Many thesis students who are using the survey as their data collection procedure, will often develop attitude, belief or perception statements on the topic of their thesis, such as mothers’ attitudes toward prenatal care, college students’ female beauty ideal, students’ perceptions of the credibility of TV ads on alcoholic drinks, etc. Developing attitude statements shouldn’t end there. After the survey has been done and the attitude data encoded in a spreadsheet and uploaded in SPSS, a key step is to calculate the reliability score of the scale.
In surveys, scales are often used to measure respondent beliefs, perceptions and attitudes. Results are then used to make inferences and judgments on intervention points that need to be addressed in subsequent upscaling initiatives. When a scale is used, it is a standard procedure in the social sciences to determine its reliability. A popular measure of reliability is Cronbach’s alpha which determines the internal consistency or average correlation of items in a survey instrument to gauge its reliability. Internal consistency estimates how consistently the respondents have responded to the items within the scale. The closer the Cronbach’s alpha is to 1, the higher the internal consistency (Gliem & Gliem, 2003). In social science, the widely accepted alpha is 0.70 or higher for a set of items to be considered a scale.
We calculated the Cronbach’s alpha of the belief and attitude scale on non-rice habitats used in a baseline survey instrument. As items were worded either positively or negatively, reverse scoring was done to negatively worded statements. The items within the scale were drawn from focus group discussions (FGDs) with rice farmers in China, Thailand and Vietnam. We used a Likert-type scale, as follows: Definitely not true, in most cases not true, may be true, in most cases true and always true. The attitude scale on ecological engineering included these items:
Uses of non-rice habitat
Maintaining non-rice habitat
Cronbach’s alpha from survey data
Using SPSS, reliability of the attitude scale was computed. Table 1 shows that the reliability tests performed on the belief and attitude data generated an acceptable Cronbach’s alpha of higher than 0.70, with relatively higher alpha for data from Vietnam farmers. This suggests that the items in the attitude scale are related enough to combine into an attitude scale or index. For each data set, only 16 to 22 out of 25 items were used in the analysis. Some items were excluded in the computation to increase the reliability coefficient.
Table 1. Cronbach’s alpha obtained from attitude scale data in target countries.
|Country||Cronbach’s alpha||Number of items|
|China (Jinhua and Lingui)||0.713||16|