Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Topics - Georg II

Pages: [1] 2
Questions and Answers / ConQuest Output Question
« on: April 03, 2017, 06:39:20 AM »
Dear All,
In ConQuest output under the variance/covariance matrix you can find the variance for each dimension. Next to each variance between parentheses there is another value. What is that value? The manual doesn’t say anything about it.
Georg II


Dimension                    1                 2                 3                 4                 5

Dimension_1                                  0.000*            0.000*            0.000*            0.000*         
Dimension_2                0.000*                              0.000*            0.000*            0.000*         
Dimension_3                0.000*            0.000*                              0.000*            0.000*         
Dimension_4                0.000*            0.000*            0.000*                              0.000*         
Dimension_5                0.000*            0.000*            0.000*            0.000*                           
Variance                   0.182  ( 0.009)   0.240  ( 0.012)   0.401  ( 0.020)   0.273  ( 0.013)   1.523  ( 0.074)

Questions and Answers / Different item estimates across models?
« on: January 17, 2017, 05:42:54 PM »
Dear colleagues,

I’d be thankful if you replied this question.
I separately fitted two models to a data set: a unidimensional and a within-item two dimensional model. The item difficulty estimates of the items which simultaneously load on a second dimension in the two-dimensional model differ noticeably from the values they have in the unidimensional model. The items which load on only one dimension have the same estimates across the two analyses. What does this mean?
Thanks for your help

Questions and Answers / Multidimensionality & reliability
« on: January 26, 2014, 09:43:46 AM »
I have a scale with 25 items. I fitted a unidimensional model and a 5-dimensional model.
The unidimensional model fits significantly better. The correlations between dimensions are above .95 in the multidimensional model.
The strange thing is that EAP reliability of the unidimensional scale with 25 items is .82 while the EAP reliabilities of the subscales, each having only five items, are above .92. I could understand it if the multidimensional model had a better fit. Because then one could argue that forcing irrelevant items which belong to five different scales into a single dimension lowers reliability. But when the unidimensional model has a better fit why should it have lower reliability with 25 items than subscales which have only five items?
Isn’t this strange?

Georg II

I have analysed the same test once with a unidimensional model and once with a multidimensional model using ConQuest. In the two dimensional model the items are assigned to two dimensions. The two-dimensional model fits significantly better. However, item difficulty parameters are very different in the two models. What does this mean?
I remember I read somewhere that when item parameters across the two models are very different, it means that the better fitting model has done a better job of accounting for the data and was necessary for modeling the data. Is that right? I can’t locate the reference now. Would be thankful if you directed me to the appropriate source.

Georg II

Questions and Answers / Interpreting MNSQ and T fit indices
« on: August 19, 2013, 12:20:38 PM »
Dear colleagues,
How should weighted and unweighted MNSQ and T fit indices be interpreted in ConQuest?
Are the conventional .70-1.30 for MNSQ and +-2 for T applicable?
How should the confidence intervals (CI) be used? Do they help in deciding on the fit of individual items or just help have a feel of the accuracy of the MNSQ indices?


Questions and Answers / Common scale in multidimensional analyses?
« on: August 19, 2013, 12:18:24 PM »
Dear colleagues,
Are person and item parameters in a multidimensional analysis in ConQuest on a common scale?
If the mean of persons on dimension 1 is .50 and on dimension 2 is .90 can we conclude that dimension 1 is harder or persons are more proficient on dimension 2?


Questions and Answers / Step calibrations on the map
« on: October 26, 2012, 07:19:44 AM »
Dear all,
In a unidimensional analysis of a Likert scale with both rating scale and partial credit model, ConQuest does not provide the step calibrations on the Wright map. I get only the estimates of the items and persons.
The title of the map is " Terms in the model (Excl step terms)". Do I need a certain command to get the step calibrations on the map too?

Georg II

Questions and Answers / EAP reliability
« on: October 25, 2012, 08:21:27 AM »
Hallo Ray,
Can you please explain EAP reliability?
How is it computed? Can you please explain its relation with alpha and its features.
Please provide a source so that I can read up on it.

Georg II

Questions and Answers / Within-item multidimensional models
« on: July 26, 2012, 12:25:20 PM »
Dear Ray,

In within-item multidimensional models if we make an item load on say, five or more dimensions, aren’t we running any risks?

What consequences does it have if we make items load on 6-7 dimensions (in terms of convergence, computation, sample size requirements, fit, precision of estimates, etc.)?
I just feel that there must be some limitations from a practical point of view (if not theoretically).
Can you please comment?

Georg II

Questions and Answers / EAP file
« on: July 24, 2012, 10:08:11 AM »
Hi Ray,
The EAP estimates file contains three columns for each dimension.
Can you please expalin what each colum refers to?
Under what conditions do we prefer to use EAP ability estimates instead of MLE or WLE?


Questions and Answers / EAP's and PV's
« on: July 23, 2012, 12:13:53 PM »
1. How are EAP ability estimates different from MLE and WLE?
2. When do we used EAP estimates?
3. How are PV's different from WLE, MLE, and EAP?
4. When do we used PV's?

Georh II

Questions and Answers / When constarin correlations to zero?
« on: July 01, 2012, 07:20:08 AM »
Constraining correlations to zero changes the difficulty and ability parameter estimates slightly and the deviance substantially. When dimensions are free to correlate fit is better (deviance is smaller) compared to when correlations are set to zero.

When and under what conditions should we constrain correlations to zero?


What's the reason of convergence failure for cases?
In a within-item multidimensional model ConQuest gives a message that there convergence failure. The person parameters are some unreadable mixture of some large numbers and characters, as shown below.
What should I do to get person ability estimates?
Georg II

1.#QNAN   -1.#IND0    1.#QNAN   -1.#IND0    1.#QNAN   -1.#IND0    1.#QNAN   -1.#IND0    1.#QNAN   -1.#IND0    1.#QNAN   -1.#IND0    1.#QNAN   -1.#IND0
0.30120   -1.#IND0    1.25060   -1.#IND0    0.30120   -1.#IND0    0.30120   -1.#IND0    1.25060   -1.#IND0    1.25060   -1.#IND0   -5.60935   -1.#IND0
13.10653 47453132.81213   12.83898 47453132.81213   13.10653 47453132.81213   12.83898 47453132.81213   12.83916 47453132.81213   12.83916 47453132.81213  -12.79817 47453132.81213

Questions and Answers / Model comparison with zero correlations???
« on: June 29, 2012, 05:43:40 AM »
Hallo Ray,
I’m comparing the fit of two models: a 5-dimensional within item model and a 4-dimensional between item model.

1.Are these two models nested? Can deviance be used to compare them?

In the 5-dimensioanl model I have set all the correlations equal to zero while in the 4-dimensional model, dimensions are allowed to correlate. Deviance shows that the 4-dimensional model fits better.
However, when in the 4-dimensional model the correlations among the dimensions are set to zero, the 5-dimensional model fits better. 

2.Which model fits better, the 4-dimensional or the 5-dimensional?

I just thought that in order to make all the conditions equal to make a fair comparison I have to set the correlations in the 4-dimensional model to zero, too. Besides, the dimensions in the  4-dimensional model are test format dimensions. We don’t expect test formats (for example, multiple-choice and gap-filling) to be correlated.

3.Does this make sense?

Georg II

Questions and Answers / zero correlation
« on: June 27, 2012, 04:59:07 AM »
Hello Ray,
In multidimensional models what happens if we force all correlations among dimensions to be zero?
How is this different from the (same analysis) when we allow the dimensions to correlate?

Georg II

Pages: [1] 2