I would be really grateful if anyone could help with the following questions:
1. Is there a way to calculated squared correlations on STATA after performing a correlation command? This is for the Fornell and Lacker’s test. For instance, I used this correlation command corr factor1 factor2 factor3 factor4
and then I found I could run correlations with 2 of the constructs in each corr command , e.g., corr factor1 factor2 display r(rho)^2
I was wondering if you are aware of a command that squares all the corr coefficients in one go.
Also, is there a way to get the significance level? SPSS provides significance at 0.01(**) and 0.05(*) but I cannot find a way to get this properly on STATA. I tried these
pwcorr factor1 factor2 factor3 factor4 , sig star(.05) and pwcorr factor1 factor2 factor3 factor4, sig star(.01) but the * sign on the correlation coefficient values is a single asterisk and does not separate them clearly on significance level.
2. When do we set an error term 0.10 please? Is it only when we have a single item construct or a higher-order construct? And to do this on STATA is the command
sem ([email protected] -> var1 var2 var3)
Also, is error term the same as the initial values please?e.g., sem (VAR -> [email protected] [email protected] [email protected] [email protected])
I am a bit confused about the difference between the two and cannot recall my notes.
3. When do we add standardize iter(1000) method(mlmv) nolog to the sem command? Would it be appropriate to run the following for instance (all first order constructs)?
sem (VAR -> var1 var2 var3) ///
(Factor -> fac1 fac2 fac4r fac5) ///
(MA -> MA1 MA2 MA3 MA4 MA5 MA6) ///
(ID -> ID1 ID2 ID3 IDf4 ID5 ID6 ID7), standardize iter(1000) method(mlmv) nolog
4. Following question 3 when I remove standardize iter(1000) method(mlmv) nolog command from the sem I get different results.
Results when standardize iter(1000) method(mlmv) nolog is used
Standardized | Coefficient std. err. z P>|z| [95% conf. interval]
----------------+----------------------------------------------------------------
Measurement |
aeth1 |
AEth| .7100429 .0347624 20.43 0.000 .6419098 .778176
_cons | 4.754638 .1973182 24.10 0.000 4.367902 5.141375
--------------+----------------------------------------------------------------
aeth2 |
AEth | .8203105 .0282439 29.04 0.000 .7649535 .8756676
_cons | 4.899959 .2028655 24.15 0.000 4.50235 5.297568
--------------+----------------------------------------------------------------
aeth3 |
AEth | .7935696 .0294832 26.92 0.000 .7357837 .8513556
_cons | 4.37813 .1830129 23.92 0.000 4.019432 4.736829
--------------+----------------------------------------------------------------
Results when standardize iter(1000) method(mlmv) nolog is not used
OIM
Coefficient std. err. z P>z [95% conf. interval]
Measurement
aeth1
AEth1 (constrained)
_cons 5.281646 .0624897 84.52 0.000 5.159168 5.404123
aeth2
AEth 1.17141 .0961538 12.18 0.000 .9829519 1.359868
_cons 5.518987 .0633612 87.10 0.000 5.394802 5.643173
aeth3
AEth 1.223204 .0992947 12.32 0.000 1.02859 1.417818
_cons 5.322785 .0683922 77.83 0.000 5.188739 5.456831
5. Following question 4 I cannot recall which signs reflect standardised loadings/st. lamda, t-values? Should I refer to the command including standardize please?
If so, let’s see aeth1. Is is correct to use standardised loadings/st. lamda = .7100429, stand error= .0347624 and t-value 20.43 from the first table please?
6. I have very high t-values, some of them reaching 69. Is there a different command I should add please? I tried to set the 1st item of each construct at 1 but then the stand.loading/coeeficient is 1. Should I fix the item to set the scale and how should I do that please?
Thank you very much for your time,
Nana
1. Is there a way to calculated squared correlations on STATA after performing a correlation command? This is for the Fornell and Lacker’s test. For instance, I used this correlation command corr factor1 factor2 factor3 factor4
and then I found I could run correlations with 2 of the constructs in each corr command , e.g., corr factor1 factor2 display r(rho)^2
I was wondering if you are aware of a command that squares all the corr coefficients in one go.
Also, is there a way to get the significance level? SPSS provides significance at 0.01(**) and 0.05(*) but I cannot find a way to get this properly on STATA. I tried these
pwcorr factor1 factor2 factor3 factor4 , sig star(.05) and pwcorr factor1 factor2 factor3 factor4, sig star(.01) but the * sign on the correlation coefficient values is a single asterisk and does not separate them clearly on significance level.
2. When do we set an error term 0.10 please? Is it only when we have a single item construct or a higher-order construct? And to do this on STATA is the command
sem ([email protected] -> var1 var2 var3)
Also, is error term the same as the initial values please?e.g., sem (VAR -> [email protected] [email protected] [email protected] [email protected])
I am a bit confused about the difference between the two and cannot recall my notes.
3. When do we add standardize iter(1000) method(mlmv) nolog to the sem command? Would it be appropriate to run the following for instance (all first order constructs)?
sem (VAR -> var1 var2 var3) ///
(Factor -> fac1 fac2 fac4r fac5) ///
(MA -> MA1 MA2 MA3 MA4 MA5 MA6) ///
(ID -> ID1 ID2 ID3 IDf4 ID5 ID6 ID7), standardize iter(1000) method(mlmv) nolog
4. Following question 3 when I remove standardize iter(1000) method(mlmv) nolog command from the sem I get different results.
Results when standardize iter(1000) method(mlmv) nolog is used
Standardized | Coefficient std. err. z P>|z| [95% conf. interval]
----------------+----------------------------------------------------------------
Measurement |
aeth1 |
AEth| .7100429 .0347624 20.43 0.000 .6419098 .778176
_cons | 4.754638 .1973182 24.10 0.000 4.367902 5.141375
--------------+----------------------------------------------------------------
aeth2 |
AEth | .8203105 .0282439 29.04 0.000 .7649535 .8756676
_cons | 4.899959 .2028655 24.15 0.000 4.50235 5.297568
--------------+----------------------------------------------------------------
aeth3 |
AEth | .7935696 .0294832 26.92 0.000 .7357837 .8513556
_cons | 4.37813 .1830129 23.92 0.000 4.019432 4.736829
--------------+----------------------------------------------------------------
Results when standardize iter(1000) method(mlmv) nolog is not used
OIM
Coefficient std. err. z P>z [95% conf. interval]
Measurement
aeth1
AEth1 (constrained)
_cons 5.281646 .0624897 84.52 0.000 5.159168 5.404123
aeth2
AEth 1.17141 .0961538 12.18 0.000 .9829519 1.359868
_cons 5.518987 .0633612 87.10 0.000 5.394802 5.643173
aeth3
AEth 1.223204 .0992947 12.32 0.000 1.02859 1.417818
_cons 5.322785 .0683922 77.83 0.000 5.188739 5.456831
5. Following question 4 I cannot recall which signs reflect standardised loadings/st. lamda, t-values? Should I refer to the command including standardize please?
If so, let’s see aeth1. Is is correct to use standardised loadings/st. lamda = .7100429, stand error= .0347624 and t-value 20.43 from the first table please?
6. I have very high t-values, some of them reaching 69. Is there a different command I should add please? I tried to set the 1st item of each construct at 1 but then the stand.loading/coeeficient is 1. Should I fix the item to set the scale and how should I do that please?
Thank you very much for your time,
Nana
Comment