Two Questionnaires

Home Forums Default Forum Two Questionnaires

Viewing 15 posts - 1 through 15 (of 16 total)
  • Author
    Posts
  • #1918

    Dear Readers

    i have two questions with respect to data analysis in SPSS.

    Question 1

    I have One Questionnaire on which the data is collected from Employees, now i have other questionnaire on which the data is collected from Customers, now the data that is collected from Employees has 200 Respondents, and i have collected data on another questionnaire from 100 respondents, i want to correlate the variables from Employees Questionnaire and Customers Questionnaire As i see it, i can add variables in one SPSS file and then correlate them, but in that case there are 100 less respondents for Customers, how shall i do this. suggestions needed

    Question 2 I have collected data on a questionnaire from Employee, the questionnaire is close ended, now i want ot see the influence of Organizational Learning Culture (i have 16 variables, 3 factors) and have 250 respondents, i have the sales figures for 10 years, how do i correlate the questionnaire data (Primary) and sales data. Regards

    #1933

    Dear friend
    Correlation is not the tool for your problem. You have two sets of data, one from employees and other from customers. Correlation is tool that may be used in a case where the same sample provides data on two or more variables. Also, there needs to be a linear relation between the variables used for correlation. Your second question can be answered if you describe the problem and data more explicitly.

    #1932
    Rafael Garcia
    Participant

    K.Kalyanaraman’s comment is correct. Correlation is inappropriate. In order to do correlation the cases must be from the same source (people, cities, states, companies, whatever). Since these are two different samples (I’m assuming) they do not have the same cases in both samples and therefore cannot be correlated in any principled way. 

    As near as I can tell, at best you can do descriptive statistics and call it a day.

    Best,

    Rafael

    #1931
    Anonymous
    Inactive

    Dear Khavaja

    There is not any problem because you do relation between variables no between respondents,so you must gather indexes of variables and do relations.in fact you can compare the  employees responses and customers responses or there relations.

    #1930

    Dear Mojtaba

    thx for your response, but what if i have 200 replies from Employees and 100 from Customers, in this case i am short by 100 replies from customers.

    and what about my Question 2, i am doing research on Influence of HRM practices on Financial Performance, where data on HRM practices is gathered through questionnaire while financial performance is the sales of the business from last 10 years.

    Regards

    #1929
    Rafael Garcia
    Participant

    Mojtaba, I would love for more of a description on how to do this. 

    My understanding is as follows:

    A correlation is the covariance/(product of the standard deviations). The covariance is derived by summing the product of the raw deviations for each case. Using standard Pearson’s r, you would need matched cases. In fact, The entire point of the correlation coefficient is to tell you how the deviations are moving on average, or put another way, how cases that respond one measure will tend to respond on another. Pearson’s r is about individual differences, so there need to be ‘individuals’.

    Are you suggesting something else, because I have no idea how someone would go about showing that two variables collected from two different samples are associated with each other quantitatively? 

    Best,
    Rafael

    #1928
    Rafael Garcia
    Participant

    With respect to Question 2, there is a lot that can be done with a simple regression.

    #1927
    Anonymous
    Inactive

    About your first question  as I told you the number of respondent is not important because you work with variables.

    about your second question ,you must use the mean of sell in past ten years and do relation with the questionnaire questions. 

    #1926
    Rafael Garcia
    Participant

    You can also do a Mixed GLM looking at between effects (the HRM practices), the within effects (time), and the interactions.

    #1925

    good sir…thanks

    #1924

    I too have difficulty in using rate questions in closed ended Questionnaire. Will you suggest how to use rate questions on scale point like 1—10  in closed ended questionnaire

    #1923

    Dear Rafael

    I found a way out for Question 2, For Instance i have Earning Per Share from ten Banks, and i have Questionnaires filled from employees of ten banks, i will take mean score of Questionnaire on each of the eight bank and correlate then with Earing Per Share for instance look at the following table. I have calculated the Mean of Internal Service Quality for every Bank

    What do u think, is this the right way. Please reply soon as i have to submit my phd proposal and i am stuck beacuse of this analytical glitch

     BankID                                

     Average Internal Service Quality

     Earning Per Share

    1

    3.5

    2.1

     2

     3.6

     2.2

     3

     4.0

     3

     4

     2.8

     2

     5

     3.8

     2.5

     6

     3.9

     2.45

     7

     4

     3.1

     8

     4.1

     3.2

    #1922

    Dear Rafael

    I found a way out for Question 2, For Instance i have Earning Per Share from ten Banks, and i have Questionnaires filled from employees of ten banks, i will take mean score of Questionnaire on each of the eight bank and correlate then with Earing Per Share for instance look at the following table. I have calculated the Mean of Internal Service Quality for every Bank

    What do u think, is this the right way. Please reply soon as i have to submit my phd proposal and i am stuck beacuse of this analytical glitch

     BankID                                

     Average Internal Service Quality

     Earning Per Share

    1

    3.5

    2.1

     2

     3.6

     2.2

     3

     4.0

     3

     4

     2.8

     2

     5

     3.8

     2.5

     6

     3.9

     2.45

     7

     4

     3.1

     8

     4.1

     3.2

    #1921
    Rafael Garcia
    Participant

    Khawaja,

    Yes, this is how I would probably do it. If you think the mean scores are important, you need to put them on the same level of analysis. If you only have 10 banks, though, the power is going to be very low (you’re not likely to detect significance). This may or may not be an issue depending on your committee. I would just report the r or r^2 and state that you are aware of the limitation and that acquiring more bank data would allow for significance to be detected. You may want to do a power analysis (using the effect size you gather from these data) and state how many banks you would need for your effect. 

    Hope that helps.

    Raf

    #1920

    Dear Rafael

    Thank you so very much for your prompt response, so if i increase the sample size of banks i will gain more power in analysis.

    regards

Viewing 15 posts - 1 through 15 (of 16 total)
  • The forum ‘Default Forum’ is closed to new topics and replies.