Five-yearly review of employment conditions


Article S V 1.02 of our Staff Rules states that the CERN “Council shall periodically review and determine the financial and social conditions of the members of the personnel. These periodic reviews shall consist of a five-yearly general review of financial and social conditions;” […] “following methods […] specified in § I of Annex A 1”.
Then, turning to the relevant part in Annex A 1, we read that “The purpose of the five-yearly review is to ensure that the financial and social conditions offered by the Organization allow it to recruit and retain the staff members required for the execution of its mission from all its Member States. […] these staff members must be of the highest competence and integrity.” And for the menu of such a review we have: “The five-yearly review must include basic salaries and may include any other financial or social conditions.”
The next Five-yearly review has to be concluded in December 2015, when Council will decide on proposals brought forward by the Director-General after consultation with the Staff Association and the Member States in the framework of the Standing Concertation Committee (SCC) and TREF.

In fact the list of topics to be considered has to be defined by the CERN Council in June 2014, upon suggestion of the Director-General. Therefore, in the autumn of year n-2 before the end of each five-yearly review cycle the Staff Association organizes a survey amongst staff to find out their wishes regarding the items they would like to see reviewed under the heading “other financial and social conditions” (knowing that a salary comparison is mandatory). Thus, in November 2013, we organized such a survey, whose main results were presented in public meetings in February 2014. The complete results are also available on the web.

In Echo 189 we showed some statistics on participation, showing that the 55% of staff who responded to the survey were representative of the whole CERN staff population. In the following issues of Echo we will go through the results of the various chapters of the survey in detail. But first, in this issue, we explain the methods we used to analyse the replies in the 1383 questionnaires fully filled out (Fig. 1)

 Fig. 1: Analysis of the questionnaire data

Calculational method of analysis: correlation coefficients

This first method tries to answer the question “who thinks what” by calculating, with the help of the Mathematica program, correlations for types of answers, based on the personal information data available. For each individual reply all personal data (dimension 1) is crossed with all possible replies in chapters of the survey, i.e., Priorities, Contract Policy, MARS, Working Time, and Family Policy (dimension 2). Each of the possible answers for dimensions 1 and 2 has been given a unique identifier (hash code), so that we obtain N {=dimension 1*dimension 2} possible points (x,y) for one reply. Doing this for all replies yields about one million points. Note that specific (x,y) pairs can occur several times (e.g., some “males” may have replied to some questions in the same way). Therefore, we encode this number of occurrences in a 3rd dimension so that we obtain a three-dimensional set of points for each chapter in the survey: p3d = (x: personal-info-hash, y: mars-reply-hash, z: occurrence-count). The calculation of all these points is rather time-consuming, so it is calculated once and saved for later use in the analysis.

One can now select a pair of personal data (e.g., “male” and “female”), corresponding to two cuts through the p3d surface, and compute the similarity of these two curves. The result is a Pearson’s correlation coefficient, which measures the shape-similarity of the curves in question, i.e., the degree of similarity in the replies given by “males” and “females” regarding their opinion about the existing MARS, or about wishes for a future contract policy, etc., for all chapters of the survey. Table 1 shows an example of such an analysis for the questions relating to the acceptance of the current MARS system, with only a subset of rows shown. These correlation coefficients are rather relative measures of similarity, and the information is contained in the comparison between them. If the correlation is close to 1 (green in the table), answers of the two selected subgroups are similar. The more the value diverges from 1, the more different the opinions are (light brown to darker brown in the table).

Table 1: Example of correlation coefficients for replies to“ satifcation with existing MARS system”


In Table 1, we observe that the opinion of men and women are quite similar (row 1). Staff in the middle of their careers (in their forties and fifties) have similar views (row 2), whereas those in earlier stages (in their twenties) and the later stages (in their sixties) of their career have quite different opinions (row 3). Similarly, an increase in difference in level of education (from 4 row 5 to ) seems to lead to larger differences in opinion. Staff in career paths E and F have similar view on MARS (row 6), those in career paths F and G somewhat less (row 7), whereas there is no correlation whatsoever between the views of staff in career paths A and G (row 8). There also seems to be quite a difference of opinion between LD and IC contract holders (row  9).
This method assumes full independence of all replies, which is only approximately true. Moreover, the number of entries in a given sample (e.g., career path A in row 8 above) can be quite small so that the value of the calculated correlation coefficient has a non-negligible uncertainty. Therefore, an informed interpretation of the correlation coefficients should take these limitations into account.

Graphical method of analysis: plots

The second method of investigation uses a graphical representation of the answers and looks for differences in response patterns with respect to personal criteria. To ease the visual inspection, the possible replies were combined into three categories: “globally agree” (green in Fig. 2, grouping “totally agree” and “agree”), “globally disagree” (red in Fig. 2, grouping “partially agree” and “disagree”), and “No opinion” (black in Fig. 2). We can thus easily observe any variation in the green-red pattern between the various populations. For instance, in Fig. 2, which is concerned with the question “The current MARS suits me”, we see that staff in earlier parts of their career (ages below 40), with higher technical training, and in career path C seem to be somewhat less negative than other subgroups. The only significant difference is between LD and IC holders. Note that this method provides only some visual guidance. Thus, observed differences should be studied in more detail by the correlation coefficient method described previously.  


 Fig. 2: Graphical detailed analysis of answer

Structuring the comments

To complement the information we got from analysing the answers to the questions we also studied in detail over a hundred pages of comments. Each comment in the survey relates to a specific question. To allow for a more systematic analysis, we first group the questions by “Themes” and we relate every comment to one of the chosen themes (e.g., Fig. 3 shows the six themes we chose for the MARS chapter). On top of that, for a given theme we categorize each comment in terms of a “sentiment” (judgment). We also extract possible proposals or critics from the comments. Finally, we structure the information pictorially by grouping the comments according to whether they answer some specific questions (Fig. 4).



 Fig. 3: Analysing the MARS comments



Fig. 4: Pictorial view of comments about promotion and advancement


During the three months since the period for participating in the survey ended, the members of the Staff Council, and in particular those active in the Commission “Employment Conditions”, have been hard at work to analyse the replies using the three-prong approach of Fig.1 as explained in this article.
The results are still being studied carefully and will be complemented by the feedback received during our recent public meetings. You, as a staff member, can also continue to provide us with complementary feedback by contacting one of your delegates in the Staff Council or by sending an email to
Based on all this input your Staff Association representatives will be able to have a clearer picture of the wishes of all staff in their discussions with Management and Member States in the coming months.



par Staff Association