Re: Advice requested regarding method of data analysis


Edd Grant wrote:
> Hi,
>
> I am currently writing my degree dissertation. I have performed an
> experiment which has obtained data from participants, the data
> contains opinions of the participants relating to 2 websites. An
> example question and answer is:
>
> Q: Do you feel that one of the websites provides better services than
> the other?

When the question is phrased that way, I would have been inclined
to answer "yes" or "no", and you would have to ask "which one?"
to the "yes" response to find out whether Site A or B was considered
to have "better services".

> Possible answers: Site A, Site B, Either, Neither
>
> The user is required to select one of the available answers which I
> believe makes this 'ordinal' data?

A hybrid of two binary (nominal) responses perhaps, but definitely
not "ordinal".

>
> I am looking for a method of analysing my data to enable me to answer
> the following question:
>
> Participants will, to some 'significant degree', find Website A to
> have offer better services. I realise that I need to define exactly
> what a 'significant degree' is and was hoping that there is some
> method I can use to do this? i.e. for a sample size of x, at least y
> candidates must give answer z for it to be statistically significant.

This kind of information is sometimes approached by asking questions
the respondants to rate the sites on a "likert" scale of 1 to 5, or
1 to 7, other others; or implicitly define your 'significant degree'
by putting it in your QUESTION to solicit:

Strongly Agree Agree Unable to Rate Disagree Strongly Disagree

>
> My sample size is 50 so quite small.

You should have a talk with those who use samples of size 1 to test
expensive items like nuclear bombs. "It's not the size that counts".
:-) Please don't try to run over me with your car if you see me
crossing a street.


> Does anyone know what methods I
> should be using here? So far I have looked at hypothesis testing but
> it seems no more useful in this scenario than simply looking at the
> number of responses for each site i.e. I can test the hypothesis that
> more people like website A than website B, but I can do that by
simply
> looking at the frequency of each response. I may have grossly
> misunderstood this so please let me know if I have overlooked
> something potentially useful here.

A sample survey course will help. You definitely need a mentor to
help you with the statistical elements of your dissertation.

>
> So, your thoughts and comments would be appreciated, I'm doing a S/W
> Engineering degree and am not familiar with many statistical terms so
> please go easy on me to begin with :-).

Thanks for the warning and tip. :^)

The remarks below are my global opinion on non-statisticians thinking
they can practice statistics (let alone well) by merely reading a
book or a chapter of a book on some statistical method, not realizing
that it takes as long to train a statistican to do his/her job
moderately well (say at the Ph.D. level of statistics) as it takes
a medical school to train a brain surgeon.

Would you let your good friend operate on your brain after s/he read
a book on brain surgery? :-)

I had posted this in ngs:

RF> Can you imagine what happens if every Tom, Dick and Harry start
RF> cutting up skulls in brain surgery after reading a paper back
RF> book on brain surgery? :-)

RF> My professional opinion about this statistical "mal-practice"
RF> was expressed in the flagship journal of the American Statistical
RF> Association (JASA 1982, 489-491) over two decades ago,

*> "I am less perturbed by the poor substantive quality of this
*> book than by the fact that we are witnessing the emergence of
*> a subculture of economists and social scientists, who are no
*> more qualified or equipped to practice statistics than law
*> law or medicine, yet who nonetheless do practice it among
*> their circles of nonstatisticians, without much visible signs
*> of protest from the community of statisticians. I feel
*> obliged to register my strongest protest against this type of
*> malpractice, < ... >"

> Thanks in advance of all your help.
>
> Edd

Nada.

-- Bob.

.

Relevant Pages

  • Question about use of 1-Sample Sign test
    ... I have collected some data and am trying to find a method of attaching ... Possible responses include: ... Prefer Website A ... to perform statistics without having the required knowledge!! ...
    (sci.stat.edu)
  • Why another list such as as jobs_sas?
    ... There are so many online job ... finders related to statistics and sas programers. ... Most companies have compensations for any hiring via internal ... You can simply go to your company's website, ...
    (sci.stat.math)
  • Why another list such as as jobs_sas?
    ... There are so many online job ... finders related to statistics and sas programers. ... You can simply go to your company's website, ...
    (sci.stat.consult)
  • Re: good stats table/formula needed
    ... LWn wrote: ... I want to bookmark such website for daily ... which has been for years a standard reference for formulas as well as ...
    (sci.stat.math)
  • Re: good stats table/formula needed
    ... LWn wrote: ... I want to bookmark such website for daily ... which has been for years a standard reference for formulas as well as ...
    (sci.math)