+ Reply to Thread
Results 1 to 6 of 6

Thread: HELP Dissertation Research - Which Statistical Test Do I Use?

  1. #1
    Points: 70, Level: 1
    Level completed: 40%, Points required for next Level: 30

    Posts
    3
    Thanks
    0
    Thanked 0 Times in 0 Posts

    HELP Dissertation Research - Which Statistical Test Do I Use?




    Hi,

    I'm currently completing my Dissertation paper and I'm stumped on which statistical test I should be using!.

    I have 30 completed questionnaires which I had random people rank from 1-11 which factors they prioritised when selecting an external wall material for a house. The factors were things like Aesthetics, Strength, Cost, Durability etc.

    I have now totaled all of the results in a table as shown attached. Can someone please advise me which test I can run on this data set? My supervisor advised using the Friedman Test, but i dont know if this would be applicable?

    Thank you in advance!

    Alex
    Attached Images  

  2. #2
    TS Contributor
    Points: 41,667, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Downloads
    gianmarco's Avatar
    Location
    Italy
    Posts
    1,381
    Thanks
    237
    Thanked 305 Times in 229 Posts

    Re: HELP Dissertation Research - Which Statistical Test Do I Use?

    Hello,
    I am not going to address your question; I got no experience with that kind of research field and analysis.
    However, I had a couple of spare minutes to play around with your dataset to graphically display the relation between priorities in selecting the external material and the frequency with which people assigned the rankings.

    In other words, I was courious to know if there was a relation between priorities and the ranking. Therefore, I treated that rankings (columns) as levels of a categorical variable, and applied Correspondence Analysis. I am attaching a scatterplot in which rows and columns are displayed in a low-dimensional space, which portays the association between them. I will not go into the details, so please disregard the numbers and symbols that show up within parentheses.

    Indeed, it would seem that priorities such as 'Builder Reccomendation', 'Availability of Materials', and 'Environmental Concerns' tend to be associated to higher ranks. It is also interesting to note that 'Cost' seems not to be a priority in material selection. Other priorities, such as 'Strength' and 'Thermal Quality' seem associated to medium-to-low rankings. In general, the 1st dimension (horizontal axis) is actually capturing a large amount of data variability (called 'inertia' in CA terms), and is actually lining up rankings roughly from lowest (left-hand side) to highest (right-hand side). The two dimensions togheter are retaining about 60% of the deviation from independence. Given that your table has dimensionality 10, that means that even if you are dropping 8 dimensions you are still retaining an 'important' amount of information about data association.
    Finally, the association in the table is significant, as the chi-sq test of independence would indicate.

    Hope this will prove interesting.
    Best
    Gm
    Attached Images  
    http://cainarchaeology.weebly.com/

  3. #3
    Points: 70, Level: 1
    Level completed: 40%, Points required for next Level: 30

    Posts
    3
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: HELP Dissertation Research - Which Statistical Test Do I Use?

    Thank you GM for your reply.

    This graph shows exactly what I wanted to try and express. Are you able to provide a little more detail about the numbers in this graph?.

    I'm a statistical novice and still in my early learning stages.

    Any extra information would be much appreciated.

    Alex

  4. #4
    TS Contributor
    Points: 41,667, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Downloads
    gianmarco's Avatar
    Location
    Italy
    Posts
    1,381
    Thanks
    237
    Thanked 305 Times in 229 Posts

    Re: HELP Dissertation Research - Which Statistical Test Do I Use?

    Hello,
    first of all, I do not want to lead you astray suggesting the use of CA. If your advisor suggested the Friedman test is possibly because (s)he believes that you need an inferential framework. I am not familiar with that particular test, but I can imagine that (s)he wants you to assess to what extent the choice of materials is different regarding to the ranks. You may want to go deeper into that test, possible googling a little first.

    Secondly, if you want to grasp the basics of CA, you may want to read this article, which introduces CA from a marketing research perspective. Once you have some basic understanding of CA, you can better understand the plot I attached to my first reply. Below, I paste the help documentation of the R function I used to produce that plot:

    ca.plot() provides the facility to produce:
    (2) a scatterplot with advanced labels. If the user's interest lies (for instance) in interpreting the rows in the space defined by the column categories, by setting the parameter 'cntr' to "columns" the columns' labels will be coupled with two asterisks within round brackets; each asterisk (if present) will indicate if the category is a major contributor to the definition of the first selected dimension (if the first asterisk to the left is present) and/or if the same category is also a major contributor to the definition of the second selected dimension (if the asterisk to the right is present). The rows' labels will report the correlation (i.e., sqrt(COS2)) with the selected dimensions; the correlation values are reported between square brackets; the left-hand side value refers to the correlation with the first selected dimensions, while the right-hand side value refers to the correlation with the second selected dimension. If the parameter 'cntr' is set to "rows", the row categories' labels will indicate the contribution, and the column categories' labels will report the correlation values.

    Best
    Gm
    http://cainarchaeology.weebly.com/

  5. #5
    TS Contributor
    Points: 19,199, Level: 87
    Level completed: 70%, Points required for next Level: 151
    CowboyBear's Avatar
    Location
    New Zealand
    Posts
    2,080
    Thanks
    123
    Thanked 429 Times in 330 Posts

    Re: HELP Dissertation Research - Which Statistical Test Do I Use?

    Quote Originally Posted by alexfinch89 View Post
    I have now totaled all of the results in a table as shown attached. Can someone please advise me which test I can run on this data set?
    It depends on what you are trying to find out. What are your research questions or hypotheses?
    Matt aka CB | twitter.com/matthewmatix

  6. #6
    Points: 70, Level: 1
    Level completed: 40%, Points required for next Level: 30

    Posts
    3
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: HELP Dissertation Research - Which Statistical Test Do I Use?


    Sorry, I should have added in some more detail.

    I'm basically looking at characteristics of external walls in residential construction and seeing which types have advantageous factors ie Double brick construction has high levels of durability, strength etc.

    Then I conducted a 30 person questionnaire to try and determine which factors a sample of our population prioritize, to see if there is alignment between what the demand side want/need and what the supply side is most commonly constructing.

    Alex

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats