In this paper an attempt was made to s tudy sampling errorsrnfrom socioeconomic variables of three different areas of Gondarrnreg ion under three different sampling techniques . On the basisrnof the variances (standard errors) of each of the variables,rndesign effects and rate of homogeneity values were computed.rnDepending on these values of sampling errors the accuracy andrnefficiency achieved in different sample designs on qualitativernand quantitative socioeconomic variables were compared. Lookingrnat the pattern of the sampling errors , out of the threerntechniques stratified random sampling was f ound to be highlyrnsuperior for both discrete and continuous var i ables , in all ofrnthe three a reas. On the other hand these comput ed values ofrnsampling errors s howed that for continuous variables , thernperformance of Cluster sampling was much better than that ofrnSimple random sampling.