A first-of-a-kind study uncovers online profile connections to performance.
Social networking sites can reveal a lot about a person, but do they offer insight and value to the hiring manager? How do they measure up when compared with the best candidate selection systems? Can online profile information predict future performance—to a statistically significant degree?
To investigate the application value of online profile data, our team delved into LinkedIn. We explored the relationship between LinkedIn-captured profile information, structured performance evaluations, and DDI’s own pre-employment assessment results.
We chose LinkedIn because, unlike Facebook and Twitter, the networking site is career-focused and relatively free of personal, non-work-related clutter.
In addition, LinkedIn has become increasingly ubiquitous. According to the company’s website, from 2008 to 2012 LinkedIn grew from 32 million to 200 million members; two new people sign up every second. It’s also the social site through which many more hiring managers have selected new employees (Figure 1). And, unlike privately submitted résumés, LinkedIn profiles are more open to scrutiny, more verifiable, and less prone to embellishment.
Data Meets Data
Our study could not have been conducted without access to DDI’s robust database. As a result of large-scale assessment projects conducted over four decades, we have detailed proficiency and personal attribute data on individuals and job families. Included in this knowledge repository are self-reported ratings of key (work) competencies, including Active Learning, Decision Making, and Customer Focus. We looked at recent sets of data (assessment results paired with performance evaluations) across a range of jobs, and identified participants who also had LinkedIn profiles. One hundred fifty professionals and 437 sales workers from 11 organizations comprised our sample.
Our first area of inquiry: What can be learned about someone’s real-life work style from their online profile? Are there links between personal attributes and profile data? The answer to the latter question is mostly yes. We found that decision-making style (the tendency to make rational, timely, work-related decisions), followed by judgment and assertiveness (tied), were related to the greatest number of investigated profile elements (Figure 2). Additionally, participants with higher assertiveness and achievement orientation scores were more likely to be LinkedIn members, and had greater numbers of connections.
Next, we asked if the profile information related to on-the-job performance. The answer here: a resounding yes! We looked at a variety of performance competencies—Active Learning, Decision Making, etc.—and gleaned several key insights: Stronger performers were more likely to have more LinkedIn connections and more upward progressions when moving from job to job. Performance on Active Learning, Decision Making, Work Standards, and Sales Analysis competencies had the greatest number of connections to many profile data points (Figure 3). We also found one part of the profile that negatively correlated with performance: the longer a person spent in between jobs (indicated by gaps in employment), the less likely they were to be top performers.
Our results also support the adage that it is better to give than to receive. Performance was more positively correlated with the number of recommendations given by LinkedIn members than by the number they received.
The following chart shows the relative strength of several profile elements when measured against performance (Figure 4).
Conclusions and Caveats
Although the results of our LinkedIn studies are promising, we are not advocating abandoning proven methods for gathering objective, structured information to predict job performance. Although we did find several links between online profiles and performance, the strength of these links was weaker than formal selection tests also completed by the study participants.
Keep in mind that all LinkedIn profiles are not created equally. For certain demographic groups and types of positions, the prevalence of profiles varies greatly. Hourly workers are less likely than sales and professional workers to have profiles, making it impossible for a company to use LinkedIn as an across-the-board selection tool. Even when they have profiles, women and those over the age of 40 are less likely to have a large number of connections (we did not explore the reasons for this). Our findings also do not necessarily reflect universal patterns for every job and every profile.
And, although we do see some trends between certain parts of the online profiles and better job performance, not all data in a profile can be linked to performance. For example, we saw no connection between a participant’s number of skills/experiences and performance. Recruiters routinely assign unsubstantiated meaningfulness to this data, unwittingly adulterating their selection systems in the process.
Last but certainly not least, it’s important to weigh the significant legal implications of these tools. We already mentioned that many demographic groups are not represented equally online; so, in essence, those without profiles are excluded (and must be given alternate consideration).
Also, organizations must be able to demonstrate the job relevance of the data—from all sources, including online—added to the selection mix, and have a structured, legally defensible approach for reviewing social profile data. Be prepared to detail the type of profile information used, and how and why it was chosen. This will enable you to defend your process if faced with a legal challenge. The bottom line on using social media: Consider every piece of data and determine how it fits into the broader selection process in the same way you would use information gleaned from structured interviews, pre-employment tests, and simulations.
Evan Sinar, Ph.D., is chief scientist and director of DDI’s Center for Analytics and Behavioral Research (CABER).
Jamie Winter is a manager in DDI’s selection solutions group.