I am hoping to conduct a moderated mediation analysis on quasi-experimental data and would like to know what the minimum sample size is. My model is based on Hayes' PROCESS model number 45 e.g. there are two moderators acting on the path from X to Mi and two different moderators acting on the path from Mi to Y.
I've looked at the literature on sample sizes for this type of analysis but there is very very little in the way of mediation analysis (and no moderated mediation) using experimental design in the literature for the topic area. I would really appreciate some advice!
When it involves moderation test, you need to break you sample into subgroup based on your moderator. Thus, think of every subgroup must have at lease 30 samples for the experimental data (laboratory type) and at 100 samples if your study involves field study data.
Sekaran (2003) wrote:
"Roscoe (1975) proposes the following rules of thumb for determining sample size:
1. Sample sizes larger than 30 and less than 500 are appropriate for most research.
2. Where samples are to be broken into subsamples; (male/females, juniors/seniors, etc.), a minimum sample size of 30 for each category is necessary.
3. In multivariate research (including multiple regression analyses), the sample size should be several times (preferably 10 times or more) as large as the number of variables in the study.
4. For simple experimental research with tight experimental controls (matched pairs, etc.), successful research is possible with samples as small as 10 to 20 in size."
Reference
Sekaran, U., 2003. Research methods for business: A skill building approach. John Wiley & Sons.
You will find a number of different suggestions for how many observations you need per coefficient in your model (typically ranging from 5 to 10). When you conduct a moderated analysis, you need that many observations for each sub-population.
Thank you all for your responses. David, do you have a source / any recommended reading to complement your response? Thanks in advance
I personally think the 5 or 10 observations per parameter method is informal at best, but widely used. You can find earlier discussions of this topic, which include various citations, at:
https://www.researchgate.net/post/What_is_the_minimum_sample_size_per_group_for_SEM_or_Path_Analysis
https://www.researchgate.net/post/What_is_the_minimum_sample_acceptable_for_structural_equation_modelling_using_AMOS
Hii Marwood
Please visit at http://statwiki.kolobkreations.com/index.php?title=Structural_Equation_Modeling
this may helpful for you
When it involves moderation test, you need to break you sample into subgroup based on your moderator. Thus, think of every subgroup must have at lease 30 samples for the experimental data (laboratory type) and at 100 samples if your study involves field study data.
Hi,
Estimates such as number of variables * number of people or number of parameters * number of people do not seem correct to me. Even in a simple moderation test, there is a variable, a * b, next to a and b. In the calculations above, the a * b interaction variable is ignored. In addition, this interaction variable does not have a normal distribution. Besides, there is no type I error if the significance test for the interaction variable is statistically insignificant. Also, Hayes's scripts compute bootstrap confidence intervals. I think a priori sample size estimate is not very important if your parameters'sizes are big enough and their standard errors are small enough.
Can someone please let me know if mediation moderation analysis can be apply in SPSS on 65 patients data?? And if not then what is the way to apply mediation moderation analysis on that data?
If you do not have a "measurement model" then mediation and moderation can be done in SPSS with ordinary least square regression. Mediation requires the use of "path models" and moderation involves the testing of interaction effects.
Dear Sir can you please send me your WhatsApp number so I can discuss details with you as it is needed my thesis deadline is on head .
Thanks
I would like to conduct a mediation analyses, but my sample size is 46 participants. A colleague of mine said that the minimum sample for a mediation/path analysis should be 75, but I cannot find this information anywhere online. Does anyone here know whether it would be appropriate to use the mediation analysis using my sample size? Note that the sample if representative of the population.
There is no hard and fast rule about minimum ample size, but with a sample as small as 46, you will have large standard errors, and that will make it difficult to locate significant effects. This is an issue of "statistical power," so you could read more about that topic.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size