**2.2.3 Statistical analysis**

The proportions of the people expressing discriminatory attitudes toward people living with HIV/AIDS were calculated using percentages. Cross tabulations were used to present the proportions of males and females with discriminatory attitudes toward people living with HIV/AIDS. Graphs were used to examine levels and trends in the proportions of people with discriminatory attitudes. Because comparison of percentages between the three surveys

Trends and Levels of HIV/AIDS-Related Stigma and

observed for both males and females.

Discriminatory Attitudes: Insights from Botswana AIDS Impact Surveys 229

Socio-demographic differentials showed that the proportion of the population who reported that they would not care for a family member sick with HIV and AIDS decreased as age increased, although slightly higher proportions of people aged 50 and above tended to portray higher levels of HIV-related stigma and discrimination. A similar pattern was

Fig. 1. Levels and trends in the proportion of males who expressed discriminatory attitudes

toward people living with HIV and AIDS

may not be reliable because of differences in the age structures in the three sample populations, direct standardization procedure was used to eliminate the compositional effects or confounding. Standardization involved taking the 2001 population in 10-year age groups from 10 to 64 years as a standard and applying to it the specific proportions expressing discriminatory attitude for the populations being compared. This produced the number of expected population expressing discriminatory attitudes which was compared with the actual number of people expressing discriminatory attitudes in the standard population. The ratio of expected divided by observed gave the standardized proportion. The standardized proportions were used to examine the levels and trends in the population expressing discriminatory attitudes toward people living with HIV/AIDS.
