person statistics using snowIRT

Discuss statistics related things
Post Reply
lauramar
Posts: 15
Joined: Wed Oct 27, 2021 11:11 am

person statistics using snowIRT

Post by lauramar »

Hello, would anyone happen to know why I cannot run person statistics for snowIRT dichotomous analysis? I have watched videos where the presenters have that option but I'm using snowIRT version 4.0.0 and it is not available. I'm using jamovi version 2.2.2. Many thanks.
User avatar
seol
Posts: 158
Joined: Mon Dec 04, 2017 8:40 am
Location: Seoul, Korea
Contact:

Re: person statistics using snowIRT

Post by seol »

Hi lauramar

Person statistics will be displayed in the datasheet not result screen. Could you share the example of this issue here? or send me an e-mail:snow@cau.ac.kr.
Cheers
Seol
lauramar
Posts: 15
Joined: Wed Oct 27, 2021 11:11 am

Re: person statistics using snowIRT

Post by lauramar »

Hi Seol

Thank you so much for your quick reply and suggestion, it was just as you said.
And one more question, if I may. Do I understand it correctly that the program does not compute person measures when there is missing data for some of the items?

lauramar
User avatar
seol
Posts: 158
Joined: Mon Dec 04, 2017 8:40 am
Location: Seoul, Korea
Contact:

Re: person statistics using snowIRT

Post by seol »

Hi lauramar

Nope, it depends on the number of sample size and the estimation algorithm to compute item/person measure.

Best
Seol
Last edited by seol on Fri Oct 29, 2021 5:29 am, edited 1 time in total.
User avatar
jonathon
Posts: 2613
Joined: Fri Jan 27, 2017 10:04 am

Re: person statistics using snowIRT

Post by jonathon »

hey seol,

would it be appropriate for snowIRT to automatically remove these missing values?

jonathon
lauramar
Posts: 15
Joined: Wed Oct 27, 2021 11:11 am

Re: person statistics using snowIRT

Post by lauramar »

seol wrote:Hi lauramar

Nope, it depends on the number of sample size and the estimation algorithm to compute item/person measure.

Best
Seol
Hi Seol,

I would like to ask about the results of the analysis. I ran the same data in jMetrik using Rasch and the results are different. Would you be able to comment how I should interpret this outcome? My background is in language assessment and I don't understand why this might happen. Could it be related to the format of the data? By this I mean that in Jamovi I used correct and incorrect responses (0 for incorrect and 1 for correct) but in jMetrik I used the responses that were chosen by the students (eg, A, B or C in case of multiple choice questions) and specified the correct one for the analysis.

Many thanks in advance,

Lauramar
User avatar
seol
Posts: 158
Joined: Mon Dec 04, 2017 8:40 am
Location: Seoul, Korea
Contact:

Re: person statistics using snowIRT

Post by seol »

Hi Lauramar

snowIRT module uses Marginal Maximum Likelihood estimation to calculate item and person statistics with Rasch model and the estimation method is totally based on the TAM package in R. Different estimation method will produce differernt values.

With Regards
Seol
lauramar
Posts: 15
Joined: Wed Oct 27, 2021 11:11 am

Re: person statistics using snowIRT

Post by lauramar »

Hi Seol,

I would like to ask for your advice on how to include certain participants in my analysis. I have a group of people who took a language test. The test has 6 tasks. Most of the participants took the test with all six of the tasks. However, some participants were only given tasks 1-3. I want to use snowIRT to look at item difficulty. What do I do with the participants who only saw tasks 1-3? Can I include them in the analysis?

Many thanks in advance
Lauramar
User avatar
seol
Posts: 158
Joined: Mon Dec 04, 2017 8:40 am
Location: Seoul, Korea
Contact:

Re: person statistics using snowIRT

Post by seol »

Hi Lauramar

snowIRT module treated missing values such as like na.,omit(data) way. . . you have to select cases before running analysis.

Cheers
Seol
Post Reply