011) might

be true of those from more deprived areas, who might not have awareness of what

the product being advertised actually is.

IMPLICATIONS AND FUTURE RESEARCH DIRECTIONS

This study has shown that it is possible to combine national data on countries with

individual observations in order to understand social factors, such as Internet ac-

cess, education and intelligence. One might argue that in case of geographical fac-

tors, there are a number of interacting variables in the dataset used. This could be

generalised to conclude that in order for Big Data studies to be effective that theyneed to amalgamate with other data sources, including linked data, as an essential

component of providing an overall picture of the social phenomenon being studied.

The study has presented a model for understanding the links between online advert

metric and geo-demographic data. Future research will have to uncover whether this

relates online to the advertisement associated with QPress (Figure 1), or whether it

can be applied more generally to other crowd-funding projects.




113

DISCUSSION

Big Data is the use of datasets that are too big for traditional business analysis tools

used in industry. This chapter has explored the role Big Data can play in analysingthe effectiveness of crowd-funding projects, using the data from such a project,

which aimed to fund the development of a software plug-in called ‘QPress.’ Using

an ANOVA, the website metrics of impressions, clicks and average position, were

found by this chapter to be significantly connected with geographical factors. To

understand these further, they were combined with other country data to perform

t-tests in order to form a geo-demographic understanding of those who are displayed

advertisements inviting participation in crowd-funding.

The question as to whether it was possible to replicate the findings of an earlier

study now has an answer. It appears clear that the factors such as productivity,number of rooms in house, education, intelligence and NEETs, are indeed impor-

tant to understanding what affects the extent to which an online communication is

provocative. The study looked not at regions as with was the case with the earlier

study, but countries, and it was still found that these facts play and important part

in understanding the geo-demographics that exist and are measurable in relation to

human behaviour.

The chapter has presented a model linking the geo-demographic factors identi-

fied with the online advertising matrix. It shows that a high number of impressions

is associated with low rooms per house and low ad expenditure. It also shows thathigh clicks are associated with high numbers of young people not in education,

employment or training (NEETs), and that a high cost campaign is associated with

lower intelligence. It found that the average position of an online advert is low ad-

vert expenditure, high rooms per house and high Internet access, while low average

position is associated with high education outcomes and low productivity.

It can there be concluded that an online advertising campaign, like the one for

the crowd-funded application called QPress, has its success dependent on geo-

demographic factors. It is clear that when considering how to advertise to a locality,

these will have to be taken into account. When advertising to countries with lowproductivity and education outcomes, an advertising method based on cost-per-

impression or cost-per-click would not be the most effective option, due to it being

more likely they will be received by websites that’s are clicked on by people with

little interest in what is being advertised. Methods based on cost-per-interaction,

such as where an ad is only paid for when a person buys from a website, leaves a

comment or likes a post, would be most suitable to these economies. Cost-per-click

models would be most suited to geographies where productivity is high, along with

education outcomes and intelligence being the same. This may be because people in

these economies are least likely to click on adverts, meaning a cost-per-impression




114

model would have least return on investment. Cost-per-impression models seem to

be most suited to developed economies, such as Japan, Canada and the USA for the

reason that the number of impressions is low, but number of clicks is high. These

economies have the highest spending on online advertising this results in fewerimpressions and greater competition for a good average position. It is therefore

necessary for advertisers to know their target market as bidding high on a keyword

most will search on to get higher up the page may pay off more than bidding on

many keywords of little relevance.

REFERENCES

Arthur, C., Ball, S., Bray, P., Grossman, W. M., Queree, A., & Rowe, J. (2001). Thehutchinson dictionary of computing and the internet (4th ed.). Oxford, GB: Helicon

Publishing Ltd.

Bishop, J. (2011). Assisting human interaction (AU/GB2011/2011266844 ed.). AU:

PCT/GB2011/050814.

Bishop, J. (2012). Lessons from the emotivate project for increasing take-up of big

society and responsible capitalism initiatives. In P. M. Pumilia-Gnarini, E. Favaron,

E.

Pacetti, J. Bishop, & L. Guerra (Eds.), Didactic strategies and technologies for

education: Incorporating advancements (pp. 208–217). Hershey, PA: IGI Global.

doi:10.4018/978-1-4666-2122-0.ch019

Bishop, J. (2013a). The art of trolling law enforcement: A review and model for

implementing ‘flame trolling’legislation enacted in great britain (1981–2012).

International Review of Law Computers & Technology, 27 (3), 301–318. doi:10.10

80/13600869.2013.796706

Bishop, J. (2013b). The effect of deindividuation of the internet troller on criminal

procedure implementation: An interview with a hater. International Journal of

Cyber Criminology, 7 (1), 28–48.

Bishop, J. (2014a). Digital teens and the ‘antisocial Network’: Prevalence of trouble-

some online youth groups and internet trolling in great Britain. International Journal

of E-Politics, 5(3), 1–15. doi:10.4018/ijep.2014070101

Bishop, J. (2014b). Internet trolling and the 2011 UK riots: The need for a dualist

reform of the constitutional, administrative and security frameworks in Great

Britain. European Journal of Law Reform, 16(1), 154–167.



115


Bishop, J. (2014c). Representations of ‘trolls’ in mass media communication: A

review of media-texts and moral panics relating to ‘internet trolling’.

International Journal of Web Based Communities, 10(1), 7–24. doi:10.1504/

IJWBC.2014.058384

Bojke, C., Castelli, A., Street, A., Ward, P., & Laudicella, M. (2013). Regional varia-

tion in the productivity of the english national health service. Health Economics,

22(2), 194–211. doi:10.1002/hec.2794 PMID:22334404

Bonacchi, C., Bevan, A., Pett, D., & Keinan-Schoonbaert, A. (2015). Developing

crowd-and community-fuelled archaeological research: Early results from the Mi-

croPasts project.

Brechner, I. (2013). The value of long-term client and search agency relationships

in a digital world. Journal of Digital & Social Media Marketing, 1(3), 258–264.

Chaney, E., & Hornbeck, R. (2012). Economic Growth in the Malthusian Era:

Evidence from the 1609 Spanish Expulsion of the Moriscos, Chu, J. L., & Snider,

C. E. (2013). Use of a social networking web site for recruiting canadian youth

for medical research. The Journal of Adolescent Health, 52(6), 792–794.

Cowpertwait, J., & Flynn, S. (2002). The internet from A to Z. Cambridge, GB:

Icon Books Ltd.

Crumlish, C. (1995). The internet dictionary: The essential guide to netspeak .Alameda, CA: Sybex Inc.

Cukier, K., & Mayer-Schoenberger, V. (2013). Rise of big data: How it’s

changing the way we think about the world, the. Foreign Affairs, 92, 28.

Dahlen, M., & Bergendahl, J. (2001). Informing and transforming on the web: An

empirical study of response to banner ads for functional and expressive products.

International Journal of Advertising, 20(2), 189–205.

Dahlén, M., Ekborn, Y., & Mörner, N. (2000). To click or not to click: Anempirical study of response to banner ads for high and low involvement products.

Consumption Markets & Culture, 4(1), 57–76.

doi:10.1080/10253866.2000.9670349

de-la-Peña-Sordo, J., Santos, I., Pastor-López, I., & Bringas, P. G. (2013).

Filtering trolling comments through collective classification. Network and

system security

(pp. 707–713). Springer.




116

del Rio, Maria de la Cruz, Garcia, J. A., & Rueda-Armengot, C. (2014). Evolution

of the socio-economic profile of the entrepreneur in galicia (spain). Business and

Management Research, 3(1), 61.

Dushnitsky, G., & Marom, D. (2013). Crowd monogamy. Business Strategy Review,

24(4), 24–26. doi:10.1111/j.1467-8616.2013.00990.x

Etzkowitz, H. (2013). Silicon valley at risk? sustainability of a global innovation

icon: An introduction to the special issue. Social Sciences Information. Information

Sur les Sciences Sociales, 52(4), 515–538. doi:10.1177/0539018413501946

Feldmann, N., Gimpel, H., Kohler, M., & Weinhardt, C. (2013). Using crowd funding

for idea assessment inside organizations: Lessons learned from a market engineering

perspective. Paper presented at the Cloud and Green Computing (CGC), 2013 Third

International Conference On, pp. 525-530. doi:10.1109/CGC.2013.88

Ferrer-Roca, N. (2014). 2. business innovation in the film industry value chain:

A new zealand case study. International Perspectives on Business Innovation and

Disruption in the Creative Industries: Film, Video and Photography,, 18.

Graham, M., & Shelton, T. (2013). Geography and the future of big data, big data

and the future of geography. Dialogues in Human Geography, 3(3), 255–261.

doi:10.1177/2043820613513121

Hardaker, C. (2010). Trolling in asynchronous computer-mediated communication:From user discussions to academic definitions. Journal of Politeness Research.

Language, Behaviour . Culture (Canadian Ethnology Society), 6 (2), 215–242.

Hardaker, C. (2013). “Obvious trolls will just get you banned”: Trolling versus

corpus linguistics. In A. Hardie & R. Love (Eds.), Corpus linguistics 2013 (pp.

112–114). Lancaster: UCREL.

Harvey, D. (2011).The enigma of capital: And the crises of capitalism. Profile Books.

Herodotou, H., Lim, H., Luo, G., Borisov, N., Dong, L., Cetin, F. B., & Babu, S.

(2011). Starfish: A self-tuning system for big data analytics. Paper presented at the

CIDR, 11. pp. 261-272.

Huang, L. (2013). Visual analysis on online display advertising data. Paper presented

at the Large-Scale Data Analysis and Visualization (LDAV), 2013 IEEE Symposium

On, pp. 123-124. doi:10.1109/LDAV.2013.6675170

Hurwitz, J., Nugent, A., Halper, F., & Kaufman, M. (2013). Big data for dummies.

For Dummies.




117

Ikeda, M. M., & Marsumaru, M. (2012). Leadership and social innovation initia-

tives at the grassroots during crises. Japan Social Innovation Journal, 2(1), 77–81.

doi:10.12668/jsij.2.77

Jansen, E. (Ed.). (2002). NetLingo: The internet dictionary. Ojai, CA: Netlingo Inc.

Jansen, E., & James, V. (Eds.). (1995). NetLingo: The internet dictionary. Ojai,

CA: Netlingo Inc.

Kaisler, S., Armour, F., & Espinosa, J. A. (2014). Introduction to big data: Chal-

lenges, opportunities, and realities minitrack. Paper presented at the System Sciences

(HICSS), 2014 47th Hawaii International Conference On, pp. 728-728.

Katircioğlu, S. T. (2014). Estimating higher education induced energy consumption:

The case of northern cyprus. Energy,66 , 831–838. doi:10.1016/j.energy.2013.12.040

Kshirsagar, V., & Ahuja, R. S. (2015). Crowd-funding and its perspectives. Indian

Journal of Applied Research, 5(1).

Kunkel, S. (2015). Green crowdfunding: A future-proof tool to reach scale and

deep renovation? World sustainable energy days next 2014 (pp. 79–85). Springer.

Labrinidis, A., & Jagadish, H. (2012). Challenges and opportunities with big data. Pro-

ceedings of the VLDB Endowment ,5(12), 2032–2033. doi:10.14778/2367502.2367572

Landázuri, A. M., Mercado, S., & Terán, A. (2014). Sustainability of residentialenvironments. Revista Suma Psicológica, 20(2), 191–202. doi:10.14349/sumap-

si2013.1463

Laven, M. (2014). Money evolution: How the shift from analogue to digital is trans-

forming financial services. Journal of Payments Strategy & Systems, 7 (4), 319–328.

Li, H., & Bukovac, J. L. (1999). Cognitive impact of banner ad characteristics: An

experimental study. Journalism & Mass Communication Quarterly,76 (2), 341–353.

doi:10.1177/107769909907600211

Loader, B. D. (2007). Young citizens in the digital age: Political engagement, young

people and new media. Routledge.

López-Feldman, A. (2014). Shocks, income and wealth: Do they affect the extrac-

tion of natural resources by rural households? World Development, Malgonde, O.,

& Bhattacherjee, A. (2014). Innovating using big data: A social capital perspective.

Paper presented at the Twentieth Americas Conference on Information Systems,

Savannah, GA.




118

McCosker, A. (2013). Trolling as provocation: YouTube’s agonistic publics. Con-

vergence (London), , 1354856513501413.

Mora-Ruiz, M., Penilla, R. P., Ordonez, J. G., Lopez, A. D., Solis, F., Torres-Estrada,

J. L., Rodriguez, A.D. (2014). Socioeconomic factors, attitudes and practices as-

sociated with malaria prevention in the coastal plain of chiapas, mexico. Malaria

Journal, 13(1), 157-2875-13-157. doi:10.1186/1475-2875-13-157

Norcliffe, G., & Mitchell, P. (1977). Structural effects and provincial productivity

variations in canadian manufacturing industry.The Canadian Journal of Economics.

Revue Canadienne d’Economique, 10(4), 695–701. doi:10.2307/134300

Odesola, I. A., & Idoro, G. I. (2014). Influence of labour-related factors on construc-

tion labour productivity in the south-south geo-political zone of nigeria. Journal of

Construction in Developing Countries, 19(1), 93–109.

Olwig, K. F., & Valentin, K. (2014). Mobility, education and life trajectories: New

and old migratory pathways. Identities, (ahead-of-print), 1-11.

Ordanini, A., Miceli, L., Pizzetti, M., & Parasuraman, A. (2011). Crowd-funding:

Transforming customers into investors through innovative service platforms. Journal

of Service Management , 22(4), 443–470. doi:10.1108/09564231111155079

Pfaffenberger, B. (1995). Que’s computer & internet dictionary (6th ed.). India-

napolis, IN: Que Corporation.

Phillips, W. (2012). This is why we can’t have nice things: The origins, evolution

and cultural embeddedness of online trolling. Unpublished Doctor of Philosophy.

Oregon, USA: University of Oregon.

Rutz, O. J., & Bucklin, R. E. (2011). From generic to branded: A model of spillover

in paid search advertising. JMR, Journal of Marketing Research, 48 (1), 87–102.

doi:10.1509/jmkr.48.1.87

Rutz, O. J., Bucklin, R. E., & Sonnier, G. P. (2012). A latent instrumental variables

approach to modeling keyword conversion in paid search advertising. JMR, Journal

of Marketing Research, 49(3), 306–319. doi:10.1509/jmr.10.0354

Sagiroglu, S., & Sinanc, D. (2013). Big data: A review. Paper presented at the Col-

laboration Technologies and Systems (CTS), 2013 International Conference On,

pp. 42-47.

Shiller, R. J. (2013). Capitalism and financial innovation.Financial Analysts Journal,

69(1). doi:10.2469/faj.v69.n1.4




119

Shim, K. (2012). MapReduce algorithms for big data analysis. Proceedings of the

VLDB Endowment , 5(12), 2016–2017. doi:10.14778/2367502.2367563

Singh, S. K. (2011). Organizational innovation as competitive advantage during

global recession. Indian Journal of Industrial Relations, 46 (4), 713–725.

Smith, R., Eaton, A., Arshad, H., & Ricketts, P. (2014). Global crowd funding to

increase accessibility for small-scale biodigester projects. Boiling Point, (62), 2-5.

Snijders, C., Matzat, U., & Reips, U. (2012). Big data: Big gaps of knowledge in

the field of internet science. International Journal of Internet Science, 7 (1), 1–5.

Stern, J. S. (2013). Characteristics of content and social spread strategy on the

indiegogo crowdfunding platform. (Master of Arts, University of Texas at Austin).

Suhonen, T. (2014). Field-of-study choice in higher education: Does distance mat-

ter?Spatial Economic Analysis,9(4), 355–375. doi:10.1080/17421772.2014.961533

Ubierna, F., Arranz, N., & Fdez de Arroyabe, J. (2014). Entrepreneurial intentions

of university students: A study of design undergraduates in spain. Industry and

Higher Education, 28 (1), 51–60. doi:10.5367/ihe.2014.0191

Walter, T. (2014). New mourners, old mourners: Online memorial culture as a

chapter in the history of mourning. New Review of Hypermedia and Multimedia,

(ahead-of-print), 1-15.

Wilson, M. W. (2014). Morgan freeman is dead and other big data stories. Cultural

Geographies, 1474474014525055.

Wilson‐Ching, M., Pascoe, L., Doyle, L. W., & Anderson, P. J. (2014). Effects of

correcting for prematurity on cognitive test scores in childhood. Journal of Paedi-

atrics and Child Health, 50(3), 182–188. doi:10.1111/jpc.12475 PMID:24617343

Woerner, S. L., & Wixom, B. H. (2015). Big data: Extending the business strategy

toolbox. Journal of Information Technology, 30(1), 60–62. doi:10.1057/jit.2014.31

Yuan, Y., Wang, F., Li, J., & Qin, R. (2014). A survey on real time bidding advertising.

Paper presented at the Service Operations and Logistics, and Informatics (SOLI), 2014

IEEE International Conference On, pp. 418-423. doi:10.1109/SOLI.2014.6960761

Zhang, J., Li, J., Jiang, Y., & Zhao, B. (2014). The overview of the information

technology industry chain in big data era. Future information technology (pp.

429–432). Springer.




120

Zhu, Q., Kong, X., Hong, S., Li, J., He, Z., & Lewandowski, D. (2015). Global

ontology research progress: A bibliometric analysis. Aslib Journal of Information

Management, 67 (1).

KEY TERMS AND DEFINITIONS

Appalachia: A geographic and cultural region of the Mideastern United States.

The population in media is portrayed as suspicious, backward, and isolated.

Big Data: Big Data is the term used to describe information which is of a form

that it is difficult to analyse using traditional business software or where a data col-

lected for one purpose is then used for another to improve a business’s offerings.

Clicks: The number of times an advert has been clicked on by a user on a given

website(s).

Average Position: The extent to which an advert is offset from the top of the

advert stream where it is competing for the top spot with other advertisers, usually

based on a bidding price.

Google Ads: A platform that enables business to advertise on Google websites

and those websites that Google has agreed for its adverts to be displayed on.

Impressions: The term that refers to how many time an advert has been displayed

on a given website(s).

OECD: An international economic body whose data is used to understand the

nations it covers.

QPress: A software application that allows for the collection and exporting of

Q-methodology q-sorts via the Internet.

the role of geo-demographic big data for assessing the effectiveness of crowd-funded software...

Documents