Hi Roldan,

Before you selected those 70+ items and run EFA on them, you must have identified theoretical or conceptual model underlying these indicators from the extant literatures. The model should guide your interpretation and understanding of the items - whether theoretically, they can be grouped into second order or even third order, or not at all. I would advise not to over-rely on statistics (even though it is a powerful tool that tells you whether groups of items have good fit for valid model) because, in the end, the factors and overall model should make sense. Hope this helps.

Hello Roldan,

Whether a second-order solution should be pursued is a question you should likely have answered already (with appropriate justification), as Hanif's answer suggests.

From a practical perspective, however, if your EFA used an orthogonal rotation scheme (or no rotation scheme), then your extracted factors will be uncorrelated and there will be no second-order factor(s). You must agree (and implement) an oblique rotation solution, allowing the factors to be correlated, before any higher-order factor solution may be pursued.

Good luck with your work!

You should depend on a theoretical base as mentioned above

if so you could use higher order factor analysis procedures (ex: Schmid Leiman solution) to detect the structure of the data.

you can do your analysis using FACTOR which is an easy used freeware, its link:

http://psico.fcep.urv.es/utilitats/factor/Download.html

you can also check Navruz and his colleagues paper:

Article A Review of Higher-Order Factor Analysis Interpretation Strategies

good luck