Our Methodology

The Numbers Behind the Stars

Better Buying uses a 0 to 100 point scoring system to calculate overall scores and scores for each category with a larger number indicating better purchasing practices. After scoring each question and category of purchasing practices with a proprietary scoring methodology, the star “grading” is applied to the scores as follows (see Table 1).
TABLE 1: Stars and Corresponding Numerical Scores

The same “grading standards” are applied to each category score and to the overall weighted score. Better Buying uses the weighting system outlined below to determine the weight of each purchasing practices category to the overall score.

TABLE 2: Weights of Each Purchasing Practice Category

Basic descriptive statistical analysis is conducted for the scores and the responses to each question. Means for the purchasing practice categories are also based on scales from 0 to 100. In all cases, smaller means reflect poorer purchasing practices while larger means reflect better purchasing practices. Standard deviation (SD) reflects the variability of scores around the mean; in this case how widely spread the buyer ratings were. A larger SD means there is quite a range of scores that have been reported about buyer purchasing practices.

Tests of differences between a buyer’s score and the industry benchmark (the average of all submitted ratings) are conducted with an independent sample t-test using the benchmark (the average score for all submitted ratings) to determine if there was a significant difference between the rated buyer’s scores and the benchmark. A p-value of .05 is used. P-values indicate the chance that the differences observed in data are likely due to chance. A p-value of .05 means that there is less than or equal to a 5% chance that differences reported as “significantly different” would not hold if data from all suppliers was available. Because the rated buyer’s scores are also included in the benchmark and the number of submitted ratings is relatively small, the likelihood of finding a significant difference is reduced.