Colleagues
I have a researcher who has undertaken analyses of barriers and drivers to change in a Public Sector Organisation (Pakistan). Sample size was 300, alphas and normality were all good, for regression data was homoscedastic and VIFs all lay between 1-10 (no multicolinearity). 11/12 aspects of change were measured and variables (IV) subjected to EFA followed by SEM. Two DV variables were also used in the study (intention to adopt and level of change success). The best resolved factor model (11 factors) was also run in AMOS against the two DVs.
The 11/12 IV measures were also separately collapsed and subjected to Multiple regression against the two DVs. Best models obtained were a six IV forecasting intention to adopt and a five IV forecasting level of change success. Both models were also run in AMOS
In reporting these outcomes for publication it is apparent that the factor models (with GFIs below 0.900 and CFIs below 0.95) have reasonable fit but are not, what is considered ‘well-fitting’? Whereas the regression model indicators do fall into this ‘well-fitting’ category (with exception of TLI values).
Comments appreciated?
SEM for Factor analysis model as illustrated in AMOS Model Fit: chi2 (224) = 613.457 p < .05; CFI = .933; GFI = .857; TLI = .918; RMSEA= .062 and NFI = .900
Factor model to forecast first dependent variable (Level of adoption to change) Model Fit: chi2 (224) = 613.457 p < .05; CFI = .933; GFI = .857; TLI = .918; RMSEA= .062 and NFI = .900
Factor model to forecast second dependent variable (Level of success to change) Model Fit: chi2 (240) = 642.572 p < .05; CFI = .933; GFI = .858; TLI = .917; RMSEA= .061 and NFI = .900
Multiple regression model in AMOS (level of adoption, dependent variable) - Model Fit: chi2 = 6.495 p < .05; CFI = .993; GFI = .995; TLI = .795; RMR .000 and NFI = .992
Multiple regression model in AMOS (Level of success dependent variable) - Model Fit: chi2 = 4.939 p < .05; CFI = .995; GFI = .995; TLI = .904; RMR .006 and NFI = .994
Dear Robert , Sudhir et all
The most important analysis in the results fo a model is to see if it has a real interpretation (theory that support it).
You should not depend mainly on statistics (like GFI et.c). They are important, but the final decision is yours. iIt is always better to have a model that doe not fit so well the data ( as yours in some statistics) but that has a real interpretation then to have a very good model in terms of fit but that is it not connected with reality.
I hope this helps and this is of course my opinion and the way i always analyse data.
Have a nice time
Helena
Dear sir
Your value of GFI for factor analysis seems little low, its .85 and we expect this >.90
you will have to produce references of the studies finding this value significant and acceptable. rest is fine I guess.
Dear Robert , Sudhir et all
The most important analysis in the results fo a model is to see if it has a real interpretation (theory that support it).
You should not depend mainly on statistics (like GFI et.c). They are important, but the final decision is yours. iIt is always better to have a model that doe not fit so well the data ( as yours in some statistics) but that has a real interpretation then to have a very good model in terms of fit but that is it not connected with reality.
I hope this helps and this is of course my opinion and the way i always analyse data.
Have a nice time
Helena
Respected Robert, I totally agree with Helena, that what assumption(theory) we have taken into account is more important, although your model shows that it is fit, as you know for its fitness require, absolute fit, incremental fit and parsimonious fit. for which you have to check RAMSEA, CFI and Chi2/df values. for the robust check you can apply Pearson correlation test to find out relationship between IV and Dvs. so its better to check whats the theory says in context of this relationship( IV and DVS). For example OCB is strongly related to behavioural leadership (Poohongthong et al., 2014; Walumbwa & Schaubroeck, 2009).
The information in RG about SEM:
Search
Giovanna Gabetta asked a question:
What is the technique one can use to make SEM images of a wax deposited on a metal sample?Is metal sputtering a suitable technique to prepare the surface?
Follow Answer
Abdillah Imron Nasution asked a question:
Preparation of Bacteria-SEM study: Can I replace gluteraldehide with hydrogen peroxide+high Level of disinfectant?-
Follow Answer
Imran Patel asked a question:
How do we read Energy dispersive spectroscopy report for SEM?Please find the EDS report attached here.
Please comment soon.
EDS report.pdf
Follow Answer
Son Thanh Nguyen asked a question:
What is the reason of strange Si peak in SEM/EDS analysis?I am using SEM/EDS analysis to investigate the composition of my sample (which was sintered by Y, Ti, O and Fe). Unexpectedly, there is a peak of Si coming out. I tried to analyze the sample few times, at different domains, but the strange peak still come out. Now I am trying to explain the appearance of these strange peak. Could you tell me your idea? ( I also attach here a spectrometry of my EDS analysis)
[more]
https://www.dropbox.com/s/imln8lut6132u0k/150mag.jpg
150mag.jpgShared with Dropbox
Follow Answer
Arvind Kumar Dixit asked a question:
How can we do phase analysis of complex alloy system (which have 14 elements) by using SEM and EDS?I want to do phase analysis of complex alloys system which is prepared by melting 14 elements in its pure and oxide form.
Follow Answer
Manju Taxak asked a question:
In a line scan of SEM-EDX analysis of homogeneous alloy, up and down lines in EDX patterns are observed instead of a single line. What does this indicate?...
Follow Answer
Changqi Xu asked a question:
What is difference between GSE and SE detector in SEM?What is difference between GSE and SE detector in SEM?
Follow Answer
Paul Sauer asked a question:
I have a structural equation model that is being tested with empirical data that was collected at two separate points in time (time period 1 and time period 2) from the same set of respondents. How can I test changes in the model parameters from time period 1 to time period 2?...
Follow Answer
Muhammad Imran Arshad asked a question:
What is difference between sem and tem?Aa
Follow Answer
Abbas Rahdar asked a question:
Can anyone tell me about the ZAF method application in the SEM?Can anyone tell me about the ZAF method application in the SEM?
Follow Answer
Shahid iqbal Iqbal asked a question:
Can someone assist me with my sample preparation for SEM analysis?Please assist me with your kind advice on the following:
We have prepared AAO templates in phosphoric acid by two steps anodization. And we have deposited NiZn by electropolishing. Now we are unable to prepare our samples for SEM analysis because when we try to dissolve AAO templates, it causes our NiZn to dissolve. Please advice me how to prepare our sample for SEM. Any method?
Sincerely,
[more]
Follow Answer
Mohd Azmir Mohd Nizah asked a question:
What is the "Rule of Thumb's" for SEM analysis to be more reliable?For example, in terms of scale that should be used, i.e. continuum or ordinal or nominal; in terms of normality of data, and else...
Follow Answer
Manoranjan Arakha asked a question:
How many days can we store the fixed bacteria slides for SEM analysis?I am preparing the bacteria slides using glutaraldehyde and tannic acid and storing them in normal environmental condition for SEM analysis. My question is that how many days we can store it before SEM analysis so that the structure of the bacteria will not change.
Follow Answer
Ramendra Pratap Singh asked a question:
Why is the over identified model required in case of SEM? Is it possible to get proper result from Just identified model?What is the logic behind over-identified model for Structure equation analysis.Whether it is possible to get result from just identified model and why it is essential to use over-identified model in case of hypothesis testing?
Follow Answer
Gansukh Erdenedelger asked a question:
Why, during SEM EDX, we cannot detect Nitrogen. However on the element map Nitrogen is detected. What reason? Could someone explain?Why, during SEM EDX, we cannot detect Nitrogen. However on the element map Nitrogen is detected. What reason? Could someone explain?
Follow Answer
Moyassar Al-Taie asked a question:
Does anyone know how to compute Q2 in PLS with pls graph package?Seeking a step by step explanation to get the value of Q2 .
Follow Answer
M. Yusuf Hakim Widianto asked a question:
How to analyse porosity in the SEM pics and how to know percentage material SEM EDX? Are you using software? What software?Want to know how to calculate porosity in SEM pics, qualitative and quantitative.
Follow Answer
Evangelia Kyriazi asked a question:
SEM-EDS for fossil analysisHas anyone used SEM-EDS for the analysis of fossils, especially fossil wood?
Can anyone recommend any relevant papers?
How many kV would you recommend?
What magnification would you recommend for the observation of the deterioration of fossil wood?
[more]
Follow Answer
Chinmaya Kumar Sahoo asked a question:
Can anyone suggest free scientific image measurement software?I require free scientific image measurement software to measure the distance and area of an SEM image. I have been using Gwyddion but it is creating problem with Windows 7
Follow Answer
Md Mahfuzur Rahman asked a question:
How can I analyse moderating variables in SEM?When the moderating variables have a hypothetical relationship with all the independent variables?
Follow Answer
12345…348
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software