Skip to first unread message
Wisarut Srisintorn
Apr 16, 2019, 12:09:39 AM4/16/19
to lavaan
Greeting,
I tried to fit a one-factor CFA model with 21 ordinal items (grading 1 - 5). There was 900 subjects in my dataset.
My model was:
I tried to fit a one-factor CFA model with 21 ordinal items (grading 1 - 5). There was 900 subjects in my dataset.
My model was:
ws_model <- 'ws =~ ws1 + ws2 + ws3 + ws4 + ws5 + ws6 + ws7 + ws8 + ws9 + ws10 + ws11 + ws12 + ws13 + ws14 + ws15 + ws16 + ws17 + ws18 + ws19 + ws20 + ws21'
And the command and error message were:
> ws_fit <- cfa(ws_model, data = df, ordered = c('ws1','ws2','ws3','ws4','ws5','ws6','ws7','ws8','ws9','ws10','ws11','ws12','ws13','ws14','ws15','ws16','ws17','ws18','ws19','ws20','ws21'))
Warning message:
In pc_cor_TS(fit.y1 = UNI[[i]], fit.y2 = UNI[[j]], method = optim.method, :
no convergence
After searching in google and self trouble-shooting, I found that the error occurred from lavCor() where:
> lavCor(df %>% select(ws1:ws21) %>% map(ordered) %>% bind_cols())
......(long result omitted)...........
Warning message:
In pc_cor_TS(fit.y1 = UNI[[i]], fit.y2 = UNI[[j]], method = optim.method, :
no convergence
Therefore, I would like to ask if this warning (no convergence) pointed to any problem in my data?
Best regards,
Wisarut
Best regards,
Wisarut
Terrence Jorgensen
Apr 16, 2019, 5:15:56 AM4/16/19
to lavaan
I would like to ask if this warning (no convergence) pointed to any problem in my data?
Not sure if the data are a problem per se (perhaps if you have rarely endorsed categories, check 1-way tables in lavTables() output). But this is certainly a lot of 5-category variables for "only" N=900 (there are 5^21 = 4.768372e+14 possible response categories, making estimation of polychoric correlations a bit more difficult). First, check if this still happens in the development version, in case it was an issue that was already fixed:
install.packages("lavaan", repos = "http://www.da.ugent.be", type = "source")
If so, consider the distributions of responses across categories in lavTables() output. If thresholds are approximately symmetric and there are not many rarely endorsed categories (e.g., proportions < 5% of the sample) in 1-way tables, then you would obtain similar results using a robust ML estimator by treating the data as continuous.
Terrence D. Jorgensen
Assistant Professor, Methods and Statistics
Research Institute for Child Development and Education, the University of Amsterdam
Terrence Jorgensen
Apr 16, 2019, 5:17:08 AM4/16/19
to lavaan
Warning message:In pc_cor_TS(fit.y1 = UNI[[i]], fit.y2 = UNI[[j]], method = optim.method, :no convergence
I just noticed this was a warning, not an error. What is returned by summary(ws_fit)?
Wisarut Srisintorn
Apr 16, 2019, 7:00:52 AM4/16/19
to lavaan
Greeting Dr.Jorgensen,
The same warning code still occurred after update lavaan package to latest version.
This is the result of summary(ws_fit). It is very long table. So I does not include the intercept (all 0.000), and scale y* (all 1.000)
The same warning code still occurred after update lavaan package to latest version.
lavTables(ws_fit) showed a lots of observed frequency with less than 10.... Might this be the cause?
This is the result of summary(ws_fit). It is very long table. So I does not include the intercept (all 0.000), and scale y* (all 1.000)
lavaan 0.6-4.1376 ended normally after 47 iterations
Optimization method NLMINB
Number of free parameters 105
Used Total
Number of observations 869 915
Estimator DWLS Robust
Model Fit Test Statistic 7075.575 6324.195
Degrees of freedom 189 189
P-value (Chi-square) 0.000 0.000
Scaling correction factor 1.138
Shift parameter 107.640
for simple second-order correction (Mplus variant)
Parameter Estimates:
Information Expected
Information saturated (h1) model Unstructured
Standard Errors Robust.sem
Latent Variables:
Estimate Std.Err z-value P(>|z|)
ws =~
ws1 1.000
ws2 0.994 0.027 36.328 0.000
ws3 1.351 0.038 35.683 0.000
ws4 1.431 0.041 35.095 0.000
ws5 1.394 0.039 35.511 0.000
ws6 1.424 0.039 36.795 0.000
ws7 1.183 0.036 33.041 0.000
ws8 0.929 0.040 23.312 0.000
ws9 0.784 0.041 19.235 0.000
ws10 1.292 0.040 32.470 0.000
ws11 1.330 0.038 34.640 0.000
ws12 1.042 0.034 30.403 0.000
ws13 1.232 0.036 34.022 0.000
ws14 1.226 0.035 34.718 0.000
ws15 1.360 0.041 33.555 0.000
ws16 1.352 0.040 33.727 0.000
ws17 0.859 0.036 24.015 0.000
ws18 0.996 0.036 27.316 0.000
ws19 0.957 0.039 24.545 0.000
ws20 1.057 0.040 26.460 0.000
ws21 1.137 0.038 29.994 0.000
Thresholds:
Estimate Std.Err z-value P(>|z|)
ws1|t1 -1.328 0.059 -22.351 0.000
ws1|t2 -0.816 0.048 -16.967 0.000
ws1|t3 -0.543 0.045 -12.101 0.000
ws1|t4 0.654 0.046 14.209 0.000
ws2|t1 -0.883 0.049 -17.959 0.000
ws2|t2 -0.187 0.043 -4.371 0.000
ws2|t3 0.146 0.043 3.423 0.001
ws2|t4 1.079 0.053 20.398 0.000
ws3|t1 -1.509 0.066 -22.937 0.000
ws3|t2 -0.913 0.050 -18.384 0.000
ws3|t3 -0.279 0.043 -6.467 0.000
ws3|t4 0.845 0.049 17.404 0.000
ws4|t1 -1.722 0.076 -22.772 0.000
ws4|t2 -1.121 0.054 -20.822 0.000
ws4|t3 -0.468 0.044 -10.569 0.000
ws4|t4 0.883 0.049 17.959 0.000
ws5|t1 -1.833 0.082 -22.346 0.000
ws5|t2 -1.294 0.058 -22.163 0.000
ws5|t3 -0.654 0.046 -14.209 0.000
ws5|t4 0.862 0.049 17.652 0.000
ws6|t1 -1.833 0.082 -22.346 0.000
ws6|t2 -1.379 0.061 -22.584 0.000
ws6|t3 -0.591 0.045 -13.027 0.000
ws6|t4 0.716 0.047 15.312 0.000
ws7|t1 -2.357 0.131 -18.029 0.000
ws7|t2 -1.566 0.068 -22.980 0.000
ws7|t3 -0.878 0.049 -17.898 0.000
ws7|t4 0.754 0.047 15.954 0.000
ws8|t1 -2.039 0.097 -21.040 0.000
ws8|t2 -0.944 0.050 -18.803 0.000
ws8|t3 -0.433 0.044 -9.833 0.000
ws8|t4 1.126 0.054 20.873 0.000
ws9|t1 -1.547 0.067 -22.972 0.000
ws9|t2 -0.321 0.043 -7.412 0.000
ws9|t3 0.129 0.043 3.017 0.003
ws9|t4 1.528 0.067 22.958 0.000
ws10|t1 -2.063 0.099 -20.853 0.000
ws10|t2 -1.425 0.063 -22.750 0.000
ws10|t3 -0.540 0.045 -12.034 0.000
ws10|t4 1.079 0.053 20.398 0.000
ws11|t1 -2.171 0.109 -19.909 0.000
ws11|t2 -1.685 0.074 -22.863 0.000
ws11|t3 -0.866 0.049 -17.714 0.000
ws11|t4 0.812 0.048 16.905 0.000
ws12|t1 -1.236 0.057 -21.787 0.000
ws12|t2 -0.574 0.045 -12.697 0.000
ws12|t3 -0.013 0.043 -0.305 0.760
ws12|t4 1.171 0.055 21.275 0.000
ws13|t1 -1.849 0.083 -22.269 0.000
ws13|t2 -1.314 0.059 -22.278 0.000
ws13|t3 -0.650 0.046 -14.143 0.000
ws13|t4 0.939 0.050 18.744 0.000
ws14|t1 -2.017 0.095 -21.213 0.000
ws14|t2 -1.342 0.060 -22.422 0.000
ws14|t3 -0.567 0.045 -12.565 0.000
ws14|t4 1.084 0.053 20.452 0.000
ws15|t1 -2.313 0.125 -18.498 0.000
ws15|t2 -1.500 0.065 -22.924 0.000
ws15|t3 -0.650 0.046 -14.143 0.000
ws15|t4 1.043 0.052 20.014 0.000
ws16|t1 -2.313 0.125 -18.498 0.000
ws16|t2 -1.401 0.062 -22.671 0.000
ws16|t3 -0.543 0.045 -12.101 0.000
ws16|t4 1.084 0.053 20.452 0.000
ws17|t1 -1.617 0.070 -22.963 0.000
ws17|t2 -0.944 0.050 -18.803 0.000
ws17|t3 -0.364 0.044 -8.355 0.000
ws17|t4 1.038 0.052 19.958 0.000
ws18|t1 -1.974 0.092 -21.518 0.000
ws18|t2 -1.143 0.054 -21.026 0.000
ws18|t3 -0.199 0.043 -4.642 0.000
ws18|t4 1.274 0.058 22.043 0.000
ws19|t1 -2.203 0.112 -19.612 0.000
ws19|t2 -1.509 0.066 -22.937 0.000
ws19|t3 -0.643 0.046 -14.012 0.000
ws19|t4 1.014 0.052 19.676 0.000
ws20|t1 -2.171 0.109 -19.909 0.000
ws20|t2 -1.425 0.063 -22.750 0.000
ws20|t3 -0.557 0.045 -12.366 0.000
ws20|t4 0.981 0.051 19.274 0.000
ws21|t1 -2.237 0.116 -19.281 0.000
ws21|t2 -1.596 0.069 -22.976 0.000
ws21|t3 -0.829 0.048 -17.155 0.000
ws21|t4 0.781 0.048 16.400 0.000
Variances:
Estimate Std.Err z-value P(>|z|)
.ws1 0.599
.ws2 0.604
.ws3 0.268
.ws4 0.179
.ws5 0.222
.ws6 0.188
.ws7 0.439
.ws8 0.654
.ws9 0.754
.ws10 0.331
.ws11 0.291
.ws12 0.565
.ws13 0.392
.ws14 0.398
.ws15 0.258
.ws16 0.268
.ws17 0.704
.ws18 0.602
.ws19 0.633
.ws20 0.552
.ws21 0.482
ws 0.401 0.024 16.823 0.000
Reply all
Reply to author
Forward
0 new messages