TPC-Journal-V1-Issue3

212 The Professional Counselor \ Volume 1, Issue 1 Next, these 100 items were then subjected to an item analysis that resulted in identifying 45 items with item discrimination indices of 0.50 or higher. These 45 items were then further subjected to a factor analysis and an alpha coefficient. An arbitrary decision was then made to use a 0.5 or higher factor loading in examining items. A strict convention of 0.5 or higher was used in order to identify the most discriminating items. The factor analysis identified the same items the item analysis identified. The alpha coefficients were as follows: overall= 0.924; straights= 0.723, gays= 0.653. The factor analysis also identified four factors that were consistent with issues identified by both the literature and the initial 100 items. After reviewing the items in these factors, they were labeled as: 1) attraction to same sex 2) attraction to opposite sex 3) self-acceptance of gay behavior/attitudes 4) fears Item Analysis Anastasi (1988) pointed out that items on an instrument may be analyzed qualitatively, in terms of their content and form, and quantitatively, in terms of their statistical properties. The item analysis performed on the initial 100 items of the Sexual Orientation Scale focused on a quantitative analysis and more specifically on the measures of item difficulty and item discrimination. Item difficulty refers to the percentage of subjects that endorse certain items on the scale. The closer the difficulty level approaches 0.50, the more differentiations the items can make (Anastasi, 1988). Item discrimination refers to how effective the item discriminates between the two groups. Therefore, the higher the item discrimination score, the more effectively the item will differentiate between the two groups (gays/straights). Table 1 shows how the sample was grouped in order to establish an item-to-total score correlation, which is identified as a useful exercise to select items. Table 1 Statistics for Grouping the Sample Quintile Sample Size Proportion Score Range 1 st 42 0.20 87–95 2 nd 42 0.20 72–86 3 rd 43 0.21 51–71 4 th 39 0.19 48–50 5 th 42 0.20 36–47 Based on the total score, the respondents were divided into quintiles (groups of approximately 40 subjects). A total score was established by assigning a value of 1 to true responses and a value of 2 to false responses. A true response indicated the way a gay man would respond. An item difficulty, identified in the item analysis as proportion of subjects that responded correctly to the item (PROP) and item discrimination, identified in the item analysis as a point biserial correlation coefficient (RPBI), were calculated for each item. Table 2 outlines the item difficulty and item discrimination score and the scoring key of the initial 100 items. The asterisks identify the scores for the 43 items on the final version of the Sexual Orientation Scale. Of the final 43 items, approximately 76.7% of the items have a difficulty score that range from 0.40 to 0.60. Since Anastasi (1988) stated that the closer the difficulty level approaches 0.50, the more differentiations the item can make, it is safe to infer that the