When I performed path analysis with Lisrel and Amos, the significance of parameters can be determined by t-test. But, after I performed that in Stata, the significance level determined with z-test. How do I calculate the significance level parameters in path analysis with z-test? Is there any manual way/formula to calculate it? Is there any reference book?
Thank You
Hello Dedy,
They're the same form of test: result = parameter (path coefficient) estimate divided by standard error. Carl Joreskog (co-author of LISREL) referred to these as z-tests; others call them t-tests. To answer your question, if you presume a z-statistic, the corresponding non-directional ("two-tail") probability of z>z(observed) is given by:
Excel: 2*(1 - NORM.S.DIST(z, TRUE)) where "z" is the z-statistic
Stata: zprob = 2*(1 - normal(z)) where "z" is the z-statistic
Good luck with your study!
Oops, left off stata's internal command:
In Stata: generate zprob = 2*(1 - normal(z))
list zprob
But, you probably knew that!
t test is the same score for n=120, so that if you have sample more than 120 is approximate the same as z=t. but if you applied SEM PLS is recommended used bootstrapping methods for n=500. go ahead to apply SmartPls 3 is more powerful. Yoiu can used formative construct as model extension in your researach model. Hair et al (2016) and Ringle et al (2015) will guidance you to find best fits of the results. SEM PLS study more focus on measurement steps rather than complicated relationship that might be only the game because of difficulty in statistical assumption of SEM. SM PLS is guarantee more happier ..
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude