Data Science Vidhya

Step by step approach to analyze data to publish research papers in Top journals

• Posted on February 24, 2022
• Ateeque Shaikh
• No Comments
• Marketing Research

1. Model – The first step in analyzing the data is to look at the hypothesized model.

Variables in the model are as given below

PCO – Proactive customer orientation

RCO – Responsive customer orientation

PV – Perceived value

CB – Cross buying

WOM – Word of mouth

1. Discuss the hypotheses – directionality of the hypothesis
2.  
3. Discuss the constructs – operational definition of the constructs
4.  
5. Discuss how the constructs were measured – scale items used to measure the construct
6.  
7. Design of the questionnaire – Probably show a sample questionnaire
8.  
9. Show the data and relate it back to the questionnaire and the scale items. Do the coding of the scale items if required and store it in excel sheet.
10.  
11. Pre-requisite for understanding the analysis – Descriptive statistics, Instrument reliability and validity, correlation, factor analysis – Exploratory and confirmatory, regression.
12. 1. Let us start our analysis
    2.  
13. The first is to do reliability analysis – why this is important – entire set of variables and then construct wise. Look at the Cronbach alpha numbers
14.  
15. Exploratory factor analysis (If you are developing scale – confirmatory factor analysis (if you are using established scales)
16.  
17. This information is required because you need to also carry validity assessment i.e. convergent validity and discriminant validity
18.  
19. CFA AMOS – or measurement model in IBM AMOS. This will give factory loading. From factory loading, you can calculate the scale composite reliability, Average variance extracted (AVE)
20.  
21. Convergent validity (Factor loading, AVE, Scale composite reliability calculation), Discriminant validity (Correlation, AVE,  Fornell Larcker ratio)
22.  
23. Infer the discriminant validity and convergent validity – Generate the tables convergent and discriminant validity tables.
24.  
25. Structural model testing – Model fit indices and the threshold, important indices, Hypotheses testing. A single-headed arrow is used to represent a hypothesized structural relationship between one construct and another.
26.  
27. Why residual is to be added in structural model? Ans –
28.  
29. Output analysis of structural equation modelling
30. Chi-Square –
    • Non-significance of Chi-square value implies that there is no difference between the assumed covariance matrix and the covariance matrix of the data. This would result in ideal model fit indices.
    • If the Chi-square value is significant then we need check for model fit indices. There are various thresholds of the model fit indices
31.  
32. Modification index was used to identify error covariance between items of within construct and between constructs. Modification indices are used to delete the items in the measurement model
33.  
34. Then look at the significance level of different hypothesis in the model.

Further readings

What is p-value?

Why SEM is preferable to regression?