Information sharing using voting
This is a resource file which supports the regular public program "areol" (action research and evaluation on line) offered twice a year beginning in mid-February and mid-July. For details email Bob Dick bd@bigpond.net.au
... in which multiple (and perhaps cyclic) voting is offered as a robust procedure for exchanging information about preferences in a group setting
A Polish translation by Valeria Aleksandrova is available here.
Contents
- Introduction
- Two
votes and one vote
- When
consensus is difficult
- Choosing
criteria
- Using
criteria
- Other
consensus producing procedures
- Notes
Many group processes generate information lists which are quite lengthy. This has to be reduced in some way. This document 1 examines some robust methods for doing so. (By robust I mean that they continue to work in inexperienced hands and difficult situations.)
Here's an example of a situation that may be familiar to you. You have collected information from a group of people. It's a long list -- too long to be able to act collectively on it. You'd like to reach agreement on the top item.
You may have used discussion for this purpose. If so, you'll know that, as a process for taking all views into account, it doesn't work very well.
When there are high levels of cohesion and trust and openness, discussion can be an enjoyable and effective way of arriving at a consensus. But this is a tall order: such groups are rare. It is safer and easier to reduce or avoid the amount of face-to-face interaction.
Voting is one way of doing this. Sometimes a consensus exists already, and needs only to be made public. A single round of voting may then be simple and effective. The level of consensus can be most easily identified by using a voting procedure which allows people multiple votes.
You can increase the usefulness of a single round of voting by giving careful instructions beforehand, when you collect the information. Interposing work by pairs or threes during information generation can build more overlap into the information while making it hard for any one person to exert undue influence.
Giving each group member a number of votes increases the overlap between one person's votes and another person's votes. More information can be gained from a single round if a distinction can be made between two levels of voting (as with the two-vote/one-vote method described below). You can increase the overlap further by reducing the weight people can give to their own items.
Two votes and one vote
I'll assume you've already collected a list of information publicly.
You've recorded the information on newsprint (or something similar), with a working margin on the left hand side of the sheet. The items have been numbered from 1 to whatever.
1. Let l be the number of items in the list, as they say in statistics texts. Let n be the total number of participants. Then x lies between the number of items divided by the number of people, and four times that amount. That is ...
(l/n) is less than x is less than 4(l/n)
If l is not evenly divisible by n, round upwards. For example, suppose there are 100 items and twelve participants. Then x lies between 9 (100 divided by 12, and rounded upwards) and 36 (four times the amount). I determined this equation by trial and error. It works quite well on average.
When participants are similar to each other and the topic is relatively uncontentious, choose the lower limit. Where they differ widely, or the information is more contentious, stick to the middle of the range. If you are in doubt, the middle or upper limit is the safer choice. In a small group, err on the high side. By the way, there is no point in x being larger than one-third of the list.
If you want to spread out the whole list of items, choose a larger x. On the other hand, a smaller x is more likely to identify just a few items from a long list.
2. Be clear in your own mind about the aims of the next step. Then choose criteria accordingly. You can ask yourself, "Which items of information will provide the best starting point to the next phase?" A common criterion is importance. Be specific.
3. Explain that it is necessary to reduce the list to a smaller number of items in readiness for the next phase. A voting procedure will be used to do so.
Explain that you do not intend to combine items. (It requires discussion, which can be difficult to facilitate). The multiple vote allows the overlap in items to be taken into account. In any event, there is nothing to stop people taking all the items on the list into account when they later move on to action planning. You merely wish to identify the items that are seen as most important (or whatever).
If this does not satisfy the people who want to combine items, ask them instead if they will settle for multiple rounds of voting.
4. Participants work individually and without discussion. Ask them to choose the x items (as calculated earlier) which best fit the criteria, excluding items which they themselves contributed to the public list. (They can vote for items which were on their private list but which they didn't contribute because someone else had already done so.)
Inform them that when the time comes they can give two votes to each of these chosen items. They will do this by holding up two hands. Draw their attention to the numbering; suggest that they write down the numbers of the chosen items. Explain that they are then less likely to change their mind in the light of other people's voting.
5. Now ask them to choose, and note down, the x items which are the next best fit to the criteria. They can include their own suggestions in this list of x items. Inform them that when the time comes, they can give one vote to each of these chosen items. They will do this by holding up one hand.
6. Explain that you will read out the items one at a time, each with its identifying number, from 1 to l. When you come to an item that was one of a person's first x choices, she holds up two hands. When you come to an item from the second list of x, she holds up one hand.
"Two votes, two hands [which you demonstrate], for the x most important items. One vote, that is one hand [again with a demonstration], for the x next most important items."
Don't be afraid to say it three or four times. Otherwise someone will wait until it is all finished, and then say "I held up one hand for my first choice. Was that what I was supposed to do?"
7. Ask participants to stick to their chosen items no matter how many or few other people vote for those items. Point out that it can be important to know that at least one or two people included an item on their first or second list.
8. Read out the items slowly and clearly, each with its number. For the first item, trigger the raising of hands by then saying "Right, two votes for one of your first x items, one vote for one of your second x items. How many votes for number 1?"
It helps if you point to hands as you count them. People will otherwise drop their hands before you are finished counting. Count aloud, so that people have an opportunity to correct you if you make a mistake. Record the number of votes (preferably in a different colour) in your working margin directly on the newsprint.
Don't use up all the space in your working margin, as you may need some of it for a repeat vote.
9. When the list has been exhausted, look for natural gaps in the order of priority of items. You will often find that somewhere between one and six items may cluster together in total votes, followed by a gap to another cluster of items. Underline the items up to the first gap in priority. Then use dotted underlining for the next block of items.
10. If convergence on a few items doesn't emerge, cross out the items receiving only 1 or 2 votes and repeat the process from the beginning. Do the same if some participants believe that overlap between items has distorted the voting.
11. Don't count the total number of votes. They seldom add up to what they are supposed to. If someone starts to count the votes, say with a laugh, "It's easier not to count the votes, as they seldom add up properly." (I have developed the habit of marking items attracting a lot of votes "Lots". Counting is then impossible.)
12. You may now wish to transfer the priority items to a separate sheet of newsprint, in order of priority. This makes later reference to them simpler. I suggest, however, that you also continue to display the full list.
In a small group it is simpler for people to record their votes directly on the list. You can then use more than just two levels of voting.
Even in a group which is open enough for discussion to work, a simple voting procedure can be used to give a quick interchange of information about priorities. Discussion may then work well enough for the actual decision-making.
In this instance you may find it useful to explain that this isn't an exercise in majority rule. The intention is to make a decision which as far as possible takes all points of view into account.
When consensus is difficult
A single voting round is often adequate if information collection is carefully planned and done. It suffers from a major drawback, however. There is no interchange of information before people vote. If there is not a clear consensus at the beginning, one voting round may not be enough.
Multiple rounds of voting can then be used. They allow an interchange of information without the problems of discussion. They give people an opportunity to change their vote from items which have no chance to those where their vote can have some effect.
As in the procedure above, a long list can be shortened between rounds by removing items which receive no votes, or very few votes.
Even with multiple voting rounds, much less time is likely to be taken up than by discussion. Every person has an equal say in decision making: competition is therefore controlled. Some people may change their mind during a round when they see they are the only person voting for an item. But for the most part, it is difficult for one person to sway the group. Conformity is thus reduced.
In theory the exchange of information can be more thorough with discussion than with voting. In practice it is a different matter. Unless you manage it with sensitivity and skill, a few people are likely to do most of the talking. Those with more power or prestige are likely to have more say, whether or not they have better ideas.
Here is a simple modification which further increases the exchange of information...
Use multiple voting rounds but with different criteria each time. There are a number of ways in which this can be done. The participants can be asked to choose the criteria, and then apply them. Alternatively, you can choose the criteria or use some standard set.
However the criteria are selected, the votes under different criteria can be summed for each item, or a final round of voting can be used to decide the final priority. (I much prefer the latter.)
Choosing criteria
The simplest approach is also most economical of time. You choose the criteria. Sometimes, however, the advantages of letting group members choose the criteria are more than adequate compensation.
o Group members are sometimes in a better position to know what criteria are most important for the issue being worked on.
o A clearer consensus may occasionally be obtained with participant-chosen criteria.
o Participants may have more trust in criteria which they have chosen themselves.
o Where only partial consensus on priorities is likely, participant commitment will be higher when they have chosen the criteria. Even if a particular participant doesn't agree with other group members' choice of priorities, she will probably go along with them if she has helped to choose the criteria. If you choose the criteria, participants may blame undesired results on you, not on the merit of the items.
The last of these is the most important. The more contentious the issue, and the higher the costs of lukewarm commitment from group members, the more worthwhile it is to have criteria chosen by participants.
Choice of criteria can use the same process as choice of other items.
1. Generate a long list of possible criteria using one of the information collection procedures already described.
2. Use voting, in multiple rounds if necessary, to select the most useful criteria.
It is better to select criteria before generating and collecting information. Some participants may otherwise choose criteria which support their favoured items, making consensus on criteria difficult to achieve.
To compensate for the self-censorship of items which may result during information collection, you may wish to give more than usual attention to the groundrules for information collection. Interposing a brainstorming-style warm-up between selection of criteria and generation of information may be useful in reducing self-censorship.
On other occasions, participants may believe criteria selection is your job, not theirs. The criteria they choose, being based on less information and experience, may not suit the particular process you are using. It may deflect them from their real interests, in which case they may lose some of their enthusiasm for the task.
Where you are choosing your own criteria, there are some in common use which may suit your purpose, or may act as a guide. The criteria depend on the particular task.
If you are deciding which issues to take as a starting point for action planning, you are probably interested in the importance of items (that is, how much effect do they have), and their ease of change (how easily you could do something about them).
John Damm, a friend and ex-colleague of mine, uses four criteria in his version of the same problem analysis technique.
o Seriousness. How severe is the item?
o Urgency. How necessary is it to act soon on the problem?
o Growth potential. If the problem could be remedied, how much potential benefit might the solution generate?
o Ease of change. How easily could something be done?
When deciding which of many potential problems you should give attention to, there are again two useful criteria. Severity determines the cost if the problem does occur. Likelihood is the probability that the problem will arise. The product of the two may be used to decide if you should allow for the problem in your planning. 2
Using criteria
After criteria have been used to generate the votes you have a choice of three ways of proceeding. You can ...
o sum the votes over all criteria for each item;
o first multiply each vote by some weight which reflects the importance of each criterion, and then sum the votes; or
o treat the votes as an exchange of information; have a final voting round to determine the final priority.
I prefer the third of these. It is simpler than giving weights to criteria. Different people may attach different weights. The final vote, being a global judgment, is more likely to take into account other factors which the criteria don't address.
(This isn't usually a big issue. The results usually do not depend greatly on the approach you use. If some items are clearly more important than others, they will usually emerge whatever the method.)
Repeated voting produces convergence even when each round of voting is about a different aspect of the items. For this reason, I favour different criteria rather than repeated criteria for multiple voting rounds. It provides more information, and is more interesting.
Using the example given earlier, one would think that importance and ease of change are if anything negatively related. In practice, many people seem reluctant to give a vote for ease of change to an item that is too unimportant to consider. I don't mind this, as it increases convergence.
So, over three rounds of voting (say for importance, ease of change, and final priority) the leading items will stand out more prominently with each round. By the end of the third round it is usual for a clear priority to emerge.
Other consensus-producing procedures
Exchanging information about the items identifies the items on which there is consensus. Voting therefore need not be used for this purpose.
Cyclic structured discussion and voting
One variation is to use voting to assess the amount of consensus. However, intersperse each round of voting with a structured discussion. To prevent a debate which might polarise attitudes, groundrules are then needed for the discussion.
Here is one example of the procedure. It is assumed that the list has been already compiled.
1. A round of voting for overall priority is held.
2. Participants are given individual thinking time to prepare for the discussion which follows.
3. There is a structured discussion in which any participant may speak for 1 minute provided ...
o she contributes only concrete, factual information; and
o does not argue directly for or against any item;
and any participant may ask questions for clarification, to which any participant can reply subject to the same groundrules.
4. There are further similar rounds of voting and discussion until some preset criterion of agreement is reached.
Focus cards
You can use small systems cards (125 x 75 mm.) to allow people to contribute information anonymously. (Cards have many similar uses.) Assume that a list of items has already been collected on a number of small cards, c cards per person. The cards are placed face down in the centre of the group.
1. Each person takes, at random, an equal share of the cards.
2. On each card she places a tick (signifying agreement) or a cross (signifying disagreement). She may if she wishes write one brief sentence giving a reason for the agreement or disagreement. The cards are returned face down to the centre and shuffled.
3. This is done rapidly for (say) four rounds. The same procedure as before is now followed, with three variations.
o Any two cards which contain identical items are stapled together, and their combined ticks and crosses transferred to the topmost of the two.
o Any card which has half as many ticks as people in the group, and no crosses, is placed in an Accept pile. So is any card on which the ticks outnumber the crosses by half the number of people in the group. (Or negotiate your cutoff points with the group.)
o Any card which has half as many crosses as people in the group, and no ticks, is placed in a Discard pile. So is any card on which the crosses outnumber the ticks by half the number of people in the group.
The remaining cards are returned to the pile as before.
4. The process continues until the original number of cards has been reduced to half (or some other proportion) of its original size.
Combining collection and analysis
Information analysis and information collection together define a recurring sequence in group problem-solving or similar task activities. There are many group processes which are examples of this sequence.
To summarise the main features of the sequence, here it is in typical form.
1. Each person is given time to think individually about the issue being addressed, and to list relevant items.
2. Each person compares lists with a neighbour, looking for similarities between their lists.
3. Each person individually ranks the items in her own list.
4. Items are collected by going round the group, with each person (or each pair) offering the next most preferred item which has not already been given.
5. Extra items are collected in a final brainstorming session.
6. Each person individually chooses her x most favoured items from the list, and her x next most favoured items, where x is somewhere between the number of items divided by the number of people, or four times that quantity. She is barred from including her own items in her first x choices.
- Modified from Bob
Dick (1991) Helping groups to be effective. Brisbane,
Australia: Interchange.
[ back ]
- Kepner, C. and
Tregoe, B.B.(198) The new rational manager. Princeton:
Princeton Research Press. These criteria are used in the
process for potential problem analysis.
[ back ]
_____
Copyright (c) Bob Dick 1991-2000. This document may be copied if it is not included in documents sold at a profit, and this and the following notice are included.
This document can be cited as follows:
Dick, B. (1997)
Information sharing through voting [On
line]. Available at
http://www.aral.com.au/resources/voting.html
