Rasch measurement using user-friendly jMetrik | Powerful free software

Hi, is it possible read the data and in the end give a score from 0 to 100 in an assement with 45 question done by 30 students?

[@sajeelsaeed3388]

For dicotomous variables, it is running but for polycotomous, error popped. What to do

[@alesiapapa2875]

why for me is always ERROR?

[@soehartosoeharto8471]

Thanks prof, i hope the other Jmetrik tutorials form iRT will be avalaible in future

[@LauraRODRIGUEZRODRIGUEZ-ej4wo]

First thank you very much for the video, I have a question in a questionnaire with items with a response scale with values 0 to 100, how can you use the jmetrik program?

[@pabitrapradhan721]

Hello dr i am just wondering how to deal with ties in dataset when doing shapiro-wilk test

[@VahidAryadoust]

I am not sure if I understand the question. Do you mind clarifying?

[@frohnzy04]

Hi Vahid, Brilliant video. Thank you. I hadn't heard of this programme before. Looks straightforward and prefer the GUI and output organisation. Do you have or can you do a video on comparing test versions with jmetric? Many thanks again.

[@rogerturingan6141]

Good day sir! If the eigenvalue of the first contrast is greater than 2 but less than 2.5... is it still safe to say that there is an evidence of unidimensionality?

[@VahidAryadoust]

It seems to be falling on the unidimensional side.

[@rogerturingan6141]

I see. Thank you Sir.

[@faisalmustafa6035]

Dear Prof. I am desperate to know how ETS uses the results of 3PL model analysis in converting the score into scaled score for TOEFL. I have spent like one month learning from the internet and youtube. I am not from statistics major, but English language teaching, but I completed an introduction to statistics course and I know a bit of R. Please kindly create a video for that. Thanks for your help.

[@VahidAryadoust]

I am not sure how they do it, but the following paper might provide some relevant information: www.ets.org/Media/Research/pdf/RR-13-05.pdf

[@matthewcourtney8683]

There are various IRT models. When learning IRT, researchers are generally introduced to the one, two, three, and four parameter logistic models. The Rasch model (or 1plm) is the most basic and transforms students' percentage correct scores to a special type of log-odds transformation (sometimes called, theta). In this case, the students who get the same percentage score get the same theta estimate. There is one parameter for this model which is the item difficulty parameter (b). For the two parameter logistic model, students theta depends on the questions that they get correct. If they happen to get questions that have a higher discrimination index (for basics see item-total correlation), their score is weighted higher. With this model, we loose the property of "same percentage correct score, same theta". For this model, there are two parameters, item difficulty (b parameter) and item discrimination (a parameter). For the three parameter logistic model, a third parameter is introduced to account for guessing. With large datasets, we can actually estimate the probability that the lowest performer might guess an item to be correct despite having zero knowledge. Of course, these might be around 0.25 for four response options but depending on the quality of distractor response options, this can vary a little for each item. So, the use of the 3plm generates person ability estimates that account for both item alignment with the construct (discrimination, a parameter), and the test takers' propensity to guess (the c parameter for each item). The theta estimate provides an estimate of student ability aligned with what the student might "know" as opposed to "get correct". There is some push back against researchers in the 3plm camp as the the c parameter can influence the a parameter when the model is fit. I'm a bit of a Rasch camp guy as it's more simple for stakeholders and accessible. Our free and open-source cloud-based software handles dichotomous, partial credit, and mixed format scoring. It performs unidimensional Rasch, many-facets, fixed anchor equating, ANOVA and t-tests for ability estimates, and inter-rater reliability tests. Feel free to check out our free cloud-based software here: autopsych.shinyapps.io/version_1_0_0/ with exposition in journal here: journals.plos.org/plosone/article?id=10.1371/journal.pone.0257682

[@ericasare8864]

Hi Prof. can the jMetrik be used to analyze Likert data?

[@VahidAryadoust]

Yes, it can handle such data, too. It is more or less the same procedure as in dichotomous data.

[@matthewcourtney8683]

Our free and open-source cloud-based software handles dichotomous, partial credit (Likert style), and mixed format scoring. It performs unidimensional Rasch, many-facets, fixed anchor equating, ANOVA and t-tests for ability estimates, and inter-rater reliability tests. Feel free to check out our free cloud-based software here: autopsych.shinyapps.io/version_1_0_0/ with exposition in journal here: journals.plos.org/plosone/article?id=10.1371/journal.pone.0257682

[@angelaturingan4699]

Thank you for the useful video sir, may I ask if jMetrik also give PTMEA correlation?

[@VahidAryadoust]

From what I can recall, it does not provide PTMEA.

[@angelaturingan4699]

@@VahidAryadoust Ok sir.. thank you. in item analysis option, i can see Pearson correlation and the output has something to do with discriminant index, do you have an idea how to interpret those values?

[@VahidAryadoust]

@@angelaturingan4699 Pearson correlation = the correlation between the items. Discriminant index = in 2PL this index refers to item discrimination. In Rasch analysis, please ignore it.

[@angelaturingan4699]

@@VahidAryadoust oh.. thank you sir and it is now clear to me, does that correlation corresponds to item pair residual correlations for the assumption of local dependence?

[@VahidAryadoust]

@@angelaturingan4699 You are right; it is more precise to say the correlations correspond to the Q3 coefficients or correlation between residuals --> measures of local independence.

[@RLForester]

Thanks! Very helpful!

[@soehartosoeharto8471]

Hope racking will be release soon, hihi thanks prof

[@matthewcourtney8683]

Our free and open-source cloud-based software handles dichotomous, partial credit, and mixed format scoring. It performs unidimensional Rasch, many-facets, fixed anchor equating, ANOVA and t-tests for ability estimates, and inter-rater reliability tests. Feel free to check out our free cloud-based software here: autopsych.shinyapps.io/version_1_0_0/ with exposition in journal here: journals.plos.org/plosone/article?id=10.1371/journal.pone.0257682

[@ansafaqeer3201]

Thanks a lot.

[@matthewcourtney8683]

Our free and open-source cloud-based software handles dichotomous, partial credit, and mixed format scoring. It performs unidimensional Rasch, many-facets, fixed anchor equating, ANOVA and t-tests for ability estimates, and inter-rater reliability tests. Feel free to check out our free cloud-based software here: autopsych.shinyapps.io/version_1_0_0/ with exposition in journal here: journals.plos.org/plosone/article?id=10.1371/journal.pone.0257682

[@JosephFabs]

jmetrik looks better for me :) Thank you