Survey-based measures

The past two chapters discussed the generation, design, use and limitations of events-based measures and standards-based measures of human rights. This chapter follows suit by examining survey-based measures, the definitive features of which are their use of a large sample of individuals (usually selected randomly) who answer a set of predefined questions about either their perceptions of human rights or their experiences of human rights within their own country. As we showed in Chapter 4, many examples of events-based measures use some sort of narrative provided by a deponent, who provides a direct account of human rights violations that he or she has experienced, witnessed, or has knowledge about from accounts from family members, friends and neighbours, or other related individuals. The task of the analyst is to code the account after the statement has been taken. We also saw in Chapter 4 that the individuals who come forward as deponents are not random, but represent a ‘convenience’ sample against which particular statistical methods are used to alleviate the worst forms of bias that result from the use of such samples. In contrast, survey-based measures typically design some sort of survey instrument that comprises set questions with limited response categories, which are coded as the respondent answers the questions to the survey. Additionally, the selection of respondents is random, such that the analyst can use standard inferential statistics (see Box 6.1) to make meaningful statements about general human rights conditions relevant to the target population under examination. The key differences between the two methods are that survey-based measures use respondents instead of deponents and random instead of non-random samples. The survey instrument has been the work horse of the social sciences, polling

organizations, market research firms and analysts seeking to tap into the thoughts, perceptions, preferences, intentions and experiences of individuals since the early years of the 20th century (see Weisberg 2005: 1-16). Academic institutions have long-term funded projects on individuals and households, which use survey instruments to measure their voting preferences (e.g. the American National Election Studies and the British Election Studies); life choices, work-life balance issues, social attitudes and habits (British Social

Attitudes and The British Household Panel Survey); policy preferences, feelings, support for democracy and citizen engagement (e.g. the World Barometer and its associated regional variants, such as the Eurobarometer, Latinobarometro and Afrobarometer); and general perceptions and commitments to different sets of values (e.g. the World Values Survey). Like these more notable examples, human rights scholars and practitioners adopt the same basic methodology and use survey instruments to gauge the perceptions and measure the human rights experiences of individuals. The responses from individuals in the sample to the survey questions are then used to make inferences about the human rights situation in all or part of the country either for the time period in which

Survey research is grounded in the use of random samples. A random sample is also known as a probability sample, which means that within a population, each individual has an equal chance of being selected as part of the sample. If this is the case and the sample is relatively large, then it has a number of properties that allow for the analyst to make inferences from the sample to the population as a whole. Survey analysts are most interested in the percentage of people within a sample that give responses to questions and whether that percentage is an accurate reflection of the percentage response were a survey to be given to the whole population. The accuracy of the calculation of these percentages is a direct function of the sample size, such that accuracy increases with larger samples. For example, if an analyst wanted to be 95 percent confident in the calculation of percentage response in the population, with 3 percent margin of error, then a sample of 1,100 people from the population would be sufficient. Using such a sample would allow the following kind of statement to be made: ‘We are 95 percent confident that the proportion of the population

that agrees that the human rights situation has deteriorated in the last five years is 25 percent, plus or minus 3 percent.’ The real percentage of the population that agrees that the human

rights situation has deteriorated is somewhere between 22 percent and 28 percent. The range of values (in this case 6 percent between 22 and 28) would be bigger if a smaller sample of people were used or smaller if a larger sample of people were used. For human rights research, particularly in countries with the presence of different ethnic and religious minority groups, the generation of samples becomes more complicated in order for a fair representation of the different groups that comprise society to be achieved. In general, however, the use of random samples and survey techniques has aided in the analysis of perceptions and experiences of human rights that apply to the whole population.