A simulated dataset for longitudinal measurement invariance: the same four Wellbeing items answered by the same people on three occasions.
Format
A data frame with 500 rows and 12 columns, w1_t1 to w4_t3: items
w1 to w4 at occasions t1, t2, and t3. Population standardized
loadings at t1 are 0.80, 0.75, 0.70, and 0.65, and the loadings are equal
across occasions.
Details
Wellbeing has a latent variance of 1 at every occasion. Its mean rises from
0 at t1 to 0.30 at t2 and 0.50 at t3, in t1 standard deviations, and
it correlates .60 between adjacent occasions and .45 between t1 and t3.
Longitudinal teaching features: each item's unique factor correlates .20
with the same item's unique factor on the other occasions, and the intercept
of w3 is 0.40 higher at t2 and t3 than at t1. The w3 intercept is
therefore a known source of scalar non-invariance over time, and holding it
equal inflates the apparent change in Wellbeing. Indicators are on a
continuous rating metric (population mean 4 at t1) and rounded to two
decimals.
Examples
str(nomo_demo_longitudinal)
#> 'data.frame': 500 obs. of 12 variables:
#> $ w1_t1: num 3.61 2.59 3.59 4.81 5.69 4.68 4.97 3.79 4.16 3.46 ...
#> $ w2_t1: num 2.97 3.54 4.23 3.63 5.22 4.77 4.47 5.08 5 1.99 ...
#> $ w3_t1: num 4.65 2.62 2.64 4.54 4.95 4.67 5.74 3.79 3.93 4.13 ...
#> $ w4_t1: num 1.86 3.59 3.35 5.79 4.09 5.37 5.39 4.92 4.71 2.32 ...
#> $ w1_t2: num 3.81 4.03 4.5 4.95 5.33 3.46 5.8 6.04 3.92 3.88 ...
#> $ w2_t2: num 4.64 4.71 3.92 4.71 5.23 3.6 4.13 4.58 4.9 2.93 ...
#> $ w3_t2: num 3.34 3.26 4.61 5.48 5.64 3.32 5.07 4.38 4.83 3.32 ...
#> $ w4_t2: num 3.79 2.84 3.78 5.56 4.59 4.2 3.36 4.11 2.72 3.3 ...
#> $ w1_t3: num 3.21 5.43 3.15 6 5.53 4.54 4.9 5.31 5.16 4.21 ...
#> $ w2_t3: num 4.19 5.68 4.78 7.13 5.12 4.99 5.23 4.61 4.93 2.98 ...
#> $ w3_t3: num 4.35 5.8 4.26 6.98 5.69 3.81 4.53 4.46 5.82 3.41 ...
#> $ w4_t3: num 4.6 5.11 3.55 6.73 4.09 3.64 3.6 4.36 4.74 2.95 ...
colMeans(nomo_demo_longitudinal[c("w3_t1", "w3_t2", "w3_t3")])
#> w3_t1 w3_t2 w3_t3
#> 3.98258 4.53520 4.72832