Skip to main content

Table 5 Goodness-of-fit for multiple-group confirmatory factor analysis§: exploratory vs. validation samples†

From: Assessing urban and rural neighborhood characteristics using audit and GIS data: derivation and reliability of constructs

 

Urban (n = 7,660)

Rural (n = 3,110)

Model

# parameters

CFI > 0.95

TLI > 0.95

RMSEA < 0.05

Difference test‡

# parameters

CFI > 0.95

TLI > 0.95

RMSEA < 0.05

Difference test‡

1-factor

          

Arterial or thoroughfare

34

0.956

0.958

0.003

0.648

33

1.000

1.000

<0.001

0.972

Walkable neighborhood

32

0.953

0.953

0.002

0.743

31

0.888

0.888

0.003

0.549

Physical incivilities

21

0.983

0.984

0.001

0.669

22

0.985

0.984

0.003

0.855

Decoration

9

1.000

1.000

<0.001

0.852

9

1.000

1.000

<0.001

0.809

  1. § Data are weighted by the road segment's length.
  2. † In urban areas: 7,660 road segments (4,537 and 3,123 in exploratory and validation subsamples). In rural areas: 3,110 road segments (1,851 and 1,259 in exploratory and validation samples).
  3. ‡ Ho: Invariance (i.e. all factor loadings, thresholds and correlated errors are the same) H1: Non invariance (i.e. different factor loadings, thresholds and correlated errors by sample).
  4. Abbreviations: RMSEA root mean square error of approximation, CFI comparative fit index, and TLI Tucker-Lewis index
  5. *These are "rules of thumb" guidelines to interpret these indices (see paper for more information).
Back to article page

International Journal of Behavioral Nutrition and Physical Activity

ISSN: 1479-5868

Contact us