sAppropriate statistical tool

Home Forums Default Forum sAppropriate statistical tool

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
  • #4800


     I share a problem with Dalby. I am condcuting a  research on the impact of poverty reduction strategies in reducing poverty over a twenty year period. I am using the Likert type instrument to gather data. What is the statistical tool to use for analysis?


    Peter Betiang.


    Do you have the poverty data related to the time 20 years ago? To understand the impact of poverty reduction in effect, you have to compare the current mean (after poverty reduction initiatives) with the mean that was twenty years before. Comparison is done by Z -test. If the sample is small, use t-test or paired t-test.?

    Brian Cooper

    I have reviewed your questionnaire and the language used is very complex and requires a sophisticated understanding of English which may exclude many of your respondents. You have implied a relationship between the various items such as health centres, but you should be more explicit in the structuring of the questions used. A guide to language comprehension would to ensure that a person with approximately 6 years of basic education (primary standard) would be the basis of the language used. Currently the language used is that of a university graduate, which may exclude many of your respondents.

    What other evidence are you using to cross validate the questionnaire responses, as there will spatial variability in the impact of various programs. The experience in other regions is that the marketing approach of a particular strategy will have an impact on it acceptance to the particular community.


    Hello Dr Meesala,
    Thanks for that contribution. Its quite an eye opener. I will get to work to gather all available data, then revert to you. In any case, in carrying out the comparison, how to you control variance/ Once again thanks.

    Peter Betiang.


    Thanks Dr Brian,
    Thanks for your insight, i guess i will still need to look at that again, then further seek your advice.
    Once again, many thanks.

    Peter Betiang

    varadi vijay


    Brian Cooper

    There are many fine stats packages soem are propriety such as SPSS or SAS. But packages such as R which are opensource are also available. The trick is to skilled in one package and exploit all the capacity of the package.

    There are processes such as PCA or Factor or SEM. Simple regression, crosstabs CHAID or Tree could also be used. From your responses keeping it simple is the best path to follow.


    Dr Brian,
    I believe that is precisely what i am looking to do. But am sure you must have noticed that it is not appropriate to use just one tool to look at all hypotheses. Thanks, and i will keep at this.

    Peter Betiang
    Doctoral candidate, Department of Adult Education
    University of Calabar, Calabar

    Brian Cooper


    I wish you well in your study, half the battle of the thesis is your relationship to your supervisors. If they do not understand or relate to your area of study, you will have some problems. Ensure you have at least one you are able to have a conversation with. In many countries especially those who base the PHD program on the British model, the relationship between the candidate and the supervisor is that of the master and servant.

    If you are feeling unhappy in your journey have a look at, perhaps then the journey will become clearer.

    Brian Cooper
    Faculty of Health Sciences
    University of Sydney


    Dear Brian,
    You could’nt have captured that a problem in a more precise manner.
    i actually have some cordial relationship with my supervisor, but it is an uneasy master-servant thing. I quite find it difficult to connect properly with him. I am meant to take instructions, and not attempt to insist that my point of view on certain issues can really be useful. In any case, thANKS AGAIN for your wonderful portrayal of this truth, and thanks for your good wishes to me.
    Peter Betiang, University of Calabar, Nigeria.

Viewing 10 posts - 1 through 10 (of 10 total)
  • The forum ‘Default Forum’ is closed to new topics and replies.