Table 3

Final hierarchical model of factors affecting physical activity.

Factor

Type I P-value*

Type III P-value**

BMI

0.0273

0.0299

BMI²

0.0089

0.0136

Gender

0.5105

0.7973

Weekday

<0.0001

<0.0001

Month

<0.0001

0.0125

Weekday*Month

<0.0001

0.0002

Days On Study

0.0123

0.8453

Days On Study²

0.0442

0.0348

Days On Study³

0.0169

0.0056

Total Rain

<0.0001

<0.0001

Total Rain²

0.0019

0.0029

Mean Temperature

0.0002

0.0237

Total Snow

0.0541

0.0243

Total Snow²

0.4818

0.0394

Total Snow* Gender

0.0132

0.0129

BMI * Total Snow

0.5550

0.0227

BMI* Total Snow²

0.0375

0.0255

Maximum Wind Gust

0.0003

0.1455

Maximum Wind Gust²

0.0620

0.0211

BMI * MaximumWind Gust

0.9219

0.1089

BMI * MaximumWind Gust²

0.2046

0.0175

BMI ² * MaximumWind Gust

0.6069

0.0825

BMI ² * MaximumWind Gust²

0.0094

0.0112

Snow On Ground

0.0124

0.0124

* The Type I p-value is associated with sequential tests of hypothesis, thus order of the factor in the above table is important (see example below).

** The Type III p-value is associated with "last in" tests of hypothesis (see example below).

For example, consider the p-values for mean temperature. The Type I p-value measures the significance of the effect for mean temperature after adjusting for those effects in the table listed above mean temperature (BMI, Gender, Week, Month, Day on Study, Total Rain and the various interactions), but not those below mean temperature (Total Snow, Snow on Ground, and the various interactions). In contrast, the Type III p-value tests for the significance of the effect if it was the "last in" the model, i.e., adjusting for all factors in the table (BMI, Gender, Week, Month, Day on Study, Total Rain, Total Snow, Snow on Ground and the various interactions).

Chan et al. International Journal of Behavioral Nutrition and Physical Activity 2006 3:21   doi:10.1186/1479-5868-3-21