Glass Ceiling

What the glass ceiling means

The glass ceiling (السقف الزجاجي) is a description of a barrier that limits the rise of a group to the upper levels of an organisation without being written into any rule or announced in any decision. The barrier is called glass because no document shows it, while its effect shows in how people are distributed across the levels.

For that reason the glass ceiling is an observation about a pattern, not about an event. Each individual promotion has a reason that can be stated and defended, and the pattern appears only when dozens of decisions are examined together over several years. Anyone who looks for evidence in a single decision will not find it, and that absence proves neither that the barrier exists nor that it does not.

The top figure does not locate a glass ceiling

One way of presenting the question uses a single figure: the group’s share of representation at the top level. That figure says there is a difference. It does not say where the difference arose.

Take an organisation in which a group’s share of representation across five levels runs as follows: 44% at the first level, 38% at the second, 29% at the third, 17% at the fourth and 6% at the fifth.

The useful information lies not in these figures but in the ratio between each level and the level beneath it:

• From the first level to the second: 38 divided by 44, or 0.864.
• From the second level to the third: 29 divided by 38, or 0.763.
• From the third level to the fourth: 17 divided by 29, or 0.586.
• From the fourth level to the fifth: 6 divided by 17, or 0.353.

The erosion gathers pace at the third step and again at the fourth, and the top figure alone would have shown neither. These ratios compare representation with representation. They are not the probability that an individual will be promoted, and the two measures rest on different data: the first is calculated from current headcounts, while the second needs records of individual movement over time.

Measuring the glass ceiling against an assumed constant ratio

To estimate the size of what happened, assume that the first step’s ratio of 0.864 held across all four steps. The expected share at the fifth level is then 44 multiplied by 0.864 four times, or 24.52%.

The actual share is 6%. The gap between the two figures is the size of what needs explaining, and it is larger than the 6% taken alone suggests.

This is an arithmetical assumption, not a benchmark, since nothing about the first step makes it the “natural” level. Its use is that it turns the question from “is the figure low?” into “how far does it fall, and where?”, and the second question has an answer.

The figures in this example are assumed in order to show how such a series can be interpreted. In the sources we reviewed, we found no published reference for the Saudi market on which to base a benchmark ratio for measuring the glass ceiling.

What movement data adds

All the ratios above come from a still picture: who holds each level today. They locate the step but not the mechanism, because a loss of representation at a given step can have two different causes, and each calls for a different remedy:

• Fewer promotions. Consider a third level of 120 people, 29% of whom, or about 35, belong to the group. If 9 people a year are promoted to the fourth level and 2 of those nine are group members, their share of the promotions is 22.22%, about seven points below the 29% they hold at that level.
• More departures. If 18% of the group’s members at that level left, against 11% of everyone else, the erosion would be happening before promotion rather than at it.

The difference is practical. The first cause is addressed through the promotion criteria and the list of people nominated; the second is addressed through whatever makes staying at that level possible. Addressing the first when the cause was the second means promoting those who stayed and losing those who would have been promoted.

Small numbers call for caution at this point. Had three of the nine promoted people come from the group instead of two, the share would have been 33.33%, above the group’s share of the level rather than below it. A single year therefore cannot be relied on; promotions are added up over three years before the share is calculated.

Small gaps that compound

What makes the pattern hard to notice is that the gap at each step can be small enough to attract no attention, while its effect at the end of the ladder is large.

If the ratio at each step were 0.95, then after four steps 0.8145 of the initial representation would remain. If it were 0.75, what remained would be 0.3164. The difference between the two cases is twenty points at a single step and about fifty points at the top of the ladder.

For that reason, examining only the step with the steepest fall is not enough. Several moderate falls can compound into more than the effect of one sharp fall, and they are harder to notice because each of them looks acceptable on its own.

Small numbers are unstable

Where an upper level holds few people, its ratios are sensitive to a single event. At a level of 17 people, each person accounts for 5.88 points of the share, so one promotion or one departure can move the figure by about that much.

Three constraints on presentation follow. The number of people at each level is stated beside the percentage; the series is examined over three years or more rather than over one; and no improvement is announced on the strength of a single movement. Raising the share at the fourth level from 17% to about 23% through one promotion changes no pattern. It promotes one person.

Defining the levels comes before any figure

The whole calculation rests on a division of the organisation into levels, and that division is a decision, not a given. Three methods of dividing it give three different series:

• Grades in the pay structure, such as the grades that job classification defines. They can be the most precise of the three when the grade table is well kept, and the least reliable when one grade brings together roles that are far apart.
• Depth in the reporting structure, meaning how far a position sits from the top of the organisation. It makes levels comparable between units, but it is affected by the number of layers in each unit.
• Scope of responsibility, such as the number of people reporting to a person or the size of what that person decides on. It comes closest to the meaning of a level and is the hardest to collect.

The definition of the glass ceiling does not decide which of these methods is correct. Each works under its own conditions, and the choice between them belongs to the organisation and its structure. Whichever method is chosen, it has to be declared and kept the same across the years. A series built on grades and then continued on reporting depth is not one series, and any improvement or decline that appears in it may be the effect of the switch alone.

What the glass ceiling is not

• A narrow entry point. If the group’s representation at the first level is only 8%, a similar share at the top describes who came in, not a barrier on the way up. The ratios between levels separate the two cases: with a narrow entry point they stay close to one, and with a barrier they fall.
• A pay gap between men and women. It describes a difference in pay, whereas the glass ceiling describes a difference in how people are distributed across levels. The two can occur together or separately, and each has its own calculation and its own measure.
• A disparity at one level. Where the gap sits in one particular unit, it may reflect the nature of that unit’s specialism. The term describes a pattern that extends across levels, not a difference at one point.
• A difference in one year. A short series cannot tell a pattern from fluctuation, particularly at levels that hold few people.

Where to look for the mechanism behind a glass ceiling

The figures locate the step at which erosion occurs. Representation figures describe a distribution and do not attribute any particular gap to a cause; making that attribution needs data on individual movement and on decisions, which is a separate task from calculating ratios. The search for a mechanism turns to the procedures that govern that step:

• The criteria for promotion to that level. Are they written in terms of outputs, as performance standards can be, or in general descriptions that leave room for judgement? Judgement is where rater bias can enter.
• The nomination list and who draws it up. A gap can arise before the decision, in the list of people put forward in the first place.
• The prior experience required. A condition such as having held a particular assignment becomes a barrier if that assignment is itself handed out unevenly.
• The review panel and its rules. They are examined alongside performance calibration, since the panel is where judgements are either brought into line or fixed in place.
• The visible assignments handed out. An assignment seen by the people who decide promotions is the kind on which a reputation inside the organisation can be built, and the allocation of such assignments can precede nomination by two or three years. Anyone who wants an early look at the step examines how these assignments are allocated, not only the promotion decision.

In the sources we reviewed, we found no measurement, for the Saudi market, of the effect of any corrective measure for the glass ceiling. Provisions that bear on the glass ceiling have their own sources, and the definition of the term does not settle them.

Before the glass ceiling series is presented

Four things are presented with the ratios, and the ratios are not presented without them: the number of people at each level; the ratios between levels, not the absolute shares alone; the series over at least three years; and a fixed definition of levels that does not change from one year to the next.

The last condition carries a specific risk. An organisation that reclassified its grades between two measurements has changed the ruler and then measured with it, so any difference it finds may be the effect of the reclassification rather than of promotions. Whether the levels were redefined is the first question to ask when the figure moves suddenly in a reassuring direction.

This is an explanation of the concept, not legal advice.

Related terms

• Succession Planning
• Performance Calibration
• Ageism
• Peter Principle
• Retention Rate