Research Methodology Lecture-1

[Sukhmani Singh]

Dear Neeraj Sir,

Greetings!

Sir, I saw your video on Research Methodology Lecture-1 (Extracting Variables from the Objective), posted on 22nd March on you-tube. Sir, thank you very much for educating us the basics of research methodology in the most simple way! It was an enlightening lecture about variables and objectives.

Sir, I have a doubt and since “comments” are turned off in you-tube, that’s why I am asking here.

At 1:25:30 time you gave an example:
“To compare the happiness level and intelligence level of the boy and girl students of MBA in NIT Kurukshetra.”

….and said that we can split this into….

a) To compare the happiness level of boys and girls

b) To compare the intelligence level of boys and girls.

“Thus, there are 3 variables and 2 bivariate objectives.”

Sir, my doubt is that – Isn’t there are 3 variables and 4 bivariate objectives like:

a) To compare the happiness level of boys and girls.

b) To compare the intelligence level of boys and girls.

c) To compare the happiness level and intelligence level of boys.

d) To compare the happiness level and intelligence level of girls.

Sir, can we regard (c) and (d) as bivariate objective too?

Kindly guide,

Thanks and Regards,

Sukhmani.

 

 

 

 

 

[Neeraj Kaushik]

Dear Sukhmani

You've raised a very valid point. Every question provide us an opportunity to explore that concept in details. 

So let's discuss:

See, we can see whether the height of the boys and girls differ significantly or not?

We can also check whether the age of the kids has any effect on their height or not?

But if I wish to check age and height are significantly different or not, does this make sense?

Can we really subtract age (measured in years) from height (measured in cms)?

Like can I check the difference between gender and age? But we can certainly check the difference between the height of boys and girls.

So, plz note different variables can be checked if they have any relation while similar categories may be checked for their differences. This point is the essence of my session on English vs Statistics.

This is the reason we have almost fixed the work relation for any of the techniques (like correlation and Regression) while the word difference is associated with tests (t-test, F-test, H-test, U-test or even Blood test, Sugar test etc)

In the example mentioned by you, plz think can intelligence level be subtracted from the happiness level?

or we can see their relation? like are the more intelligent people happy? or the less intelligent people happy?  (considering that more intelligent people always find reasons to be unhappy ;-)

Hence, in this obj, we can see differences between 'whatever var' among the boys and girls. 

Yes, it could have been to check the effect of intelligence on happiness and then check whether the same holds true for the boys as well as girls? Now, this becomes a multivariate obj as we are checking whether the effect of intelligence on happiness among boys and girls is the same or not? 

But such a thing is not emerging from the present sentence.

Hope this is somewhat clear.

Plz feel free to ask any of your doubts.

Best wishes

--
Protocols of this Group:
 
1. Plz search previous post in group before posing the question.
2. Don't write query in someone's post. Always use the option of New topic for the new question. You can do this by writing to dataanalys...@googlegroups.com
3. Its better to give a proper subject to your post/query. It'll help others while searching.
4. Never write Open ended queries. This group intend to help research scholars NOT FOR WORK THEM.
5. Never write words like URGENT in ur posts. People will help them when they are free.
6. Never upload any info about National Seminars/Conferences. Send such info on personal emails. And feel free to share any RESEARCH related info.
7. No Happy New Year, Happy Diwali, Happy Holi, Happy B'day, Happy Anniversary etc. allowed on this group.
8. Few months back there was a facility for asking & sharing the Research Papers. Now there is no provision of asking for the research paper here.
 
Let’s make a better research environment.
---
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 view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/6a629a1e-cf5d-49e1-a93c-07be4778551bn%40googlegroups.com.

[Sukhmani Singh]

Dear Neeraj sir,

Thank you very much for devoting your valuable time and explaining me the concept minutely. 

Sir, this line “different variables can be checked if they have any relation while similar categories may be checked for their differences” has clarified the things a lot!

I got my answer that:

 “To compare the happiness level and intelligence level of the boy and girl students of MBA in NIT Kurukshetra.”

a) To compare the happiness level of boys and girls. (This bivariate objective is Correct- as similar categories (boys and girl) can be checked for differences).

b) To compare the intelligence level of boys and girls. (Correct- as similar categories (boys and girl) can be checked for differences).

c) To compare the happiness level and intelligence level of boys. (This bivariate objective is Incorrect as – happiness and intelligence are not related, therefore, their difference makes no sense!)

d) To compare the happiness level and intelligence level of girls. (Incorrect as – happiness and intelligence are not related, therefore, their difference makes no sense!).

_______________________________________________________________________________________________________________________________________________

Sir, as you have encouraged towards the end to feel free to ask further….so I am humbly putting some more points that have triggered my mind after going through your enlightening explanation.

Sir, as explained by you -

“In the example mentioned by you, plz think can intelligence level be subtracted from the happiness level?”- Yes, you are right sir, subtracting intelligence from happiness doesn’t make sense.

“or we can see their relation? like are the more intelligent people happy? or the less intelligent people happy?  (considering that more intelligent people always find reasons to be unhappy ;-)”  – As regards relation, Can’t we see their relation like this:

-“are the more intelligent people happy” through correlation, provided that we take IQ scores and Happiness scores i.e. both continuous scale. (High positive correlation- meaning thereby more intelligent people are happy);

-“ the less intelligent people happy” through correlation (High negative correlation –meaning less intelligent people are happy)

So, in the 4 Bivariate objectives:

a) To compare the happiness level of boys and girls. (t-test will give the answer)

b) To compare the intelligence level of boys and girls. (t-test)

c) To compare the happiness level and intelligence level of boys. (Can’t we do this comparison with Correlation, as mentioned above?... Sir, I know the objective says “compare” and for correlation the correct word is “to find relation between happiness and intelligence of boys”.)

d) To compare the happiness level and intelligence level of girls. (Can’t we do this comparison with Correlation, as mentioned above?.... Sir, I know the objective says “compare” and for correlation the correct word is “to find relation between happiness and intelligence of girls”.)

_________________________________________________________________________________________________________________

“Hence, in this obj, we can see differences between 'whatever var' among the boys and girls.” – I got your point, sir.

 ____________________________________________________________________________________________________________________

Lastly,

“Yes, it could have been to check the effect of intelligence on happiness and then check whether the same holds true for the boys as well as girls? Now, this becomes a multivariate obj as we are checking whether the effect of intelligence on happiness among boys and girls is the same or not?”

Sir, how can we solve this objective?

As per my little understanding, by taking Intelligence (High and Low) and Gender (Boy and Girl) as independent variables, and Happiness score as dependent variable, then applying 2-way ANOVA. Then conducting post-hoc tests in order to know the effect of intelligence on happiness among 4 categories (high intt, low intt, boy, girl).

Kindly guide sir as per your convenience.

Thanks and Regards,

Sukhmani.

[Neeraj Kaushik]

Dear Sukhmani

I've read your post 4-5 times and I'm clueless about what I've written and where are your observations?

You could have colour-coded your observations or just ask your queries leaving text written by me.

Asking the correct question using simple & correct words is extremely important.

But, I think I understand what is your confusion (or at least this is what I think). So let me explain a few things here:

1. In my experience of 2 decades with Statistics and Research, I found that people/scholars try to compensate for their "lack of knowledge of Statistics" with English. I saw people play with words and spend hours on which word to use: study/analyze/evaluate/.... or whether to use the work influence/effect/relation of leads to.

I definitely agree that semantically, these are all different words and comes with a context but to Statistics, it does not matter. 

The grammar (rules) of Statistics is very different. It just works in following simple principles:

(i). How many variables are there in obj

(ii). Whether it is Univariate/Bivar or Multivar obj (This is the most important point as merely the presence of 2 var does not make Bivar Obj, there has to be any connector for these 2 vars)

(iii). Role of Vars- IDV, DV, CV, Med, Mod, Marker etc

(iv). Measurement of each var-Metric / Non-metric

(v). Finally, refer Table-1/2/3 to get the appropriate statistical tool/technique (Attached herewith)

Example: I wish to check the relation between happiness and intelligence. 

I can write the same in so many different ways:

(a) To check the effect of intelligence on happiness

(b) To check the impact of intelligence on happiness

(c) To check the relation of intelligence on happiness

(d) To check the influence of intelligence on happiness

(e) To check whether intelligence leads to happiness

(f) To check the intelligence causes happiness

(g) Are intelligent people happy

(h) .......

You may go on playing with the words but let's see how statistics see this:

• There are 2 vars: happiness and intelligence.
• These must be measured separately to retain a Bivar analysis. It shd not be asked on 5 point scale with the statement, "there is an effect of intelligence on happiness". In such a case, this will become Univariate (Think from data entry point of view, how many cols will be required for entering the response for this statement).
• Since the obj claims to check the relation between the two (you may use any word by essence is to check the relation between the 2 vars), it is Bivar obj (To the different words used by us, all Statistics derive meaning is that we wish to see their relation)
• If intelligence is measured by taking IQ level and happiness level measured on a scale then both are metric. If you've no idea of DV & IDV, use Correlation. If you consider happiness as DV and intelligence as IDV, use Simple Regression
• If intelligence is measured with question IQ level-Low/High and happiness level measured on a scale and we consider happiness level as DV and IQ as IDV then use t-test to check whether there is sig diff between the happiness of low and high IQ people. If they are different then IQ and happiness are related.
• If intelligence is measured with question IQ level-Low/High and Happiness level measured on  Low happy/High Happy, then use Chi-square test to check the association between the 2 vars. 
  

I strongly recommend all scholars to first get clear that you may use any word(s) in English but from an Inference point of view, the words Relation and Difference are the same. Plz note the exact use of each word:

• There are 2 vars (no idea of DV & IDV) and both are metric, Use relation
• There are 2 vars (we know the DV & IDV) and both are metric, Use relation or a better word is effect/predict
• There are 2 vars (no idea of DV & IDV) and both are non-metric, Use association
  
• There are 2 vars (we know the DV & IDV) and DV is metric and IDV non-metric, Use difference
  
• There are 2 vars (we know the DV & IDV) and DV is non-metric and IDV metric, Use  effect/predict  

If someone's use any other word (than he/she was supposed to use) even then the choice of the statistical test/technique will be the same. Statistics is never impressed by your English :-)

I've explained the usage of relation/difference and association word in these 2 videos:

https://www.youtube.com/playlist?list=PLzUJUtTJcj8T5UJii0ItrPpkraMNF5fGC

Plz feel free to ask any of your doubts/queries (but in simple words).

Best wishes

To view this discussion on the web visit https://groups.google.com/d/msgid/dataanalysistraining/972e9167-ce6e-4a39-b9be-5ff12fefe072n%40googlegroups.com.

[Neeraj Kaushik]

As regards the doubts related to 2-way ANOVA/MANOVA are concerned, plz read the detailed posts about ANOVA given on this link

https://groups.google.com/g/dataanalysistraining/search?q=%22beyond%20anova%22

[Sukhmani Singh]

Dear Neeraj Sir,

Your every word of explanation is a Bible for learners like us. 

Thank you for explaining the concepts painstakingly. 

Regards,

Sukhmani.