Table 6

Comparison of path models.

Maximum Likelihood Analyses

Model

Path constrained to zero

χ2

df

P

AIC

Δi

wi

Evidence Ratio


1

Head length → Fitness

5.38

7

0.61

61.38

0

0.28

2

Max. swim speed → Fitness

1.98

5

0.85

61.98

0.60

0.21

1.35

3

Head ht. → Fitness

4.28

6

0.64

62.28

0.89

0.18

1.56

4

Tail muscle ht. → Max. swim speed

8.39

8

0.40

62.39

1.00

0.17

1.65

5

Tail fin ht. → Max. swim speed

1.58

4

0.81

63.58

2.19

0.09

3.00

6

Tail length → Max. swim speed

1.20

3

0.75

65.20

3.81

0.04

6.73

7

Tail muscle ht. → Fitness

0.89

2

0.64

66.89

5.51

0.02

15.70

8

Head ht. → Max. swim speed

0.36

1

0.55

68.36

6.98

0.01

32.77

9

Saturated model

--

0

--

70.00

8.62

0.00

74.33


Chi-square values (χ2), degrees of freedom (df), and the associated P value report the significance of the model. Note that model 9 is saturated, and thus the fit of the model to the data could not be tested using the chi-square statistic. In models that are not significant, the data are a good fit to the model. AIC = Akaike Information Criterion; ΔI = difference in AIC scores between the best model and subsequent model; wi = normalized relative likelihood of the model given the data; Evidence Ratio = the relative odds that a model is the best given the data.

Calsbeek and Kuchta BMC Evolutionary Biology 2011 11:353   doi:10.1186/1471-2148-11-353

Open Data