The first decision to be made is about which body will have the primary responsibility for approving or rejecting the draft constitution. Here the choice, as noted above, is frequently between the legislature and the constituent assembly. The second decision is whether there should be more than one stage of approval. The most common instance of more than one stage is when there is a referendum. Sometimes the draft can go straight from the drafter to the people for a referendum. This seems to have happened with the constitution of the European Union, and is stipulated in the Federal Transitional Charter of Somalia of 2004, but there are few other examples. The Kenyan process  erred on the side of too many approvals—a national constitutional conference (broadly similar to a constituent assembly), a national assembly, and— a requirement imposed by the courts when the process was nearly finished—a referendum. It is important that there should be ample opportunity to consider the suitability of the draft, but too many approvals place hurdles in the way of enactment, give further opportunities to the spoilers to regroup, and perhaps need to be avoided. (See part 2.7.)
The other major decision regarding the design of the process concerns the rules for decision- making; different institutions have different levels where a decision to carry on with the process or to approve the draft will be made. Here the primary question is the majority required. A simple majority may not be seen to give enough protection to minorities. The constitution- making process is an opportunity to build consensus on fundamental national values. Sometimes this is taken to mean that there must be unanimity (as in Kenya, Nepal, South Africa, Uganda and Vanuatu). If this fails, there may be a formal (Nepal) or informal (Kenya) process to settle differences. But they both, like Uganda, provided for a two-thirds majority if no consensus was achieved. A large majority is preferable if the country is deeply divided, especially on regional or ethnic lines. But it increases the risk that no constitution may be adopted. So does a rule that states that a minimum degree of support in a minimum number of communities or regions should be required in addition to an overall national vote (as in Iraq, where the draft was nearly defeated by negative votes in three governorates). (See part 2.5.1.)
Many states provide for a referendum, but, contrary to what a Kenyan court thought, it is not indispensable under constitutional principles. Its use is less than universal and it is controversial. If the earlier stages of the process are participatory, with compromises to reach a consensus, then it may be unwise to put what has been achieved at risk. Referendums can be divisive, increasing tensions in society (as the experiences of Iraq  and Kenya [2005; 2010] show). However, the Maldives (in 2008) made an interesting use of the referendum to determine only one issue, which had become extremely contentious—whether the executive should be parliamentary or presidential. Once that was resolved, the constituent assembly (the Special Majlis) proceeded quickly to adopt a new constitution. If the issue to be resolved touches on self-determination, a referendum can be a useful device to discover the people’s preference (as has been proposed regarding secession in Canada, and carried out in Bougainville, New Caledonia, and South Sudan).