Table IV.
Assessment of Multimodality of F2 Distributions for Solution Intakea
| Solution | Most likely number of distributions |
χ 2 |
|---|---|---|
| 75 mM NaCl | 2 vs. 1 | 19.4** |
| 3 vs. 2 | 10.2** | |
| 4% sucrose | 2 vs. 1 | 36.4** |
| 3 vs. 2 | 21.1** | |
| 0.1 mM citric acid | 2 vs. 1 | NS |
| 3 vs. 2 | NS | |
| 10% ethanol | 2 vs. 1 | 4.2* |
| 3 vs. 2 | 19.2** | |
| 0.03 mM quinine HCI | 2 vs. 1 | 16.5** |
| 3 vs. 2 | NS |
Hypothesis testing was carried out by comparing alternative (nested) models using a likelihood-ratio test. Twice the difference in the log likelihoods between models is distributed approximately as χ2, with the degrees of freedom equal to the difference between models in the number of free parameters. Because some tests of support for additional distributions involve constraints at boundaries rather than at interior regions of the parameter space, the assumed χ2 may hold only approximately.
p < .05.
p < .01.