USAGE OF SOCIAL NETWORKS IN ORDER TO BOOST AND IMPROVE SAFETY AROUND THE YOUNGER WORKERS



Authors: Ph.D.Stefan Kovacs INCDPM ”Alexandru Darabont”


ABSTRACT
Social networks on Internet are at this time not just a mode but a way to cluster youngsters from the entire world around various subjects of discussion. One of these subjects could be staff competency in the assurance of better conditions at workplace and better work, safety-oriented improvement of the quality of human factors and so on. Moreover, in cases of emergency, messages sent through the social networks could reach more easily the receiver and could disseminate more quickly the desired information.
The paper presents some results from a research started in 2008 and dedicated to the safety utility of using social networks. It shows our opening models regarding the two way research- a free social network connection regarding safety problems and a scenario oriented experiment together with the results obtained using 2 main Romanian net forums, Face book and Twitter. We found in our research that social networks could be a good support to improve the safety conscience, to develop safety capabilities and attitudes and to quickly disseminate and communicate aspects that could lead to an imminent risk. Social networks could be easy to use- on handheld dispositive and could attract attention more than normal communication ways.

KNOWLEDGE, COMMON INTELLIGENCE, SAFETY AND SOCIAL NETWORKS-THE BASIS OF OUR RESEARCH
In the emerging economy, knowledge is the primary resource for individuals and for the economy overall; land, labour, and capital. Improving front-line worker productivity together with its confort at work and implicit safety is the greatest challenge of the 21st century. [1].
Knowledge is the main engine of economic development.But what is knowledge ?
Davenport and Prusak [2] view knowledge as a fluid mix of framed experience, values, contextual information, and expert insight that provides a framework for evaluating and incorporating new experiences and information. It originates in
the minds of knowers. In organizations, it often becomes embedded not only in documents or repositories but also in organizational routines, processes, practices, and norms. Although there is no common definition of the
term knowledge, there is a wide agreement that knowledge is social in nature. Many researchers emphasise the social, collective and distributed aspect of knowledge.
Knowledge could and would be useful just in a community, not alone. Wenger [3]
points out that knowledge does not exist either in a world of its own or in individual minds but is an aspect of participation in cultural practices.
He uses the term participation to describe the social experience of living in the world in terms of membership in social communities and active involvement in social enterprises. Participation in this sense is both personal and social. It is a complex process that combines doing, talking, thinking, feeling, and belonging. Siemens [4] defines a community as the clustering of similar areas of interest that allows for interaction, sharing, dialoguing, and thinking together. The accent is put on the term „areas of interest” which imply the sharing process, even if this sharing refers to thinking. Wenger [5]  points out that
community does not imply necessarily co- presence, a well-defined, identifiable group or socially visible boundaries. It does imply participation in an activity system about which participants share understanding concerning what they are doing and what that means in their lives and for their communities. The concept of community is very close to the concept of social network. A network is an important source of labour for the formation of a collective subject,where there is no common opinion allready in place (for example, emerging risks). It is possible to define a social network as a complex, dynamic system in which, at any given time, various knowledge exists or could be developed  in different instantiations. The heterogenous nature of a social network gives a plus value because various kinds of experience could be mixed in order to obtain a certain knowledge. For example, beyond engineers and risk specialists, in the safety knowledge domain a network could mobilize doctors, psychologues and also various other specialities which could elaborate a more complex and well sustained point of view. The network is the power of many regarding the development of new knowledge. As a special type of community inside a network , Wenger introduces the concept of communities of practice (CoP). Wenger defines CoP as groups of people who share a concern or a passion for something they do and learn how to do it better as they interact regularly. According to Wenger, a CoP is characterised by:
(a) The domain; a CoP has an identity defined by a shared domain of interest. Membership therefore implies a commitment to the domain, and therefore
a shared competence that distinguishes members from other people.
(b) The community; in pursuing their interest in their domain, members engage in
joint activities and discussions, help each other, and share information. They build relationships that enable them to learn from each other. A website in itself is not a community of practice.
(c) The practice; members of a community of practice are practitioners. They develop a shared repertoire of resources: experiences, stories, tools,
ways of addressing recurring problems, in short a shared practice. This takes time and sustained interaction. To differentiate between CoP and network, it could be said that CoP or a „cloud of friends” as these structures are nominated in this paper are included in the social network . Zager [6] explores a collaboration configuration called a coalition and notes that coalitions are temporary collaborative groups where shared concerns and interests connectconstituent individuals and teams.
This coalition leads inevitably to the idea of collective intelligence. Jenkins et al. [7] define collective intelligence as the ability to pool knowledge and compare notes with others towards a common goal. Levy [8] sees collective intelligence as an important source of power in knowledge communities to confront problems of greater scale and complexity than any single person might be able to handle. He argues that everyone knows something, nobody knows everything, and what any one person knows can be tapped by the group as a whole.




THE RESEARCH

Taking into account the above mentioned aspects, our research, started in 2008, has as a main goal the response to the following question „Could social networks like facebook, Twitter, hi5 and so on be used in order to promote and improve safety among young workers ?”. In Romania statistics show that young workers are the most accident prone group. They are also the most significant participants in the social networks- so perhaps there is a possibility to make them conscient of the risks at which they are exposed, to imprint some basic safety rules through the social networks and also to make them interact or act directly in the way to mantain and improve safety.
First we have developed a model, presented in the following  paragraph. Then, on the basis of this model we have developed two experiments:
-a „static experiment” based on facebook in which young workers were informed about an accident occured some days earlier, accident in which five young born died because of an electrical instalation fault. Our experiment subjects were asked to interact with the facebook page „Safety now!” in order to express their opinions through the message system, search for specific knowledge using the links given on page and also participate at various pro-bono safety trainings performed at the Romanian National Occupational Safety Research Institute (INCDPM)
-a „dinamic experiment” based on Twitter in which the young workers were exposed to a simulated event (an accidental crude oil spill) and were asked to help and to contribute as much as possible to the stoppage of the spill.
We have used two significant statistical lots, one of 200 persons for the first experiment and one of 500 persons for the second experiment.
The methods used for our research were of two types:
A.the first category (methods used for network data collection) aims to provide a data set that helps study the effects of the interaction with a safety subject on different aspects of social activities.
To achieve this aim, the following methods are used:
-Socio-centric: to examine sets of relationships between actors that are regarded for analytical purposes as bounded social collectives.
-Ego-centric: to select focal actors (egos), and identify the nodes they are connected to.
B.The second category (methods used for network data visualisation and analysis ) aims to render data in easily understood formats, thus making complex information usable and interpretable.
To achieve this aim, the following are used:
- graphs: to visualise relationships among members of  the social networks used in the experiment
- matrices: to visualise large or dense „clouds of friends”
- maps: to manage a wide amount of data and information;



THE MODEL
In order to perform the experiments we have developed a model of interaction and learning in social networks, model that has as a basis [9] and other research papers that will be also mentioned.
Our model starts with a  canonical model of social learning which comprises a set of agents I , a finite set of actions A, a set of states of  context Ω, and a common payoff function U(a,ω), where a is the action chosen and ω is the state of context. Each agent i receives a private signal σi(ω), a function of the state of  context  ω, and uses this private information to identify
a payoff-maximizing action.The set of agents is not homogeneous and  could have the following strategies at the social network interaction:
-do nothing;
-interact weakly (access the subject but nothing more)
-interact strongly (access the subject and participate at the interaction with its own contributions);
-act;  the agent has the freedom of decision- in the same time he must maximize its payoff or , at last, not be the looser.
This set of agents is comprised in the „cloud of friends” developed inside a  social network  like facebook or Twitter.
For each such a network  some results could be ennounced, like:
Uniformity of behavior. Initially, diversity of private information leads to diversity
of actions.  Over time, as agents learn by observing the actions of their neighbors,
some convergence of beliefs is inevitable. A central question is whether agents can
rationally choose different actions forever.We postulate that each agent has its own power  of decision and would choose the best strategy that fits him.For example, if he is an expert in oil spills he could offer his help- acting; if not he could choose to express himself on a strong interaction – like joining the call to boycot the institution seen as guilty of the spill. Here, this uniformity takes the form of herd behavior.Smith and Sørensen [10] have shown that, in the SSLM, herd behavior arises in finite time with probability one. Once the proportion of agents choosing a particular action is large enough, the public information in favor of this action outweighs the private information of any single agent.So each subsequent agent “ignores” his own signal and “follows the herd.”
Optimality. It is interesting whether the common action chosen asymptotically is
optimal, in the sense that the same action would be chosen if all the signals were public
information. In special cases, it could be shown show that asymptotically the optimal action is
chosen but, in general, there is no reason why this should be the case.
There are some specific social network principles, the most interesting for our model being the following:
-the welfare improvement principle. Agents have perfect recall, so expected utility is non-decreasing over time. Welfare- as seen fit by the specfic  social network- is continously improving. If a subject- for example safety- is considered of interest by the social network- its agents are promoting inside the network (here we speak about an informal promotion) by developing the subject.
- the imitation principle. If agent i can observe the actions of agent j , the simplest  strategy available to him is to imitate whatever j does.  Since i and j have different information sets, their conditional payoffs under this strategy may be different. However, on average, i must do as well as j . If safety is considered „cool” by a critical mass of agents, it would be cool for every other agent which wants to get his position in the social network. 
A social network is represented by a family of sets [11]  {Ni : i = 1, . . . ,n}, where
Ni = {1, . . . , i 1, i + 1, . . . ,n}.For each agent i , Ni denotes the set of agents j _= i who can be observed by agent i. We can think of Ni as representing the  “cloud of friends .”. The social network determines the information flow in the proximity of the „cloud of friends” given a specific context and also the existence of self- control objectives inside the network . Agent i can observe the action of agent j if and only if j is included in  Ni . Agents have perfect recall so their information set at each date includes the actions they have observed at every previous date.
Utility and Usability.
If the social network has no immediate utility, it could imply:
-a status utility: agent m  rises his status among its friends included in the „cloud of friends” by belonging to the social network. In this respect, the competent authorities (like HSE or the Romanian Labour Inspection) could contribute by revarding the agent which is involved in a weak interaction, a strong interaction or takes an action by himself (like announcing the authorities at the signs of an imminent danger) with honorific labels or badges which could increase the social status among the network members.
-an emergent learning process- agent m  lis prepared to learn about something that could be offered by the network.
If a social network agent could not use the information he gets, then the social network is also useless. So, the social network could:
-inform  on the current level of understanding about a subject;
-imprint some subjects into the network present and future topics;
-facilitate a learning process;
-give some”user” feedback regarding possible hazardous manifestations at the workplace or in the vicinity.
The figure 1 gives a global image of the model.

Figure 1 The global image of the model

THE FACEBOOK EXPERIMENT
Description of the experiment
The facebook experiment regarding safety was centred on the idea that this social network is the perfect way to disseminate safety around not homogenous groups of youngsters which are, or could be interested, in safety problems.
In our experiment, facebook was conceived as a disseminator for various safety events and happenings, taking into account that the facts- presented partly in the Romanian mass-media- were not safety related.
We have found a statistical significant lot of 200”friends”, young workers in the main functional enterprises in Bucharest and also in the government agencies.
The statistical significant lot was chosen taking into account their common facebook affiliation, the non-homogeneity of the group (otherwise the experiment could be biased by the common organisational and safety culture); their apparent un-interest in traditional safety classes and also their desire to act for status significant causes like environment protection.
The facebook account”Safety now” was built specially for this experiment. Using the above presented model and given a subject of interest- a recent occupational accident in which five infant babies died because their Intensive Care unit took fire because the lack of surveillance and safety control-we have tried to establish how many from the statistical significant lot would interact weakly, how many would interact strong and how many would act effectively- the action proposed being a request for specialised electricians to check their electrical installations, especially the air conditioning ones.
The experiment was done between mid of 2008 and mid of 2010, totally 24 months.
The figure 2 gives the schema of the experiment. 


Figure 2 Schema of the facebook experiment





Figure 3 Links on the facebook page




Figure 4 Closed spaces course invitation on the facebook page

RESULTS

The table 1 shows the results of the experiment

Table 1- Results of facebook experiment
Attribute taken into account
Value
Statistical significant lot
200 workers, 80 male, 120 female, average age:25 years, average education: college, all having facebook accounts
Participants at events developed for the experiment
150
Participants at pro-bono safety training at class
80 (the training was focused on the confined spaces subject and was about 2 hours long)
Interacting weakly (by accessing links)
120
Interacting strongly (by sending messages of support, compassion or with various solutions to the problem that led to the accident)
170 (initially we intended to use a tracer to differentiate every participant. Taking into account facebook characteristics this was not possible- so there is a probability that the weak interactions also interact strongly)
Accessing the safety forums
40

Conclusions of the facebook experiment
The facebook experiment shows that participants at a social network like facebook, with a not homogenous group are interested in safety problems having a trigger event (the fire in which the five babies died). Counters were used in order to number the participants at various interaction forms. Also, a password allowed working just with the statistical significant lot and not with strangers.
We found an interested”cloud of friends” which acted as a group without a visible reward. The apartenence at this safety cantered group was enough for them.
Also, we found that facebook could be a very valuable link to safety knowledge on-line, if this knowledge is essentially needed and is available in a friendly format. Accessing the Safety now! Facebook profile, those in need could access easily the desired knowledge. We have computed an involvement coefficient (IC) as follows:
Ic=Number of persons that interact/Total number of the statistical significant lot (1)
As we could not differentiate among the participants we choose the larger number so
Ic= 170/200= 0.85. So, 85% of the participants were interacting somehow (we consider it strong interaction as we received 170 different messages from the”cloud of friends”).In our simulation of the above mentioned model we have estimated an Ic of 25…35% so the results were well above our expectations.

THE TWITTER EXPERIMENT
Description of the experiment
In this event we simulated an undesired event, a spill in one of the pipes that are transporting crude from the Constanta port to the Pitesti refinery. The event was placed spatially in a natural reservation zone, with a lot of damage that an oil spill can do.
Here we have chosen a statistical significant lot of 500 Twitter members located mostly in the supposed spill region. A very brief description of the simulated event was twitted expecting the possible responses. The lot of twitter members was following the initial tweet, retwitting it 350 times.  The main actions after the twitt are summarised in the Table 2
Table 2 Actions determined by the Twitter experiment
Type of action
Number of subjects
Announce of the Competent Authority (in our case, the Inspectorate for Emergency Situations)
120
Give solutions to stop or mitigate the spill
60
Offers to volunteer to clean the environment after the spill
300
Gives bad ratings on various forums for the company supposed guilty for the spill
375
Call to boycott the fictive company that was  supposed guilty for the spill
220

Figure 5 summarizes the Twitter experiment



Figure 5 The Twitter experiment



CONCLUSIONS

Our research has shown that social networks- facebook and others- are a possible way to imprint safety concepts and rules among young workers. The classic way of safety training generally implies little interaction, being an expositive one. The worker- and young workers generally- are more pro-active, willing to imply them into action. The desire to be cool and to gain the support for the group or “cloud of friends” could lead towards good safety behaviour- if this is properly supervised.
An integrative solution would be to open and maintain facebook safety nodes, with each node of the network centred on a specific topic. Each node should develop its own “cloud of friends” and propagate the ideas of safety around the network.
We found that, properly presented, safety could be cool for youngsters that are socializing through such a network. On the other side, specialization could be a bit difficult, given the extreme diversity of network participants.
In the wake of this research we are preparing a proposal on the next year FP7 calls in which we want to research heterogeneous- context depending learning, using social networks and being centred on the individual status improvement- inside the network and also usability outside it.
One new possibility that we tried since the initial  writing of this paper was the blog. Our blog received in 10 months more than 5000 visitors- we found this a good result considering the specificity of our domain. 

BIBLIOGRAPHY


1. P. F. Drucker, ,1999, Knowledge Worker Productivity:The Biggest Challenge. California Management Review, Vol.1 No. 2, pp. 79-94.
2. T. H.Davenport, & L. Prusak, ,1998, Working Knowledge: How Organizations Manage What They Know. Harvard Business School Press, Boston, MA, USA.
3.  E. Wenger, ,1998, Communities of Practice.Cambridge, Eng.: Cambridge University Press.
4. G. Siemens, ,2006,. Knowing Knowledge, Lulu.com, ISBN: 978-1-4303-0230-8.
5. E. Wenger, 1998, Communities of practice.Learning as a Social System. The Systems Thinker, 9(5). Community Intelligence Labs.
6. D. Zager, , S.Whittaker, , & Schwarz, H. ,2002,Collaboration as an Activity. In: Computer Supported Cooperative Work 11: 181–204, 2002.
7. H.Jenkins, et al. ,2006, Confronting the challenges of participatory culture. MacArthur
Foundation, 2006.
8. P. Levy ,1999, Collective Intelligence: Mankind’s Emerging World in Cyberspace. New York:Perseus.
9. D.Gale, S.Kariv, 2003, Bayesian learning in social networks, in Games and economic behaviour, 45(2003) pg. 329-346
10. L.Smith, P.Sørensen, 2000, Pathological outcomes of observational learning. Econometrica 68, pg. 371–398
11. V.Bala, , S.Goyal, 1998. Learning from neighbors. Rev. Econ. Stud. 65, 595–621.


Comentarii