Item Response Theory

[Neeraj Kaushik]

Dear Friends

Item Response Theory (IRT) is a modern psychometric approach used in scale development that focuses on how individual items perform across different levels of the underlying trait. Unlike Classical Test Theory (CTT), which depends on total test scores and sample-specific statistics, IRT provides item-level parameters (difficulty, discrimination, and guessing) that are largely sample-independent. This allows for more precise measurement, better item selection, and fairer comparisons across groups. As a result, IRT leads to more reliable, valid, and flexible scales, especially useful in adaptive testing and advanced measurement applications.

I took an online session for a selective audience this past Sunday. The same was recorded for the learning of other scholars too. I am sharing the same here.

In the first video, I discussed the fundamentals of constructs and different scale types (like Summated Rating Scale, Semantic Differential Scale, Cumulative Scale).

Item Response Theory Part-1 Construct Fundamentals: https://youtu.be/gfAJoVRXI9c

Happy Learning

Neeraj

[Neeraj Kaushik]

Dear Friends

In the second video, I discussed the computation steps for Exploratory Factor Analysis (using Principal Component Analysis) and differentiated between the terms factor and construct. 

Important thing to remember is that in Item response theory, we use the Principal Axis factoring method for computing Exploratory Factor Analysis.

Item Response Theory Part-2 EFA (PCA) Fundamentals: https://youtu.be/r09YkJNXE2M

Paper on Scale Development:

Student well-being in higher education: Scale development and validation with implications for management education

Happy Learning

Neeraj

[Neeraj Kaushik]

Dear Friends

In the third part, I've discussed the fundamentals of IRT:

A) Quality Decisions: Check the data for

1. Missing values

2. Improper Codes

3. Reverse Coding

4. Anchor points must be from lower to high agreement

5. Uni-dimensionality

 

B) IRT Models: When to Use Which

IRT models differ by item type and number of parameters

 

1PL (Rasch Model)

·         Difficulty only

·         Equal discrimination

Use when:

·         Fair, objective measurement

·         Small samples

·         Early scale development

 

2PL Model

·         Difficulty + discrimination

Use when:

·         Item quality varies

·         Psychological and educational tests

 

3PL Model

·         Difficulty + discrimination + guessing

Use when:

·         Multiple-choice, high-stakes exams

·         Guessing affects scores

 

Graded Response Model (GRM)

·         For ordered polytomous responses

·         Estimates discrimination and category thresholds

Use when:

·         Likert-type scales (e.g., attitudes, personality)

·         Responses have a clear order

 

Partial Credit Model (PCM)

·         Rasch-based polytomous model

·         Step difficulties vary, equal discrimination

Use when:

·         Rating scales with partial scoring

·         Constructed-response or rubric-based items

To begin with we will use the graded response model.

 

C) IRT Parameters:

•  Discrimination Power – Indicates how well an item distinguishes between individuals with low and high levels of the trait.
•  Thresholds – Points on the ability scale where the probability of choosing one response category exceeds the previous category.
•  Fit Statistics (Item fit & Person fit) – Measure how well items and respondents’ response patterns conform to the IRT model.
•  Plot – Trace – Shows the probability of selecting each response category across different levels of ability.
•  Plot – Item Information Curve (IIC) – Displays how much measurement precision an item provides at different ability levels.
•  Wright Map – A joint display of person abilities and item difficulties on the same scale.

Item Response Theory Part-3 IRT Basics: https://youtu.be/LAYQFr2x5DY

 

Happy Learning

Neeraj

[Neeraj Kaushik]

Dear Friends,

In the latest video, I explore the concept of Item Difficulty within Item Response Theory (IRT). Unlike traditional methods, IRT defines difficulty as the point on the latent trait scale where a respondent has a 50% probability of endorsing a specific category or answering correctly.

Understanding item difficulty is crucial for creating balanced assessments and ensuring that scales effectively differentiate between individuals at various levels of the underlying construct. This session covers how difficulty parameters are estimated and interpreted across different IRT models.

 

Item Response Theory Part-4 Understanding Item Statistics and Item Difficulty:
https://youtu.be/sKEga9A9zbM

Happy Learning

Neeraj

[Neeraj Kaushik]

Dear Friends,

In continuation of our series on Item Response Theory (IRT), the next video covers essential initial assumptions that must be validated before proceeding with modeling.

In this session, I explain how to properly check your data for missing items, analyze critical item statistics, and verify the crucial assumption of uni-dimensionality. Ensuring these steps are completed is vital to confirm your data is suitable for IRT analysis and will lead to reliable and valid measurement results.

Item Response Theory Part-5 Initial Assumptions of IRT using R-package mirt: https://youtu.be/ns64Bstkgok

Happy Learning

Neeraj

[Neeraj Kaushik]

Dear Friends,

In the next video, I have explained key IRT concepts including 

• item discrimination, 
• thresholds, and 
• fit statistics like infit and outfit
• essential visualizations such as Information Curves and Trace plots. 

These tools help determine how effectively each item distinguishes between respondents and where measurement precision is highest. 

Item Response Theory Part 6: Practical Application of IRT using the R package mirt

https://youtu.be/XlDzd6R0Vgo

Happy Learning

Neeraj

 

[Neeraj Kaushik]

Dear Friends,

Most scholars fear coding in R or Python. So here is the good news. Winstep is a GUI software specifically designed for the IRT (Rasch Analysis). Since the software is paid, its mini version MINISTEP is available free of cost. MINISTEP can handle 25 items and 75 persons. 

Download MINISTEP from this link  MINISTEP 5.10.4

 
This following video focuses on the working mechanics of the IRT, providing a step-by-step guide on how to proceed:

• How to input data
• Run the analysis, and
• Interpret the basic output.

Item Response Theory Part-7 Practical Working on IRT using MINISTEP: https://youtu.be/XlDzd6R0Vgo

Happy Learning
Neeraj

[Neeraj Kaushik]

Correct link:

Item Response Theory Part-7 Practical Working on IRT using MINISTEP:

https://youtu.be/rwO6gBMFPAo

[Neeraj Kaushik]

Dear Friends,

In the latest video of our series, I demonstrate how to perform Item Response Theory (IRT) analysis using MINISTEP, the free version of the Winsteps software. This session is specifically designed for those who prefer a graphical user interface over coding in R or Python.

Download MINISTEP from MINISTEP 5.10.4

I provide a step-by-step walkthrough on:

- Importing your data.

- Executing the analysis.

- Interpreting the primary DIAGNOSIS.

IRT using MINISTEP-1: https://youtu.be/dtV-wpX-UDE

Happy Learning,

Neeraj

[Neeraj Kaushik]

Dear Friends,

In the next video on Item Response Theory (IRT) using MINISTEP, I continue demonstrating the practical application of this powerful software.

This session focuses on two key areas: 

• explaining the concept and utility of the Wright Map, and 
• demonstrating how to quickly execute the entire IRT analysis workflow for a new construct.

IRT using MINISTEP-2: https://youtu.be/JARPlJ2jYa8

Happy Learning,

Neeraj