Last updated on: 14 August 2016

 

Here are some thoughts and ideas which involve the application of measurement theory and methods.  The suggestions below might be suitable topics for a dissertation.

 

(1) Develop a scale to measure a construct of interest to you.

 

An example would be work done by a PhD student at Boston University in the year 2010.  The student's major was "sports science".  His particular interest was in a construct called "Mental Toughness".

 

Read his research proposal by clicking here.  See his initial pool of items here.  An Excel workbook with item responses collected during pilot testing is available in Lertap 5 format here.

 

A topic very much related to the student's work involves a construct called "Mindset", developed by Professor Carol Dweck.  Read this very interesting introduction to some of Professor Dweck's work by downloading this pdf document.  The Mindset construct has been well received, and has been the subject of much research in many countries.

 

As an RMCS postgraduate student, what you might consider is the idea of developing a new instrument (in the Thai language, of course) to measure a construct similar to Mental Toughness or Mindset.  This you would do by developing a pool of items, assessing the quality and validity of your items by referring to professional judges, pilot testing the items and assessing their performance with CTT (perhaps using Lertap 5), and then, possibly, applying IRT methods with the goal of developing an online "CAT" delivery system.  Such work would parallel that done at RMCS by Aj. Piyathip Pradujprom and Aj. Suchada Sakolkijrungroj.

 

(2) Extend the research on the limitations of coefficient alpha.

 

Alpha has been a dominant measure of reliability and internal consistency for many years.  However, it has also been the subject of much criticism, with some scholars recently demonstrating that McDonald's omega coefficients are superior.

 

I recently completed a bit of research which compared alpha and omega results for several datasets; read about it here.  I have not had time to extend my research -- what might be interesting is to look further into my comparison of the scree test with MacDonalds omega(w) index.

 

This area of research involves factor analysis, and the R programming language.

 

(3) Extend the research which has to do with IRT goodness of fit measures.

 

There is some fresh research (2015) which suggests that we should consider the use of a new goodness of fit method when using IRT models.  It's called the "PPC" method, "posterior predictive checks".

 

This is an area which needs more research.  The reference is here.  I have a printed copy of the complete paper -- please ask me for it it you're interested.  This is another topic which would probably involve the use of R.

 

(4) Create a Thai version of Professor Baker's introduction to IRT book.

 

Baker, F.M. (2001). The basics of Item Response Theory. ERIC Clearinghouse on Assessment and Evaluation.

Download a PDF copy here

 

In my opinion, this remains an excellent resource, but there are many places where it could be updated.  The numerous exercises in the book, for example, need to be refreshed -- they could be done with Excel.  The book discusses dichotomous items -- it should be updated to included polytomous items.