English-language CV data

Entries of the Communication, Computer, JobRelated and Organisational skill categories have been defined as the text corpus that will be investigated. Each CV is considered a single document, composed of all entries the respective respondent included. This analysis is limited to English-language data.

• A total of 353.5K CVs were analyzed.
• Of those, 69.3K are in English.

• 58.3K respondents writing in English have declared at least one skill in the Communication, Computer, JobRelated and Organisational categories.

• Each document (ie. all relevant skill-related entries of a CV) has been tokenized with common stop words (eg. to, by, in) removed.

• Vectors of single words (unigrams) and word pairs (bigrams) associated with each document have promptly been determined.

Keywords associated with Entrepreneurship

The Entrepreneurship Competence Framework defines a list of sub-competences associated with entrepreneurship. A list of key words and phrases related to each one of these sub-competences has been compiled and a case is made that inclusion of these words and phrases on CVs identifies a respondent that possesses the respective sub-competence.

• Similar to the CV corpus, the key words and phrases are processed to once again remove common words that encode minimal information.
• The end-result is a list of single words and word pairs meant to be cross-referenced with the CV data.

Cross-referencing the CV corpus with keywords

The key strategy in identifying CVs marked with entrepreneurship competences is to investigate inclusion of the defined set of key words and phrases in the CV free-text data.

• CVs including any one key word or phrase are initially identified.

• Following that, the frequency of the specific words and phrases appearing in them is determined.

• Next, using data derived from previous analysis in the Europass Survey, four key data sets can be composed.

• Those data sets drive the visualizations and analysis that follows.

Resulting data sets

Data has been aggregated and compiled into .CSV format. It can be downloaded for further analysis. Specifically, the resulting data sets are:

• A frequency table of the key words and phrases associated with entrepreneurship competences. It describes the number of appearances of each key word or phrase in the entire corpus of all CVs.

Download

• A frequency table of the specific entrepreneurship sub-competences. It describes the number of CVs identified with each specific sub-competence.

Download

• Demographic data of respondents identified with at least one entrepreneurship sub-competence. It includes the birth year, sex and EQF level of the CVs of those respondents.

Download

• A frequency table of the skills of the respondents identified with at least one entrepreneurship sub-competence. It includes the ESCO classification of the Skills/competences included in the Communication, Computer, Job Related, and Organisational skill categories.

Download

Number of keywords included per CV

• 34.5K CVs contain at least one mention of at least one of the keywords associated with entrepreneurship competences.
• This number represents 50% of the total CVs written in English.

Appearances of each keyword in the whole corpus

• The most popular key words and phrases in the whole sample are showcased above. Number represents total appearances of the keyword in the entire corpus of all CVs.
• Note that a keyword may appear multiple times in a single CV.
• Keywords management and leadership occupy the top spots.
• All keywords given appear at least once, except for action learning.

Number of CVs posessing each sub-competence

• The most popular sub-subcompetences are showcased above. Number represents the total number of CVs posessing each sub-competence.
• Posession of a sub-competence is marked by the presence of at least one mention of at least one of the keywords related to it. One CV may include multiple keywords related to a sub-competence.
• 25% of English CVs include keywords related to the management sub-competence. 23% include keywords related to leadership.
• Creativity is the sub-competence displaying the greatest variety in wording. Different keywords related to it are close in frequency.

Age distribution

• The mean age of respondents using keywords associated with competence in Entrepreneurship is 28.2.
• Of those respondents, 38% are under 30 years old.

Gender breakdown

• Compared to the total breakdown of respondents in the Europass Survey, those mentioning keywords related to Entrepreneurship are slightly more likely to be male.
• The percentage of respondents who did not declare their gender is similar to the one in the general pool of respondents.

Most popular ESCO Skills

• Skills reported as free-text were mapped to the standardized ESCO Skills/competences in previous steps of the analysis. The 25 most popular of those for respondents with entrepreneurship competences are showcased above.
• Compared to the general pool, more respondents are reporting skills related to project management, work in teams, and problem solving.
• Additionally, technical literacy and computer programming skills are also commonly reported.

EQF level distribution

• The education level of respondents was mapped to the standardized EQF levels in previous steps of the analysis.
• Users reporting Entrepreneurship competences have a higher level of education on average compared to the general pool of users.
• More specifically, approximately 37% have received post-graduate education (Levels 7-8), compared to 20% of all Europass Survey respondents.

What’s next?

• An expansion of the list of relevant words and phrases could lead to more precise identification of the respondents acquiring entrepreneurship competences.
• Using translated or localised versions of these key words could expand the analysis to also include languages other than English.
• A more detailed drill-down to the specific sub-competences based on the identified concepts could yield more interesting results.
• Identifying the association rules between entrepreneurship competences and the demographic features of the respondents could also help produce more conclusions.