MARCH Outcome Evaluation

Therefore, coefficient alpha will be equal to zero. If all items are perfectly reliable and measure the same thing (true score), then coefficient alpha is equal to 1. In most literatures a coefficient value exceeding 0.6 is an indicative of internal consistency across the items.

The estimated scale reliability coefficients for the scores measuring positive attitude towards PLWHAs and non-differential treatment of PLWHAs were 0.74 and 0.64, respectively. These coefficients grossly indicate the presence of internal consistencies across the scores.

Median and range of the scores were computed and high score was defined if it exceeds the median values. Thus, the scores were dichotomized as high and not-high to serve as outcome variables in multivariate logistic regression analysis. The median scores as well as the proportion with high scores were compared across the 3 levels of exposure to the MARCH intervention, as shown in Table 8. Respondents who were exposed to the MARCH intervention were significantly more likely than those who did not to exhibit favorable and positive attitudes towards PLWHAs. The results indicated that the proportion who scored above the median (i.e. high positive attitude towards PLWHAs) ranged from 47.2% among the non-exposed to 63.9% and 67.6% among the moderately-and highly-exposed respondents, respectively (p<0.01). The observed gross effect of exposure to the MARCH on respondents’ attitude towards PLWHAs did not alter after controlling for selected socio-demographic characteristics (Table 9). The odds of scoring high positive attitude towards PLWHAs was 1.8 and 2.1 times higher, respectively, among the moderately-and highly-exposed respondents, as compared to the non-exposed.

There was a modest association between exposure to the MARCH and respondents’ perception of non-differential treatment of PLWHAs. Of note, the distribution of the scores for this particular variable was highly skewed with most respondents across the three exposure categories scoring at or around a value of 3. As a result, the comparison of the median values across the groups could not reflect the true

Addis Ababa and West Hararghe