I am testing a model that has 5 variables and 4 moderation effects (details in the file attached). My current sample has 145 observations, and I would like to check if this sample size is OK or maybe I should test a simpler model.
May anyone help me with some procedure for I calculate the minimal sample size in this case? Thanks in advance.
Dear Guillermo,
What do you mean by testing this model? Do refer to test the fit of the model or do refer to test whether a certain path coefficient is significantly different from zero? Likely not the former as you face a non-linear model which makes the assessment of the model fit in terms of statistical testing quite difficult.
Best regards,
Florian
Hello
It is really hard to answer this as your description of your model with constructs and number of variables and types of relations has been described in a brief fashion.
You say little about the reliability or validity dynamics of the measurements.
Are they existing scales, new ones, tested in a new culture?
Pre.. running the study are you expecting the effects to show up in a clear strong fashion or perhaps you are unsure and may expect weak or no effects (always a possibility).
Such things might determine whether a larger sample size or not in initial testing is required.
My 0.02 cent.
Brad
Dear Florian Schuberth and Bradley Wilson
Thanks for your contributions. I am testing whether the paths coefficient is significantly different from zero. I am using ADANCO. The saturated SRMR is below 0.07 and the estimated one is below 0.09
I am testing the effects of social integration mechanisms on ACAP in firms from emerging countries, using scales previously validated. Constructs show good consistency and reliability. The effect sizes vary between low and medium.
I was wondering if my sample size (145 observatoins) is adequate, and if there is a procedure for calculating the minimal sample size for assuring the consistency of this model.
Regards,
Guillermo.
Previously validated scales in emerging countries or other countries.
How many variables for each construct?
Dear Guillermo,
have you also run the test for overall model fit in ADANCO? This would give you an indication whether the model fits the data.
Yes there are procedure to determine the sample size to achieve a certain power, see for example Section 3.2 of:
Article How to perform and report an impactful analysis using partia...
HTH
Best regards,
Florian
- Badges
- Science topic
- Similar topics
- Filing