Targets - when and how
Setting targets should be done in an open and transparent way.
When are they set?
You need to decide the features and the feature values before you ask people to register their interest.
So for instance you may decide that you wish to have a target like this:
- feature: gender
- value 1: female
- value 2: male
- value 3: non-binary or other
It is important to do this early for two reasons:
- practicality: if you want to use a target for gender, then you will need to ask people to tell you their gender when they register their interest.
- transparency: to avoid any perception that targets have been maniuplated, it is important that they are set ahead of time.
How do we set targets for representation?
The easiest way to set targets is to make them representative using official data such as a census. This is how we might set out our working when we set gender targets for an assembly of 50 people:
| Gender | Census | Number | Minimum | Maximum |
|---|---|---|---|---|
| Female | 49% | 24.5 | 24 | 25 |
| Male | 48% | 24 | 24 | 25 |
| Non-binary or other | 3% | 1.5 | 1 | 2 |
To explain: Census data says that 49% of the wider population is female. Since 49% of 50 is 24.5 this is the number of females that we would want for our assembly to be perfectly representative. We therefore set our targets to be min=24 and max=25 for this feature value.
A little extra: notice that for “non-binary or other” our calculations have resulted in min=1. In this situation we may decide to increase to min=2 (and max=2) in order to be certain that this group is definitely represented in our final set of assembly members. If our final set only has 1 person who is “non-binary or other”, then a late drop out could end up with there being no one from this group at all.
How do we set targets for inclusion?
Sometimes we may not have available census data to use in setting targets… but we may still feel that a target is important for reasons of inclusion.
For instance a democratic lottery for a council may ask people if they have “been involved with any previous engagement activities with the council”. We might then set, say, max=5 for people who answer yes to ensure that there are not too many so-called usual suspects in our pool of respondents.
Not too many!
It is important to use targets in a targeted and sparing way for several reasons, including:
- Explainability: there needs to be a good justification for asking people for information about themselves. If we ask for too much data people may get form fatigue and not complete the form, or they may be (rightly?) suspicious that there is not sufficient justification for asking for all of this information.
- Complexity: every time we add another target we make it mathematically more difficult to find a set of 50 people from our pool of respondents that satisfies all of the targets. Too many targets will make it completely impossible!
The complexity consideration also explains why we don’t do targets for intersections of different features: we have targets for gender and targets for age but we don’t have targets for the combination (e.g. the number of females aged 15-24) because we quickly end up with too much difficult mathematics.