Greetings,
I am a DBA student conducting a study about "Factors Influencing the Adoption of Payment Cards in Lebanon".
My research model, theoretically based on the Technology Adoption Model (TAM), consists of 9 variables, out of which:
- 6 constructs measured using multi-items scales adapted from literature ex. Perceived Usefulness (5 items), Perceive Ease of Use (5 items)
- 3 are nominal variables
I want to conduct a reliability analysis using SPSS, should I do so for each construct individually or for the whole questionnaire at once?
Thank you
Antoine
First, the basic reliability test in SPSS is Cronbach's alpha, which relies on correlations so it cannot use nominal variables.
Second, you can conduct separate calculations of Cronbach's alpha for each theoretical construct, but that will not assess whether there is "discriminant validity." In other words, an item you think is associated with Construct A might actually be more correlated with Construct B.
To assess whether your Constructs do indeed fit the pattern you predicted, the best approach would be Confirmatory Factor Analysis. Or you could start with Exploratory Factor Analysis, where I would recommend using Maximum Likelihood as the method for factor extraction and an oblique (correlated factors) rotation.
Note that factor analysis is also based on correlations, so once again you will not be able to use the nominal variables.
Thank you Mr. Morgan, much appreciated.
So, if I understand correctly, I can proceed as follows:
All the six constructs use Cronbach's alpha for reliability test. Conduct exploratory factor analysis for these six constructs for validity testing and use SEM analysis for model testing.
Yes, you have summarized my advice. But if the Exploratory Factory Analysis matches your hypothesized structure, then you can be almost certain that the Confirmatory Factory Analysis will as well.
Thank you for your replies David L Morgan Sindhu Singh
I conducted the reliability test for the six constructs and the result came as follows:
- Ease of Use: 0.785
- Usefulness: 0.784
- Social Influence: 0.792
- Perceived Cost: 0.635
- Perceived Risk: 0.790
- Loyalty Program: 0.888
Appreciate your feedback with respect to the Perceived Cost construct where alpha = 0.635 < 0.7. What should I do in this case.
Thank you
Hi Antonine
I would like to add, that (if I am correctly) the Confirmatory Factor Analysis is not a part of the SPSS standard package.
Regarding your question please click on the link below for more useful information.
Regards
Roland
https://www.researchgate.net/post/Is_it_mandatory_to_have_the_value_of_Cronbachs_alpha_above_70
For the scale with the low alpha, make sure you activate the option that gives you the results for "alpha if item deleted." That will show you if a single variable from the set is the cause of the low reliability.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS