Quantitative analysis alone will not adequately explain the strengths and weaknesses of media coverage. It is not enough to complain that the ruling party is receiving more media coverage than the opposition - there may be good reasons for this, for example in terms of their public support. But equally, simply counting the number of items may conceal the fact that some parties' "quota" of coverage may include items that show them in a bad light. For example, in South Africa before the 1994 elections the state broadcaster kept its own statistical record of party coverage, which showed that the African National Congress, then in opposition, was receiving extensive coverage. Yet this proportion included much negative coverage, such as the reporting of Winnie Mandela's trial for kidnapping. Hence the bare statistics do not tell all.

Extremely important aspects of election coverage are not readily susceptible to quantitative monitoring. Reporting of inflammatory speech, for example, will require close textual analysis of the approach that the media uses.

Monitors will also analyse the content of voter education material to ensure that no party political message is being conveyed. Often they will wish to compare the treatment of the same stories in different language services. There is often a quite different content to broadcasts in the colonial language - English, French, Spanish or Portuguese - and indigenous languages. The former will, to some extent, be for external consumption. Broadcasters and politicians often assume that no independent monitor is paying attention to what they say in their own languages.

One very important consideration is how far media reporting is accurate. Media monitors will measure bias by comparing media reporting to their own understanding of events derived from a variety of sources. One way of doing this is "source monitoring": the media monitors themselves attend an important newsworthy event such as a political rally or a press conference, in order to see how media coverage compares with their own perceptions. The Internet has made it easier for monitors to compare domestic coverage with international reporting on their country. The two sometimes bear little relation to each other.

Evaluating the implicit messages contained within media coverage is at the same time important, difficult and highly contentious. Under this heading come all the subtleties of language and visuals that convey a message that is understood by the audience, but sometimes not in a conscious manner. This can be most clearly shown in the use of words, whether in print or broadcast. For example, pro-government media may have the President "stating" something, while his opponent only "alleges". Reporting does not have to be inaccurate to be an improper influence on the audience's perceptions. In South Africa before the 1994 election, for example, monitors noticed that reports of ANC demonstrations always mentioned the amount of litter left behind by the participants. The message was that the ANC was disruptive and irresponsible. Foreign news items can be used to encourage a particular interpretation of domestic news. In Malawi in 1994, coverage of opposition parties on the state broadcaster was placed alongside news of the Rwandan genocide. The subliminal message was that an end to one-party "stability" would lead to bloodshed.

Television has a whole complex visual vocabulary. Figures who are regarded as authoritative - such as incumbent politicians - may be portrayed at an upward angle, while others are filmed at a level angle or from above. Figures in authority will more often address the camera directly, while others will address an unseen interviewer to one side of the camera and thus will not address the viewer directly. Ordinary interviewees - opposition members, trade unionists, the public - will usually be interviewed in the open air. Government members will be seen in their office, often shuffling papers and apparently engaged in some urgent and important activity. An office background tends to emphasise the authority and expertise of the interviewee. And so on.

Even the graphics and logos that accompany a news broadcast may convey a message. The graphic on South African television for the running story of political negotiations in 1993 showed two white men and one black man. This was later changed to one white man, one off-white woman and one black man. Neither of these reflected the actual composition of the negotiations. More blatantly, in the Zimbabwean 2000 elections, a special current affairs programme that run through the campaign period had as its logo the tower at the Great Zimbabwe ruins - exactly the same as the symbol of the ruling party.