7 deadly sins of social network analysis

(数据分析师必读)7个社交网络分析的致命坑

Social network analysis (SNA) is a method of analyzing and visualizing social structures, represented in a form of graphs. Although the method itself isn’t new, it has recently become popular,as new software is developed which makes SNA easy and cheap to carry out.However this method has some pitfalls, which may be problematic for analysts who aren’t familiar with it. Here I present a list of 7 mistakes, which keep reoccurring in the field of organizational psychology.

社交网络分析(SNA)是一种对社交结构化数据进行分析和可视化并以图表方式表示出来的技术。并且随着一些新软件的开发使得社交网络的分析变得更为容易、成本更低。SNA尽管已经不是新被提出，最近却越来越流行了。不过话说回来，对于不是特别熟悉的数据分析师来说，社交化分析还是有些坑要填。这里我列出了一个列表，包含了7个在组织心理学领域经常重犯的错误。

1.Building a network using a single question

1、用一个简单的问题建立一个网络

If you ask your participants to name key sources of useful information, you might end up with everyone pointing immediately to the internal trainer. In order to grasp a complex psychological phenomenon like social relations or informal communication paths, you need to use more sophisticated tools.

如果你要求参与者对有用的信息进行关键来源标示，你可能会发现到最后每个节点都指向内部的训练人。为了能鸟览一个复杂的心理学现象如社会关系、信息传播路径，你可能需要使用一些更复杂的工具。

I’ve already talked about this issue in one of my previous posts. If you want to measure psychological phenomena accurately, you need to get familiar with the field of psychometrics and learn methods used to test the accuracy and reliability of your measurement.

我已经在我之前的帖子里面讨论过这个问题，如果你想精确衡量这些心理学现象，熟悉相关这些心理学原理，学习用于衡量测量量准确性和可靠性的方法是很有必要的。

2.Asking the wrong question

2、提出错误的问题

A couple of years ago I was asked to conduct an SNA analysis in order to identify informal paths of knowledge transfer in a large organization. The client wanted to map the entire organization by asking employees about people, to which they turn for advice. This idea had one major flaw. You can of course ask this question in your sales team and see who is the source of useful knowledge concerning sales. You can also ask the same question in your marketing department and conduct a similar analysis there. However people working in these two departments will answer your question thinking about two different areas of knowledge. If you use this question to collect your data, the moment you draw a connection going outside the department, you can no longer make sense of your results. You can’t conclude that there is an informal path going from Mark from the sales department, through Susan in marketing, and ending on Laura in the accounting department. Based on such results, you can’t assume, that by supporting Mark you will help Laura in any way. As it often turns out in organizational research, the trick is to ask the right question, everything else are just technicalities.

我在前些年进行了一项SNA相关的分析：识别一个规模较大组织中，知识传播的隐式传播路径。客户希望对整个组织的雇员进行调研，他们遇到问题都通过什么方式获得帮助。这个方案有一个主要的缺陷：你可以在你的销售团队对这个问题进行调研，然后看看销售相关的知识是来源于谁；你也可以用同样的策略应用到你的市场部门，进行同样的分析；然而同时在这两个部门工作的同学而言，他们会同时需要考虑这两个领域的知识。如果你用这个方案来收集数据，当你开始描述部门外的情况的时候，结果并不再适用。无法断定从来着销售的Mark到市场部的Susan，最后到会计部的Laura有一条隐式的知识链。基于上述逻辑，你并不能假设，支持Mark会对Laura有什么帮助。在组织研究的领域中，诀窍在于如何提出正确的问题道，其他的方面都是术。

3.Ignoring missing nodes

3、忽略被遗失的节点

When you want to conduct a survey to collect opinions, having a response rate of 80%-90% is a very good result, which allows you to draw accurate conclusions. However, if you use a survey to collect data for SNA, losing a single person may severely lower the accuracy of your research, if this person happens to be a hub (a strongly connected node). Losing hubs is a big problem in SNA, so if you can’t map the entire network you should at least make sure that no key actor has been left out of the analysis.

当你想要统计一次调用的结果，用户的回复率在80%-90%已经很好了，这让我们可以描画一个准确的结论。然而，如果是用于SNA的调研数据，由于每一个节点都有可能是你个中心节点（一个强链接节点）丢失一些独立的节点可能会严重降低一次研究的可信度。中心节点的丢失是一个大问题，所以，如果你无法保证能完整把整个网络囊括，那至少保证没有关键节点被排除在分析之外。

4.Showing social maps to everyone

4、向所有人展示社交脉络

The market offers a number of online tools that make it possible for employees to construct and see the entire social net work of their organization. As an organizational standalone specialist whois an extreme introvert, with few social connections (or none) inside the company. Does it sound like a good idea to shine a spotlight at this guy in front of 500 other employees? On the other hand, making an SNA map publicly available is also dangerous for well-connected participants. Knowledge on who communicates with whom, where are the key paths of knowledge transfer, and where are the key players can easily be used to hurt your organization and your personnel.

市场上提供了一系列的在线工具使得员工可以统计和看到所在公司的社交网络全貌成为一种可能。由于一个组织的专家们往往是相对比较内向的，在公司只有很少的社会关系(或者没有)，你觉得把这样子的员工公开在其他的500个员工面前听起来像不像一个比较靠谱的做法？从另外一方面来说，做一个公开的SNA蓝图对于那些频繁被联系的人来说也是不安全的。谁和谁交流了什么知识，知识传播的关键路径，知识传播的关键人等这些信息很容易被用来伤害你的组织或者个人。

Programming an analytical tool is onething, but knowing how to use it responsibly is something completely different,and requires a different set of skills. Results of SNA (as any other researchresults) should be treated carefully, and reported along with a clearexplanation only to people, who can understand and use them appropriately.

知道程式化一个分析工具是一回事，然而知道如何负责地使用又是一件完全不同的事。这些都包含了不同的技能集合。SNA(或者相关的研究结果)都应该被谨慎的对待，并且单独给那些需要理解和使用这些数据的人清晰的解释。

5.Delivering trivial conclusions

5、琐碎的结论(译者注：常识一般的结论)

If you want to identify informal paths of knowledge transfer, and after surveying and mapping 300 people you conclude that they tend to turn to their managers for advice, you should probably rethink the way you use your R&D budget. As silly as it may sound,“discovering”obvious facts with SNA happens quite often both in organizational and scientific research. For instance, you can find research that concludes, that people tend to group around charismatic individuals. SNA is a powerful tool, however it’s easy to focus too much on unusual measures and pretty graphics, and forget about the main goal, which is delivering useful recommendations.

如果你想要识别知识传播的隐形路径，并且在测量和映射300个人后，你得到结论：如果这些人倾向于向管理人员获取建议，那就需要重新考虑研发(译者注：Research and Development)预算的分配。这听起来就很蠢，然而这样显而易见的“发现”在公司和科学研究机构对SNA的研究中经常出现。例如，你发现人们会围绕在有魅力的个人周围。SNA是一个很有用的工具，不过会很容易让人聚焦于不寻常的思维方式、漂亮的图形展示上，而忘了使用工具的主要目标是提供有用的建议。

6.Using SNA when a simpler method is available

6、简单问题复杂化(一个简单的方式可以实现的情况下还要使用SNA)

Asking people about their connections is quite invasive and tends to raise concerns among participants. While designing the research, it’s a good idea to consider if the same result can be achieved using different techniques. For example, you don’t need to draw a full social network to identify charismatic leaders. Similar results can be achieved using other techniques, like surveys or 360̊ assessments.Such techniques also tend to be more resistant to typical SNA-related problems,like the missing hub problem mentioned before.

通过询问人们的关系是一个侵略性很强的方式，而且容易引起参与者的的警惕性。所以，再设计研究时，如果考虑使用不同的技术结合是一个更好的方案。例如，你并不是一定要绘制一个完整的社交网络来识别很有魅力领袖，使用其他技术(如知识传播调研或者360度评估)也能获得同样的结果。这些技术往往能更有效的抵抗SNA相关的典型问题，如之前我们提到的中心节点丢失的问题。

7.Using old results to solve new problems

7、使用过时的结论解决新问题

As I mentioned before, removing one key node may change the network structure entirely. Therefore it’s crucial to keep in mind that SNA results (like all other research results for that matter) have a certain expiration date. Although establishing it precisely for SNA may be challenging, the data will eventually“go bad”, especially in organizations with high rotation.

入职前所述，删除一个节点会导致整个网络的结构。因此，我们需要时刻记住有一个关键的节点(如其他的研究结果一样)又会有一个保质期。由于数据最终会“变坏”，尤其在高效运转的组织中，建立一个SNA是件非常有挑战的事。

To summarize, social network analysis is a powerful method, capable of delivering some really valuable insights. It has however some pitfalls, which are quite dangerous and at the same time unique to this method. Bearing them in mind during the early phases of the project can save a lot of time and nerves, and help to draw insightful conclusions from this interesting yet complex method.

总而言之，社交网络分析对挖掘有价值的见解上，是一把利剑，一种能力。然而他有些独一无二的，相当危险的陷阱。在项目的初期，带着对这些问题的考虑，可以节省大量的时间和精力，并从这个有趣而复杂的方法中得出有见地的结论。

by Grzegorz Rajca

译者：沐曌