Table 2.

Linear versus quadratic modeling of ordinal and visual analog scale ratings of residue as described by Stevens (1975).

	Bolus Type	Equation (p-value)	r2	Figure
Linear Modeling	Thin Liquid	y= −22.64 + 17.48x p<0.0001	0.85
	Applesauce	y= −32.51 + 22.97x p<0.0001	0.93
	Cracker	y= −32.35 + 21.11x p<0.0001	0.91
Quadratic Modeling†	Thin Liquid	y= 14.32 −17.80x + 6.73x2 p<0.0001	0.98*
	Applesauce	y= 8.32 −12.33x + 5.88x2 p<0.0001	0.99*
	Cracker	y= 10.20 −13.94x + 6.03x2 p<0.0001	0.98*

^*Significantly different than respective r² values of the same bolus type’s linear model r² value (p<0.0001).

^†The equations listed here were done on transformed data in accordance with the described psychometric testing. The descriptive text lists other quadratic equations that can be used without transforming data.