For Step 2, the coefficients estimated from regression analyses were used to calculate the expected biomass at harvest for each pond and this was compared with the corresponding observed yield in the database. After this, a simple linear regression analysis between observed and expected shrimp biomass was performed, setting the intercept at zero. After determining that the regression slope did not differ significantly from 1 (t-test), an equivalence test ( Chow and Liu, 2004) was conducted to protect against falsely not rejecting the corresponding null hypothesis (Type-II statistical error; Zar, 2010 and Hauck and Anderson, 1986). The residuals resulting from this simple linear regression were tested for normality, using the Shapiro–Wilk W-test. For testing equivalence, tolerance error was set at 5% ( Garret, 1997).