Non-parametric tests for Phd or research papers

[maliks...@gmail.com]

Dear Sir,

In my results of Phd since the data collected was not normal, i used most of the non-parametric tests instead of ANOVA. But the supervisors in the presentation suggested me to use parametric tests instead. He advised me to convert my data to normal using some tests and then use Parametric tests. My related queries are:

1. Does using Non-parametric tests in Phd indicate poor quality research?
2. Does using Non-parametric tests in Research papers for publication indicate poor quality research?
3. Can/ should we normalize data and use parametric tests instead?
4, Can you guide me on the tool I can use to normalize data.

Thanks & Regards

Sonia Malik 

[Neeraj Kaushik]

Dear Sonia

Generally its a convention in academic community to use Parametric test irrespective of whether they are applicable or not?

Lets try to understand the basics first, then u can decide for ur own. Paramteric tests are applied only when data is metric i.e. either on Likert scale or ratio data. In Likert scale, ideally we shd get the sum of all statements related to one construct and then check for its normality of that sum (Using 1 sample KS test) and if normally distributed then apply t/ANOVA else U/H-test. In no case we shd apply t-test/ANOVA on Multiple choice questions.

Now tranformation is an attempt to convert non-normal data into normal data. But we shd do transformation of only for the ratio data. We never go for the transformation of Likert scales.

So if you've used the Likert scale statement then get the sum of each construct and then check for its normality. Some experts are of the opinion that if sample size>250-300 then we shd not worry about the normality and quoting Central limit theorm, we can say that large data is generally normally distributed & go ahead with parametric tests.

Now you decide for ur own and ask for any/every doubt.

Best wishes

Neeraj

--
You received this message because you are subscribed to the Google Groups "DataAnalysis" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dataanalysistrai...@googlegroups.com.
To post to this group, send email to dataanalys...@googlegroups.com.
Visit this group at http://groups.google.com/group/dataanalysistraining.
For more options, visit https://groups.google.com/d/optout.

[Rahman M]

Dear Neeraj sir,

 Can we also quote the rule of thumb suggested by Bulmer, M. G., Principles of Statistics (Dover, 1979) that if skewness is between - 0.50 to +0.50,  the distribution is approximately symmetric. and s Some lierature like laerd statistics (https://statistics.laerd.com/spss-tutorials/one-way-anova-using-spss-statistics.php). ANOVA can be used for approximately normally distrbuted data.

 Please suggest whether we can give the above references while applying ANOVA for such data.

 Regards

M. RAHMAN

To unsubscribe from this group and stop receiving emails from it, send an email to dataanalysistraining+unsub...@googlegroups.com.

[maliks...@gmail.com]

Dear sir
Thanks but still there are few doubts.

I am checking the normality on the sum of likert scale questions. Can transformation be done on the sum also. Is sum / mean equivalent to ratio scale?

How to do transformation to convert not normal data to a normal one?

Regards
Sonia Malik

[Neeraj Kaushik]

Dear Rahman
Yes in many studies they quote this thumb rule. Even otherwise if
there are approx equal no. of items in various groups then we can use
ANOVA even in case of violation of Normality assumption. (See in
attached last line of 1st para from the book "Discovering Statistics
using SPSS by Andy Field)
Ex: For the variable Income group
a) 10,000-20,000
b) 20,000-30,000
c) 30,000-40,000
If there are almost equal no. of respondents in all 3 categories then
we can use ANOVA irrespective of whether within each category the
responses are normally distributed or not.
Best wishes
Neeraj

>>> email to dataanalysistrai...@googlegroups.com.

>>> To post to this group, send email to dataanalys...@googlegroups.com.
>>>
>>> Visit this group at http://groups.google.com/group/dataanalysistraining.
>>> For more options, visit https://groups.google.com/d/optout.
>>
>>

> --
> You received this message because you are subscribed to the Google Groups
> "DataAnalysis" group.
> To unsubscribe from this group and stop receiving emails from it, send an

> email to dataanalysistrai...@googlegroups.com.

[Rahman M]

Dear Neeraj sir,

A sincere thanks to you for the valuable information.

Regards

M.RAHMAN

>>> email to dataanalysistraining+unsub...@googlegroups.com.

>>> To post to this group, send email to dataanalys...@googlegroups.com.
>>>
>>> Visit this group at http://groups.google.com/group/dataanalysistraining.
>>> For more options, visit https://groups.google.com/d/optout.
>>
>>
> --
> You received this message because you are subscribed to the Google Groups
> "DataAnalysis" group.
> To unsubscribe from this group and stop receiving emails from it, send an

> email to dataanalysistraining+unsub...@googlegroups.com.