My factor loadings for General Trust are very low. What can i do now?
My hypothesis is "a high level of cognitive/affective trust leads to a LOW willingness to use Robo-Advisors". How do I interpret now the results? Does a low value mean that the hypothesis is right??
Some items have low loadings and others do not. You don't say anything about the nature of the scale or where it comes from, but it looks like maybe it is multidimensional. I would take just the items of this scale and run an exploratory factor analysis to see how many factors there are and if those factors make sense. It is also possible that some items do a good job of reflecting trust, and others do not. An item analysis could be used to find an internally consistent set of items.
The diagram is, of course, rather complex. It looks like you have two trust measures - one for "general trust" and one yielding separate scores for "cognitive" and "affective" trust. Is that correct? It would be helpful to know the zero-order correlations between the general trust scale and the other(s). As well as other scale-related information, as Paul suggests.
In any case, it looks like your hypothesis is supported with regard to cognitive trust (the -.31 value is fairly robust), but not with regard to affective trust (.083 is pretty trivial). And I think there is something wrong with your general trust measure. Looking at the "loadings" of the items as if they were item-total correlations, one sees at once that many of them are quite low - and some are actually negative! If scale reliability is as low as those figures suggest, then it won't correlate with much of anything.
I hate to ask this, but is it possible you made a mistake in scoring the general trust scale? That some of those items were meant to be reverse-scored?
In general your model is complex one. A first challenge to reduce the items from general trust if they are adding less. then about your hypotheses cognitive trust have -.31 loading confirms the hypotheses but on the other hand the it is not true for effective trust (0.083). however with the moderation of (general trust*cognitive trust) the result get changed from (-.31 to 0.087).
I would throw out any item that does not load .5 or above. Try rewriting those items to get more directly at on operational notion of what trust means. Standard scales are often bad because they are not based on a behavioral instantiation of the concept
- Badges
- Science topic
- Similar topics
- Psychology
- Psychometrics