Hi,

Your question:

What's the minimum sample size to conduct the SEM analysis utilizing AMOS software?

For conduct the Structural equation model analysis using AMOS software, mininum 100 samples were needed.  Generally, SEM undergoes five steps of model specification, identification, estimation, evaluation, and modifications (possibly).  Using the above mentioned 5 steps you should do the SEM analysis by AMOS.

thans.

I am afraid there is no way to answer this question without further information. Technically you will need more cases than variables. But anything beyond that will depend on the number of variables, their standard deviations/errors, the aim of the analysis (degree of generalisation) ...

As noted, more information is needed. For SEM designs, then the number of variables in total and number of indicators is needed. For SEM designs (e.g. using AMOS), I tell my students that 100 is the absolute minimum, though 200+ is preferred. If using PLS (e.g. SmartPLS), then it is 10 observations per arrow to a construct is the minimum, and whilst PLS can work with small samples, larger samples allow a greater ability to detect smaller path coefficients as being significant. As they say, "the more the merrier"!

Hi,

one of the most central paper in this regard is

Hoogland, J. J., & Boomsma, A. (1998). Robustness studies in covariance structure modeling: An overview and meta-analysis. Sociological Methods & Research, 26(3), 329–367.

A few comments:

a) rules of thumb (as so often) are useless. Consequences of low sample size depend on the context (see the paper)

b) the "N-question" depends on the consequences: your first goal should be the test of the model (otherwise all issues like unbiasedness of parameters and efficiency are of less importance). With regard to the chi-square test, low sample size leads on one hand to low power but on the other hand to an overrejection of correct models. However, there are correction methods that can be applied with a little R function (the SWAIN correction) - see 

Herzog, W., & Boomsma, A. (2009). Small-sample robust estimators of noncentrality-based and incremental model fit. Structural Equation Modeling, 16, 1–27.

c) If your model fits, then low sample size biases your parameters. This is a concern (but as you see, its the last in the sequence). But there is no magical border of N.

I personnally would refrain from using PLS. What does it help that this method can estimate *something* with less bias and higher efficiency when it is unclear what this 'somethin'g is (and whether it reflects something reasonable (=no test of causal assumptions). 

Overall, the best bet is to conduct a simple Monte-Carlo-simulation (for instance using the simsem-package in R) in which you specify your target model as the population model an test if - given your sample size - this model could be recovered. See

Muthén, L. K., & Muthén, B. (2002). How to use a Monte Carlo study to decide on sample size and determine power. Structural Equation Modeling, 9(4), 599–620.

HTH

Holger

Hi,

Your question:

What's the minimum sample size to conduct the SEM analysis utilizing AMOS software?

For conduct the Structural equation model analysis using AMOS software, mininum 100 samples were needed.  Generally, SEM undergoes five steps of model specification, identification, estimation, evaluation, and modifications (possibly).  Using the above mentioned 5 steps you should do the SEM analysis by AMOS.

thans.

Hi all, 

You can check Hoyle [R.H. (ED.) (2012). Handbook of structural equation modeling,  Guilford Press] for a discussion on how many indicators should be included for each latent variable (p.65). Some ideas included:

̵   For a single latent variable with reflective indicators, 3

̵   When a model includes more than one latent variable and the latent variables are related, allowing for latent variables with even fewer indicators is

-  When sample size is small, estimation failures are less likely as the number of indicators per latent variables increases

I also recommend Hair et al. [Hair, Black, Babin, Anderson & Tatham. (2014). Multivariate Data Analysis, 7th Edition]. See discussion on sample size in pages 573-574. They also include some guidelines for using cutoff values for GOD indices depending on the model complexity (sample size and number of observed variables) (pp. 583-584)

Hope this helps,

Emilio

This could be of help as no direct or right answer:

http://www.quantpsy.org/rmsea/rmsea.htm 

SOPER, D. S. A-priori Sample Size Calculator for Structural Equation Models [Software]. , 2017. Disponível em: . .

for non-normal data, multiply by 2.

Well, I would say it depends on the complexity of your model and the number of constructs you're investigating including the fit indices you're using. Check Lacobucci (2010) for more information and use G Power to estimate what you may require in your study considering that the current estimates in literature are based on simulation that may not be applicable in a real life situation.

G*Power3 --> ok, for PLS-SEM

--> ok, for SEM-covariance based (LISREL, AMOS, EQS, Mplus, lavaan)

• the ruleof thumb i used to conduct research , a min of 250+ respondents . if your respondents are more than 300+ then you are in good shape .

YOU CAN SEE THIS ARTICLE

Read my post about What is a good sample size for Structural Equation Modeling (SEM) here: www.sarpublisher.com/what-is-a-good-sample-size-for-structural-equation-modeling-sem/

Hi Suthan,

nice post. However, you forget to consider a view relevant references (see below). Especially the simulations by Boomsma and Hoogland showed that you cannot really assume some fixed unit x parameter ratio.

Perhaps the references can help to improve the post...

With best regards

Holger

Boomsma, A., & Hoogland, J. J. (2001). The robustness of lisrel modeling revisited. In R. Cudeck, S. du Toit, & D. Sörbom (Eds.), (pp. 139-168). Chicago: Scientific Software International.

Green, S. B. (1991). How many subjects does it take to do a regression analysis? Multivariate Behavioral Research, 26(3), 499-510.

Herzog, W., & Boomsma, A. (2009). Small-sample robust estimators of noncentrality-based and incremental model fit. Structural Equation Modeling, 16(1), 1-27. doi:10.1080/10705510802561279

Jackson, D. L. (2003). Revisiting sample size and number of parameter estimates: Some support for the n:Q hypothesis. Structural Equation Modeling, 10(1), 128-141.

Muthén, L. K., & Muthén, B. O. (2002). How to use a monte carlo study to decide on sample size and determine power. Structural Equation Modeling, 9(4), 599-620. doi:10.1207/S15328007SEM0904_8

Nevitt, J., & Hancock, G. R. (2004). Evaluating small sample approaches for model test statistics in structural equation modeling. Multivariate Behavioral Research, 39(3), 439-478.

Reviews:

Difficult to develop generalized guidelines regarding sample size requirements for SEM (MacCallum, Widaman, Zhang, & Hong, 1999). Despite this, various rules-of-thumb have been advanced, including (a) a minimum sample size of 100 or 200 (Boomsma, 1982, 1985), (b) 5 or 10 observations per estimated parameter (Bentler & Chou, 1987; see also Bollen, 1989), and (c) 10 cases per variable (Nunnally, 1967). Such rules are problematic because they are not model-specific and may lead to grossly over-or underestimated sample size requirements. MacCallum et al. (1999) demonstrated that model characteristics such as the level of communality across the variables, sample size, and degree of factor determinacy all affect the accuracy of the parameter estimates and model fit statistics, which raises doubts about applying sample size rules-of-thumb to a specific SEM ( Wolf et al, 2013)

I used AMOS for SEM analysis and it gives solution even on te samples. Now i am confused whether the results are reliable are not. Because if 10 samples are not sufficient then why it shows reult without any warning or error??

Well generally it expeced to have atleast 150 !

Thanks

Good question. It is widely recognized that SEM is a large sample size techniques. SEM scholars (see; Kline, 2011,2016) argued that the minimum sample size is 200. On the other hand, Hair et al (2010) suggested that sample size should be five to ten times the number of indicators/items of the questionnaire.

Thanks for guidence.