In general, there is no rule how to present your validity and reliability results in your thesis. But at least you should put the results of the statistical tests you utilized.

In general, there is no rule how to present your validity and reliability results in your thesis. But at least you should put the results of the statistical tests you utilized.

Internal validity of research need to be planned during the research methodology. I am enclosing a table that shows the various threats and how to avoid. This part should be specified in the chapter on your research methodology. You may also mention during the data collection process if any of these were present or not. If they are there then there is a possibility that the data will not be valid as you originally planned.

Dear @Fatimah, there is no golden form for the thesis; however, you have to keep reading literature and theses in your field and always try to reflect that on your own work to reach a good form for your thesis.

Most probably you will find some guidelines for your particular research in the brand-new "Standards for Educational & Psychological Testing" of the American Educational Research Association. Also see the famous APA Guidelines for Publication, one of the best in its area. A much more general overview is in the attached paper

http://www.aera.net/Publications/Books/StandardsforEducationalPsychologicalTesting(NewEdition)/tabid/15578/Default.aspx

I think you can follow your school guidelines or according to your supervisor

thanks all... Prof. Rao, thank you for your original piece. God bless.

The matrix table suggested by @ nageswara is good to follow.

However, the results validity and reliability depends on the LIFE SPAN in which it is studied. Hypothesis are nothing but "Question forms of Results",

The linkage of Hypothesis with the situational correlation would bring in appeal to your research results....

"Thou Shalt Not Refer to the Validity of the Test" (see attached paper)

Why not? Read the astonishing results of an empirical study of Newton and Shaw (2013): there are more than 30 different VML's and in total they found more than 120 different 'validity' constructs (see figure).

This means the concept of validity is not a single one concept and it is very often not even well-defined (in the research literature; in textbooks about testing you will often find just a few of the simpler concepts).

Why this is so and how the concept evolved and gave birth to a multitude of distinct families and flavors of validity is eloquently described in their book which came out very recently (see link).

For you, this means: please define clearly what kind of validity you are referring to and using. If you are working in the context of CTT (Classical Test Theory), you won't have problems of finding good text books with a lot of examples of what & how to report (see e.g. the classical work of Lord & Novick). If you are working with construct validity, you are in good company, but things become a bit messier (see Messick et al) (pun intended)

http://www.amazon.de/dp/1446253236

@ Demetris :

Dear Paul, I do agree with you.

Consider for example the very often problem of multicollinearity.

In a recent discuss I argued to somebody to do it by hand, ie to take his 10 independent variables X's as vectors and create some sub-matrices in order to see which variables are collinear (by computing the ranks of those matrices)

He responded to me that he could put them in SAS and do everything...

(But next day he gave it as an exercise to his students:)!)

So, the 'dirty work' has to be done by hands and not by packages.

Thanks Paul, Krishnan and Demetris. Paul, I 'll put that in mind. I will need to confirm the validity / reliability I wanted to report.

Excuse me Fatima for placing my own questions, I like to read your questions and answers because I myself work with research, but I don’t have so much experience as some of the lectures who answer you.

Now Demetris and Paul’s answer poses me some questions…

According to your first point Demetris:

What statistical methods does not depend en on the wrong data entry? And what do you refer to wrong data? Mistakes one makes when entering the data, or the data that was gathered was no the done in the right way?

An according to Paul’s first point:

I was not taught as a student that the validity coefficient had to be less than the reliability. Why is that so? Should they not be both high? What difference does it makes if the validity is higher than the reliability, how that that influence the instrument or the data that is being analyzed?

Also what more is there to validity than just calculating the coefficient? What else dose one have to calculate or take into account?

Maybe by locating it in the Methodology as a sub-part of Instrumentation. It may be presented simply by stating the final validity and reliability values.

@ Hazel Catherina Flores Hole :

Of course, both reliability and validity (as numerical coefficients) should be as high as possible. However, I was taught that the validity coefficicient can't be higher than the reliability coefficient, or at least is strictly bounded by it. This applies to CTT (i.e, is a mathematical theorem of CTT), other test frameworks and measures may of course behave differently.

Validity methodology is complex, it comprises qualitative and quantitative methods and techniques. CTT is a very special sort of testing which by its very assumptions (sic!) reduces validity to a coefficient. However, CTT isn't the last word in test methodology. Many people claim that construct validity (one form or other) is state of the art, best practice. Comes with coefficients, no question, but has much more to say.

Dear Hazel Catherina Flores Hole, even if we are absolute careful, some wrong data will be entered in our analysis, thus the choice of 'robust' statistical techniques will help us to avoid the negative influence of an isolated wrong measurement to our whole data set. For example, the median is a better estimator of the central tendency than mean, since the latter is affected by the outliers.

Thank you both! Demetris you have reminded me that I had learnt that in class. 

Paul, can you translate what CCT means, so I can translate it into my language and look up more information?

Thanks again!

Reliability

This is normally based on two assumptions:

1. The study can be repeated

2. Two or more people can have similar interpretation by using the categories and procedures you used.

Reliability of a research work will then been seen in:

a. How precise and concise your research objectives are stated.

b. How measurable your hypotheses are.

c. How meaningful your data gathering processes have been.

d. How meaningful your sampling size is in relation to the population size.

e. How meaningful your data sampling techniques are

f. How you admit your shortfalls in the work e.g. Noise from outside affected your recording so you did not hear some of the thing recorded etc.

Validity

This necessitates that the propositions generated and tested, match the causal conditions that exist.

In this you ask may ask:

1. Does my research measure what I purport it to measure? (i.e. internal validity)

2. To what extent are the abstract constructs and postulates generated and tested by my scientific research applicable across groups? (i.e. external validity)

This I believe you would have done by:

1. Using two or more methods of data collection in your study

2. Checking the consistency of findings generated by different data collection methods

3. Checking out the consistency of different data sources within the same method

Thanks for your questions Hazel, and thanks for your responses Paul and Demetris. Thanks too to Eddie and Dennis.

apologize for late reply, now back from holidays: CTT := Classical Test Theory, i.e. complete formalization of the simplest statistical theory of test responses. The 1968 bible of CTT is:

http://www.amazon.com/Statistical-Theories-Mental-Test-Scores/dp/1593119348/

If a research uses a Case study (for social sciences) then

The validity and reliability depends on the scenario approach, acceptance and rejection. It is also important to consider the robustness of the statistical analysis (sample size, variables, correlation, variance, standard error, statistical significance of the model used in the analysis, reproducibility of results, etc.)

Reliability

This is normally based on two assumptions:

1. The study can be repeated

2. Two or more people can have similar interpretation by using the categories and procedures you used.

Reliability of a research work will then been seen in:

a. How precise and concise your research objectives are stated.

b. How measurable your hypotheses are.

c. How meaningful your data gathering processes have been.

d. How meaningful your sampling size is in relation to the population size.

e. How meaningful your data sampling techniques are

f. How you admit your shortfalls in the work e.g. Noise from outside affected your recording so you did not hear some of the thing recorded etc.

Validity

This necessitates that the propositions generated and tested, match the causal conditions that exist.

In this you ask may ask:

1. Does my research measure what I purport it to measure? (i.e. internal validity)

2. To what extent are the abstract constructs and postulates generated and tested by my scientific research applicable across groups? (i.e. external validity)

This I believe you would have done by:

1. Using two or more methods of data collection in your study

2. Checking the consistency of findings generated by different data collection methods

3. Checking out the consistency of different data sources within the same method

In my opinion, the procedures for conducting reliability and validity testing should be in the "Research Methodology" section of the thesis. The results of the reliability and validity testing should be in the ensuing "Findings and Analysis" section of the thesis.