A Goodness-of-Fit Measure for the Mokken Double Monotonicity Model that Takes into Account the Size of Deviations

Size: px
Start display at page:

Download "A Goodness-of-Fit Measure for the Mokken Double Monotonicity Model that Takes into Account the Size of Deviations"

Transcription

1 Methods of Psychological Research Online 2003, Vol.8, No.1, pp Department of Psychology Internet: University of Koblenz-Landau A Goodness-of-Fit Measure for the Mokken Double Monotonicity Model that Takes into Account the Size of Deviations Teresa Rivas Moya 1 Málaga University. Spain Based on Mokken model and Isotonic Regression, Rivas Moya (2000b) gives a Global Deviation (GD) measure from the Double Monotonicity (DM). This paper illustrates the GD measure and gives the procedure to calculate it. Several examples from responses of 294 subjects to 16 dichotomous items showed (1) the procedure which calculates the GD measure from and P 11 matrices obtained by MSP5 (Molenaar et al., 2000) (2) GD is 0 if there are no individual deviations from DM (3) an increase in the number and/or size of individual deviations from DM leads to an increase in GD measure when accumulative scales with different numbers of items are considered. The principal advantage of this measure over other indices which evaluate DM is that the procedure to calculate GD estimates the size of deviations from DM. It also takes into account the number of deviations because it obtains the measure by summing up the deviations in each item pair. This measure provides interesting complementary information to the set of indices which evaluate the DM model. Keywords: Goodness of fit, scaling, nonlinear regression Within the framework of the non-parametric item response theory, Mokken (1971, 1997) defines monotone homogeneity (MH) and double monotonicity (DM) models for dichotomous items. A set of items that satisfies unidimensionality and local independence, and whose item response functions (IRFs) are non-decreasing monotone, verifies the assumptions of the MH model. If, in addition to the foregoing conditions, IRFs do not intersect, the set of items also verifies the assumptions of the DM model. Checking whether a set of items satisfies the assumptions of the DM model is laborious. This means that in practice the fit of the model was not viable until Molenaar, De- 1 Address: Teresa Rivas Moya. Departamento de Psicobiología y Metodología. Facultad de Psicología. Universidad de Málaga. Campus de Teatinos. Málaga-29071, Spain. moya@uma.es

2 82 MPR-Online 2003, Vol. 8, No. 1 bets, Stsma and Hemker (1994) developed the software (MSP 3.0) to check if a set of items satisfies DM. Later, Molenaar, Stsma, van Schuur and Mokken (2000) developed a new improved version (MSP5 for Windows) adding news indices. Previous references together with Stsma and Junker (1996), Stsma (1998), Molenaar and Stsma (2000), Stsma (2001), Stsma and Molenaar (2002) give an overall idea about these models, the evaluation indices and their applications. As well as the indices proposed by Mokken (1971) and Rosenbaum (1984, 1987), Molenaar et al. (1994), Molenaar and Stsma (2000) also set out indices which evaluate whether a set of items satisfies the MH or DM. These authors suggest additional research regarding the detailed study of some of these evaluation indices. To evaluate if IRFs are non-decreasing monotone or satisfy single monotonicity there are: - Scalability coefficients based on the analysis of each item ( i ) or a set of items ( H ) (Mokken, 1971). - Indices for each item obtained in the entire group. H, item pairs ( H ) - Indices for each item obtained in rest score groups. These groups are defined on the scores of the remaining items. Given an increasing rest score r, the proportion of positive responses must be monotonically non-decreasing in r (Molenaar et al. 1994, p. 9; Rosenbaum, 1984). - Diagnostic Value Crit which summarize information obtained through the checking of the single monotonicity via the entire group, enabling the identification of items least fitted by the MH model (Molenaar & Stsma, 2000, pp. 49, 74). Indices which evaluate the non intersection of IRFs are usually based on the analysis of item pairs. They check the non intersection via: - P-matrices by visual inspection (Mokken, 1971) and by a count of the violations. Given an ordering on the items, a MH set of items is DM when columns and rows in P 11 matrix are monotonely non-decreasing, and columns and rows in matrix are monotonely non-increasing. Local deviations from these orders are considered violations from DM. - Indices for item pairs obtained in rest score groups. These rest score groups are determined on the remaining items. The proportion of positive responses on an item should be smaller than or equal to that of the other item in each rest score group (Molenaar et al., 1994, p. 10; Rosenbaum, 1987).

3 Rivas Moya: Goodness of Fit Measure for Mokken Model 83 - Indices for item pairs obtained in rest split groups. These groups are determined by distinguishing between counts for the low and high groups that are based on the use of cut points. (Molenaar, 1991; Molenaar & Stsma; 2000; Stsma & Junker; 1996, pp ). - Diagnostic Value Crit which summarize information obtained through the checking of the non intersection via rest score groups and via P-matrices, enabling the identification of items least fitted by the DM model. - T H a and T H coefficients. T H checks if a set of items have intersecting IRF. for the total set of items based on the transposed data matrix and H T a coefficients on the level of individuals are determined. Rules of thumb for their interpretation were based on results from a study using simulated data (Stsma & Meer, 1992). T H - Graphical representations of IRF can be shown and can be seen if pairs of items intersect. These indices enable a detailed analysis to ascertain whether or not IRFs are monotone and intersect. If, from these indices, we concentrate on those that evaluate the non intersection of IRFs, based on item pairs and on the analysis of P-matrices, the final decision is generally based upon: 1a Violations that are evaluated by means of the differences of two such proportions, when their ordering is not in agreement with the requirements of the model. A minimum violation, by default, of.02 (Molenaar et al., 1994, p. 45) or.03 (Molenaar & Stsma, 2000, p. 66) is admitted, although this boundary may be altered by the researcher. 2a Recommendations made by these authors regarding the sample size and group size; and 3a other summary statistics and test of significance. Given a set of items that satisfies the assumptions of the MH model, this paper sets out the procedure to calculate a GD measure from the DM of the set of items (or non intersection of pairs of items) based on the estimation of the size of deviation (violation) from monotonicity. (N.B.: All further references to DM refer only to the assumption of non intersection of IRFs).

4 84 MPR-Online 2003, Vol. 8, No. 1 The GD measure presents two differences with regard to the violations defined in Molenaar et al. (1994, p. 45): 1b When there is a deviation from DM, the discrepancies between the observed proportions and estimated theoretical proportions (disparities) are calculated. These disparities are the values which the observed proportions should assume to satisfy the monotonicity when, in fact, they violate it. 2b It takes into account the size and number of deviations in a set of items. Therefore, these deviations give different and additional information in respect of the violations defined in (1a). In order to define the GD measure from DM, the following concepts are linked together: 1. The concept of DM established by Mokken (1971, 1997) for pairs of responses for ( ) triples of items ( k,, ) : Let P = p ( 1,1 ), p ( 0,0) 11 jk 00 jk ( ) P = of order n n be the symmetric matrices of joint proportions of scoring both correct and failed. k=1,...,n j<k; j<k denotes the item ordering for j is more difficult than k. Then, for all item k I and all pairs ( jk, ) I, j< k: p ( 1,1 ) pik ( 1,1) and p ( 0,0 ) pik ( 0,0) (Mokken, 1997, p. 357). The difficulty of an item, δ j, being the proportion of correct responses to item j. If a set of items is MH, Mokken proposes the study of DM analyzing visually the P 11 and matrices. If the items are ordered in decreasing difficulty, a MH set of items is DM when columns and rows in P 11 matrix are monotonely non-decreasing, and columns and rows in matrix are monotonely non-increasing. Local deviations from these orders are considered violations of DM. The concept of Mokken s DM means that the items must satisfy a monotone relation in rows and columns of P 11 and matrices. Whether or not this monotone relation is satisfied can be ascertained by the Isotonic Regression, then 2. Isotonic regression quantifies the degree to which the items of each row or column of P 11 and satisfy a monotone relation. Isotonic regression method, as in nonmetric scaling, allows the quantification of discrepancies from monotonicity in each row or column, by means of the differences between the observed proportions and the estimated theoretical proportions (disparities) obtained at the slopes of GCM.

5 Rivas Moya: Goodness of Fit Measure for Mokken Model 85 To this end, the concepts of Cumulative Sum Diagram (CSD) and Greatest Convex Minorant (GCM) of isotonic regression (Barlow, Bartholomew, Bremner & Brunk, 1972) adapted to calculate disparities associated with dissimilarities in non-metric Multidimensional Scaling in Rivas Moya (2000a) are used to calculate the disparities associated with the observed proportions. As a result, associated with the observed proportions which do not satisfy the monotonicity, the pˆ disparities are estimated in such a way that, when considering the disparities, the items satisfy a monotone relation in rows and columns. If this is satisfied in P 11 and, then the set of items is DM (Rivas Moya, 2000b). The basic idea is the following: Given the P 11 matrix in Table 1, the difficulty of items in columns j:1,2,..,n induces an order in each row of P 11. Numbers in brackets, in Table 1, denote the non-decreasing order induced by the difficulty. This order can be considered a dissimilarity measure, d( i, j ), between pairs of items i, j. Index of row i can be omitted because in each row it is a constant. Then, in any row i, indices of columns i+ 1 i n or d d... d denote the dissimilarities between pairs of items. i+ 1 i+ 2 n p Table1 Observed Proportions and Dissimilarities in the P 11 Matrix i / j i 1 i 2 i 3 i 4... i n i 1 - p 12 (2) p 13 (3) p 14 (4)... p 1n (n) i 2 - p 23 (3) p 24 (4)... p 2n (n) i n-1 - p n-1n (n) δ j δ 1 δ 2 δ 3 δ 4... δ n If p p j,k: 12,..n is not satisfied in each row i of P ik 11, the DM is violated because the proportions p do not satisfy the non-decreasing monotone relation in the same way as the dissimilarities between items d( i,j ). Then p must be substituted by the corresponding disparities ˆp, and the relation is now non-decreasing monotone. The disparities will be obtained as the isotonic regression of proportion function. To calcu- ( ) late them, it is necessary to know the coordinates of points P W,P = of CSD and

6 86 MPR-Online 2003, Vol. 8, No. 1 ˆ ( ) P ˆ = W,P GCM. of GCM, and the slopes of segments joining the points of CSD and The discrepancies from monotonicity in the items of each row can now be estimated, and deviation from monotonicity in all the rows of P 11 is defined as n p ( 1,1) p ˆ ( 1,1 ). Similarly, deviation from monotonicity in all the columns of P 11 i < j is defined. Then the GD measure for the P 11 -matrix is given as: being the dissimilarity matrix. n( n- 1) because ( ) and columns. n n p ( 1,1) pˆ ( 1,1) + p ( 1,1) pˆ ( 1,1) i < j j < i D( P,, n) = (1) 11 n n 1 ( ) n n 1 2 elements in each P -matrix are compared twice: in rows The GD measure from monotonicity non-increasing D( P,, n) in P matrix is similarly defined. Justification and details can be found in Rivas Moya (2000b). These measures are bounded between 0 and 1. There is no deviation from DM if these measures are 0. The maximum deviation from DM is given when these GD measures are 1. This paper gives several examples from responses of 294 subjects to 16 dichotomous items to show 1. The procedure which calculates the GD measure from and P 11 matrices. In addition, discrepancies from DM in each row / column of the P-matrix can be seen when plotting the coordinates of CSD and GCM associated with observed and estimated theoretical proportions, respectively. 2. If there are no individual deviations from DM then p = pˆ and GD is zero. 3. An increase in the number and/or size of individual deviations from DM leads to an increase in GD measure when sets of different numbers of items are considered. Thus: 3a From two different sets of 8 items, GD is larger when the size and the number of deviations is larger.

7 Rivas Moya: Goodness of Fit Measure for Mokken Model 87 3b A set of 7 items shows a lesser number of deviations, but of larger size than those of the previous examples. Then, GD is greater or equal to the measures in previous examples. These examples show empirically that in the GD measure the discrepancies from DM are obtained by comparing the observed proportions ( p ) with the disparities ( ˆp ) calculated by isotone regression. Procedure to Calculate the GD Measure from P-Matrices Given P 11 and matrices, and items in descending order of difficulty, the steps to obtain the GD from DM are set out in columns 1 10 of a table. The procedure described should be made in each row or column of each P- matrix. Column 1. Enumerates the rows or columns of the P-matrix. Column 2. Enumerates pairs of items (, ) of a row or column Column 3. Column 4. Dissimilarities d induced by the difficulty of items, that means the order that induces the difficulty. Observed proportions p given in Table 1. The matrices being obtained by MSP5 (Molenaar, et al. 2000). Values p which do not satisfy nondecreasing monotonicity are shown in bold and with (*). Columns 5, 6, 7.Weights w, accumulated weights W proportions P j k = 1 = p, respectively. Here, weights are fixed and they assume a value of 1.( W, P ) ik j = w ik and accumulated k = 1 these points the CSD is obtained. being the coordinates of CSD. By joining Column 8. Disparities pˆ associated with the proportions. 1. If there is no deviation from monotonicity, p = pˆ. 2. If there is deviation from monotonicity, disparity is calculated as:

8 88 MPR-Online 2003, Vol. 8, No. 1 P P p ˆ = with i< j, j : i + 1,...,n (2a) W W If there are two consecutive proportions p, p + 1 which do not satisfy monotonicity, the disparity associated with both proportions is given by: P P pˆ = W W (2b) Then, the slope of segment joining (, 1) and (, ) is P ˆ and the slope of segment joining (, ) and ( +, 1) is P ˆ + 1 = pˆ. Similarly, the method of calculating P ˆ can be extended to more than two successive proportions which do not satisfy monotonicity 4. If a deviation from monotonicity is found in the last item n of the row (column), disparity calculation is made as follows: A fictitious index ( n + 1) is added and assigned the value p in + 1 = p in 1. The deviation from monotonicity is calculated in respect of p in 1 That is, if P in 1 > P in and P is the last value of row i that satisfies the monotony, then P + 1 = P + p and in in P ˆ in 1 P p in 1 in = + Win + 1 Win 1 (2c) Column 9. Ordinates ˆP of GCM, are obtained as ˆP = P, if there is no deviation from monotonicity. According to the previous situations (2a), (2b), (2c), if there is deviation from monotonicity, then ordinates ˆP of GCM are obtained in (3a), (3b), (3c), respectively. For example, to obtain (3a), the equation of the segment, with slope ˆp, joining the points ( W 1, P 1) and ( W, ˆ P ) ( P ˆ P ) = p ˆ ( W W ) 1 1, then ( ), is P ˆ = P 1 + p ˆ W W 1 (3a) Similarly, ordinates (3b) and (3c) are obtained:

9 Rivas Moya: Goodness of Fit Measure for Mokken Model 89 ( ) ( ) Pˆ = P 1 + pˆ W W 1 Pˆ + 1 = P 1 + pˆ W + 1 W 1 (3b) Pˆ in = Pin 1 + pˆ in ( Win Win 1) (3c) Abscissas ˆ W coincide with those of CSD W. The GCM which satisfies non-decreasing monotonicity is obtained by joining the points ( W, Pˆ ). Column 10. The absolute values of the differences between observed proportions p and disparities ˆp for each pair of items, ˆp p. Columns 8,9, 10, are left blank if there is no deviation from DM in the entire row or column pˆ p = p p = 0. The discrepancies in rows and columns of P 11 are summed up to obtain GD measure in P 11. Similarly, this measure is calculated in. The graphs CSD and GCM show these discrepancies in each row (column). The greater the difference between these diagrams, the greater the discrepancy from monotonicity. The following has been used to apply this procedure to the data: 1. MSP5 for Windows program of Molenaar et al. (2000) to obtain the difficulty of items and observed proportions p in P 11, matrices. 2. Excel program to calculate, pˆ,, ˆ P P, pˆ p and the GD measure. 3. Graphs CSD and GCM are obtained with a graph program. Data In order to calculate GD, data collected by Ortiz (1997) were used. A test of 12 items, comprising several levels of Numerical Inductive Reasoning each item being a series of distinct difficulty was given to 296 subjects of General Basic Education (age 8-12). The task which each subject carried out was to complete the numerical sequence of the series by either addition, subtraction, multiplication or division. Subjects responses were codified in binary form 0, 1 indicating that the series had been completed either incorrectly or correctly, respectively.

10 90 MPR-Online 2003, Vol. 8, No. 1 Different subgroups of items have been selected to illustrate the GD: In Study 1: Four items whose numerical sequence is obtained by addition or subtraction. In Study 2: The four easiest items whose numerical sequence is obtained by addition. In Study 3: (1) The eight easiest items, (2) the eight most difficult items, (3) four easy and three difficult items. Study 1: Application of the Procedure to Calculate the GD Measure The P11and matrices from the responses of 294 subjects to 4 dichotomous items are given in Table 2. The items are given in descending order of difficulty δ. Table 2 P11and Matrices P 11 Items Items δ δ Columns in P 11 are non-decreasing. Therefore, there is no deviation from DM when analyzing the columns. In other rows or columns of P 11 and matrices there are deviations. Proportions p of P matrices are included in column 4 of tables 3 and 4. In each row of Table 3, p (1,1) should satisfy the same order as d. In row 1, P 11 shows one deviation from non-decreasing monotonicity. Then, the fixed weights w = 1 are given. Accumulated weights W and accumulated proportions p are obtained to calculate the disparity p 13(1,1) associated with p 13(1,1) (column 8).

11 Rivas Moya: Goodness of Fit Measure for Mokken Model 91 ˆ (1,1) = = =.294 and the size of Then, p ( P P ) ( W W ) ( ) ( ) this deviation is pˆ p = =.016, in row 1 of the P 11 - matrix. This value close to 0 indicates that there is a small deviation from monotonicity in the items of row 1. In rows 2 and 3 there is no deviation from monotonicity, column 4 is equal to column 8. In consequence, column 7 is equal to column 9, and p ˆ p = 0. So the values p ˆ, pˆ and p ˆ p are left blank. Table 3 Deviation from DM in the P11- matrix P 11 (i,j) d (1,1) p w W P pˆ Pˆ pˆ p row 1 (1,2) 1.300* (1,3) 2.278* (1,4) 3.310* row 2 (2,3) 1.293* (2,4) 2.321* row 3 (3,4) 1.345* Total.016 In Column 7, the ordinates of CSD are: P = , P = + =, P = = In Column 9, the ordinates of GCM are: P ˆ 12 = P12 = Applying (3a) P ˆ13 = =., P ˆ = P ( ) = Figure 1 shows the CSD (columns 6,7) and the GCM (columns, 6, 9) associated with p (1,1) in row 1 where it is found that the discrepancy from monotonicity is In this figure, the graphs of CSD and GCM almost overlap. This means that there is hardly any difference between the observed proportions and disparities in row1.

12 92 MPR-Online 2003, Vol. 8, No. 1 Ordinates CSD/GCM 1,0 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 CSD_Row 1 GCM_Row1 0, Weights Figure 1. CSD and GCM (row 1 in table 3) There are deviations from DM in the rows and columns of -matrix (Table 4). In Row 2 there is one deviation from DM in the last value of the row item pair (2,4). Then, a fictitious index of 5 is included, and is given the value p 25 = p 23 =.293 (Value.293 is underlined in Table 4) and the disparity associated with this value is obtained, P P by calculating p ˆ 24(0, 0) = = = This gives a deviation of W W = In Column 4 there is one deviation from DM in item pair (3,4), then a new fictitious index of 4 is included, and is given the value p 44 = p24 =. 439 (Value.439 is underlined in Table 4). The disparity associated with this value is calculated as P P p ˆ 34(0, 0) = = =.451. The size of the deviation is W W =.012. In 00 P there is a total deviation ( p p ) ˆ = =.085. P00 Row 1 and Column 3 of Table 4 show no deviation from monotonicity. So p ˆ = are not included in this Table. In consequence, vertical Column 4 is equal to vertical Column 8, and pˆ p = 0 (similar to Table 3). GD measure is given as DP ( 11,, 4) = = in P11-matrix and as DP ( 00,, 4) = = in 00 P - matrix. GD in both matrices is close to zero, so there is no great deviation from DM. p

13 Rivas Moya: Goodness of Fit Measure for Mokken Model 93 Table 4 Deviation from DM in the Matrix (i,j) d p (0,0) w W P pˆ pˆ p Row 1 (1,2) 1.557* (1,3) 2.537* (1,4) 3.439* Row 2 (2,3) 1.293* (2,4) 2.439* * Row3 (3,4) 1.463* Column 2 (1,2) 1.557* Column 3 (1,3) 1.537* (2,3) 2.293* Column 4 (1,4) 1.439* (2,4) 2.439* (3,4) 3.463* * Total.085 Pˆ Study 2: GD is Zero if there are no Individual Deviations from DM P11and matrices from the responses of 294 subjects to 4 dichotomous items are given in Table 5 (These 4 items are different to the 4 items in Study 1). It can be seen that there is no deviation from DM in any row or column of these matrices.

14 94 MPR-Online 2003, Vol. 8, No. 1 Table 5 P11and Matrices P 11 Items Items δ δ If the procedure of Study 1 is applied, P = and p = pˆ for all p. In each row (or column) the CSD and GCM coincide and the GD measure is zero. Pˆ Study 3: Effect of Size of Individual Deviations on GD P11and matrices from the responses of 294 subjects to two sets of 8 and one set of 7 dichotomous items are given in Tables 6a, 6b, 7a, 7b, 8a, 8b. After applying the above procedure, described in Tables 3 and 4 of Study 1, only the proportions p and disparities pˆ associated with proportions not satisfying monotonicity are included. This is done with the object of summarizing the information as follows: each cell of P 11 and matrices shows two values which appear in the following order (from top to bottom). When there is deviation from DM in rows, in Tables 6a, 7a and 8a: 1. p appears in bold, 2. the estimated pˆ is shown below the figure in bold. When there is deviation from DM in columns, in Tables 6b, 7b and 8b: 1. p appears in bold, 2. the estimated p is shown below the figure in bold. For example, in the P 11 matrix of Table 6a, there is one deviation in Row 2 (pair (2,7) p. 70 and p ˆ 27 =. 705 ) and in Table 6b, there is one deviation in Column 6 of = P matrix (pair (3,6) p. 70 and p ˆ 36 =. 72 ). 36 =

15 Rivas Moya: Goodness of Fit Measure for Mokken Model 95 The sum of all deviations and GD for each P-matrix is shown at the foot of each Table. pˆ in each matrix of Tables 6a and 6b is obtained, with p ˆ p = = DP (,,8) = = and In P 11, p ˆ = = D ( P 11,,8) = = P11 p Table 6a P Matrices and Deviations from DM in Rows (8 Items) P 11 Item Item δ δ , Rows p ˆ = p ˆ = p Table 6b P Matrices and Deviations from DM in Columns (8 Items) P11, Rows p P 11 Item Item δ δ , Col p ˆ = p ˆ = p P11, Col p

16 96 MPR-Online 2003, Vol. 8, No. 1 In Tables 7a and 7b, the deviations in and P 11 are p ˆ p = = and 00 P11 DP (,,8) = = p ˆ p = = 0.21and ( P 11,,8) = = D. Table 7a P Matrices and Deviations from DM in Rows (8 Items) P 11 Item Item δ δ , Rows p ˆ = p ˆ = p P11, Rows p Table 7b P Matrices and Deviations from DM in Columns (8 Items) P 11 Item Item δ δ , Col p ˆ = p ˆ = p P11, Col p

17 Rivas Moya: Goodness of Fit Measure for Mokken Model 97 Tables 6a and 6b present fewer deviations from DM than Tables 7a and 7b, and the GD obtained from Tables 6a 6b is lower than GD from Tables 7a 7b. P 11. In Tables 8a and 8b, there is no deviation in the rows of nor in the columns of P, ˆ = = In 00 In 11 p and D (,,7) = = p P, ˆ = = P11 p and D ( P 11,,7) = = p Table 8a P Matrices and Deviations from DM in Rows (7 Items) P 11 Item Item δ δ , Rows p ˆ = 0 p ˆ = p P11, Rows p Table 8b P Matrices and Deviations from DM in Columns (7 Items) P 11 Item Item δ δ , Col p ˆ p = 0.16 p pˆ = 0 P11, Col

18 98 MPR-Online 2003, Vol. 8, No. 1 In Tables 8a and 8b there are only 5 deviations from DM, but these deviations are greater than those in Tables 6a, 6b and 7a, 7b. These deviations are reflected overall in the GD measure, which in Tables 8a, 8b is greater than or equal to any of the others. From these examples, when different set of items are considered, it can be seen that a few large deviations have more effect than many small deviations in this goodness of fit measure. In order to analyze visually the deviation from DM, for example in row 1 of P 11 (Table 8a), values P and Pˆ associated with p and pˆ are given in Table 9. There are deviations in consecutive item pairs ( 1,4) and (,5) value pˆ associated with p 14 and 15 p is ( ) ( 5 2) = Applying (2b), the and applying (3b), P ˆ14 = ( 3 2) = and P ˆ15 = ( 4 2) = the items there is no deviation from monotonocity, P ˆ = P.. As in the rest of Table 9 P and P ˆ of Row 1 in P 11 (Table 8a) Item pairs Weights p P pˆ Pˆ ( 1,2 ) ( 1,3) ( 1,4 ) ( 1,5) ( 1,6 ) ( 1,7 ) CSD and GCM of row 1 in P 11 are plotted in Figure 2. 0,8 Ordinates CSD/GCM 0,7 0,6 0,5 0,4 0,3 0,2 CSD_Row1 GCM_Row1 0, Weights Figure 2. CSD and GCM (row 1 in P 11, table 8a)

19 Rivas Moya: Goodness of Fit Measure for Mokken Model 99 It can be seen from Figure 2 that deviations from DM in this row are important. Comparing Figures 1 and 2, it can be seen that the deviation from DM in row 1 of Table 3 (0.016) is less important that that in row 1 of Table 8a, this deviation being Row1 ( p p ) ˆ = = In the same way, deviations in other rows and/or columns can be calculated and visualized. Discussion Mokken s concept of DM and isotonic regression are linked together to define a goodness of fit measure (GD) from DM in P 11 and matrices. In this way, the study of non decreasing monotonicity (in rows and columns of P 11) and non increasing monotonicity (in rows and columns of ) is made by isotonic regression. This method is applied in the same way as it is applied to define the goodness of fit measure stress in non-metric multidimensional scaling. The GD measure is shown, explaining the procedure to calculate it in general terms. From the examples presented in this work, the GD measure has been checked with real data. It shows: 1. The empirical procedure for calculating the measure (Study 1), 2. the minimum value is 0 if there is no deviation from DM (Study 2), and 3. a comparison of results of several types of data is given. On one hand, given two sets with the same number of items (8) but of varying difficulty, the example shows that the greater number of deviations, the more the GD increases. On the other hand, given two sets with a differing number of items (8 and 7) but also of varying difficulty, the example shows that a few large-size deviations result in a greater or equal GD measure than more deviations of smaller size. The advantages of this measure over others that investigate deviations from DM are: 1. It takes into account the size and number of deviations. The size of the deviations is obtained by calculating the differences between the observed and estimated theoretical proportions (disparities), the latter being obtained by Isotone Regression. Thus, in P 11, the disparities which make the GCM satisfy the non-decreasing monotonicity are considered in order to ascertain whether the discrepancies from

20 100 MPR-Online 2003, Vol. 8, No. 1 DM are great or small. In addition, deviations from DM in each row or column can be analyzed visually by plotting the CSD (associated with observed proportions p ) and the GCM (associated with estimated proportions pˆ ). The greater the difference between these Diagrams, the greater the discrepancy from DM. 2. The procedure given to calculate the measure allows us to know the deviations between pairs of items, all the items of one row or column of or P 11, or global deviation in each P-matrix. This last measure gives the global deviation from nondecreasing monotonicity, on one hand, and the global deviation from non-increasing monotonicity, on the other hand. Further studies are necessary to 1. Prove that in some cases, given the same number of items, a few large-size deviations can result in a greater deviation measure than more deviations of smaller size 2. Compare GD empirically with other indices which evaluate DM. 3. Determine the bounds of GD using simulated data in order to see if there is low, medium or high deviation from DM. 4. Extend this measure to polytomous items. References Barlow, R. E., Bartholomew, D. J., Bremner, J. M. & Brunk, H. D. (1972). Statistical inference under order restrictions. London: John Wiley and Sons. Mokken, R. J. (1971). A theory and procedure of scale analysis with applications in political research. Berlin: Walter de Gruyter, Mouton. Mokken, R. J. (1997). Nonparametric models for dichotomous responses. In W. J. Van der Linden & R K. Hambleton (Eds.) Handbook of modern item response theory (pp ). New York: Springer. Molenaar, I. W. & Stsma, K. (2000). User s manual MSP5 for windows. Groningen: iec ProGAMMA. Molenaar, I. W. (1991). Fit investigation in the multicategory Mokken scale. (Unpublished manuscript). Molenaar, I. W., Debets, P., Stsma, K. & Hemker, B. T. (1994). MSP 3.0. A program for Mokken scale analysis for polytomous items. Groningen: iec ProGAMMA.

21 Rivas Moya: Goodness of Fit Measure for Mokken Model 101 Molenaar, I. W., Stsma, K., van Schuur, W. H. & Mokken, R. J. (2000). MSP5 for Windows. A program for Mokken scale analysis for polytomous items. Groningen: iec ProGAMMA. Ortiz, A. (1997). Razonamiento inductivo numérico: Un estudio en Educación Primaria. Unpublished doctoral dissertation. Granada University, Spain. Rivas Moya, T. (2000a). Calculating isotonic regression of the distance function in nonmetric multidimensional scaling model. Methods of Psychological Research, 5(3), 1-8. Rivas Moya, T. (2000b). Goodness of fit measure based on sample isotone regression of mokken double monotonicity model. In H. A. L Kiers, J.-P. Rasson, P. J. F. Groenen & M. Schader (Eds.), Data analysis, classification and related methods (pp ) Berlin: Springer Verlag. Rosenbaum, P. R. (1984). Testing the conditional independence and monotonicity assumptions of item response theory. Psychometrika, 49, Rosenbaum, P. R. (1987). Comparing item characteristic curves. Psychometrika, 52, Stsma, K. & Junker, B. W. (1996). A survey of theory and methods of invariant item ordering. British Journal of Mathematical and Statistical Psychology, 49, Stsma, K. & Meer, R. R. (1992). A method for investigating the intersection of item response functions in Mokken s nonparametric IRT model, Applied Psychological Measurement. 16, Stsma, K. & Molenaar, I. W. (2002). Introduction to nonparametric item response theory. London: Sage Publications. Stsma, K. (1998). Methodology review: Nonparametric IRT approaches to the analysis of dichotomous item scores. Applied Psychological Measurement, 22, Stsma, K. (2001). Developments in measurement of persons and items by means of item response models. Behaviormetrika, 28,

On the Use of Nonparametric ICC Estimation Techniques For Checking Parametric Model Fit

On the Use of Nonparametric ICC Estimation Techniques For Checking Parametric Model Fit On the Use of Nonparametric ICC Estimation Techniques For Checking Parametric Model Fit March 27, 2004 Young-Sun Lee Teachers College, Columbia University James A.Wollack University of Wisconsin Madison

More information

Number of cases (objects) Number of variables Number of dimensions. n-vector with categorical observations

Number of cases (objects) Number of variables Number of dimensions. n-vector with categorical observations PRINCALS Notation The PRINCALS algorithm was first described in Van Rickevorsel and De Leeuw (1979) and De Leeuw and Van Rickevorsel (1980); also see Gifi (1981, 1985). Characteristic features of PRINCALS

More information

Journal of Statistical Software

Journal of Statistical Software JSS Journal of Statistical Software May 2007, Volume 20, Issue 11. http://www.jstatsoft.org/ Mokken Scale Analysis in R L. Andries van der Ark Tilburg University Abstract Mokken scale analysis (MSA) is

More information

over the parameters θ. In both cases, consequently, we select the minimizing

over the parameters θ. In both cases, consequently, we select the minimizing MONOTONE REGRESSION JAN DE LEEUW Abstract. This is an entry for The Encyclopedia of Statistics in Behavioral Science, to be published by Wiley in 200. In linear regression we fit a linear function y =

More information

University of Groningen. Mokken scale analysis van Schuur, Wijbrandt H. Published in: Political Analysis. DOI: /pan/mpg002

University of Groningen. Mokken scale analysis van Schuur, Wijbrandt H. Published in: Political Analysis. DOI: /pan/mpg002 University of Groningen Mokken scale analysis van Schuur, Wijbrandt H. Published in: Political Analysis DOI: 10.1093/pan/mpg002 IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's

More information

Contents. 3 Evaluating Manifest Monotonicity Using Bayes Factors Introduction... 44

Contents. 3 Evaluating Manifest Monotonicity Using Bayes Factors Introduction... 44 Contents 1 Introduction 4 1.1 Measuring Latent Attributes................. 4 1.2 Assumptions in Item Response Theory............ 6 1.2.1 Local Independence.................. 6 1.2.2 Unidimensionality...................

More information

What is an Ordinal Latent Trait Model?

What is an Ordinal Latent Trait Model? What is an Ordinal Latent Trait Model? Gerhard Tutz Ludwig-Maximilians-Universität München Akademiestraße 1, 80799 München February 19, 2019 arxiv:1902.06303v1 [stat.me] 17 Feb 2019 Abstract Although various

More information

Applied Psychological Measurement 2001; 25; 283

Applied Psychological Measurement 2001; 25; 283 Applied Psychological Measurement http://apm.sagepub.com The Use of Restricted Latent Class Models for Defining and Testing Nonparametric and Parametric Item Response Theory Models Jeroen K. Vermunt Applied

More information

Number of analysis cases (objects) n n w Weighted number of analysis cases: matrix, with wi

Number of analysis cases (objects) n n w Weighted number of analysis cases: matrix, with wi CATREG Notation CATREG (Categorical regression with optimal scaling using alternating least squares) quantifies categorical variables using optimal scaling, resulting in an optimal linear regression equation

More information

Pairwise Parameter Estimation in Rasch Models

Pairwise Parameter Estimation in Rasch Models Pairwise Parameter Estimation in Rasch Models Aeilko H. Zwinderman University of Leiden Rasch model item parameters can be estimated consistently with a pseudo-likelihood method based on comparing responses

More information

monte carlo study support the proposed

monte carlo study support the proposed A Method for Investigating the Intersection of Item Response Functions in Mokken s Nonparametric IRT Model Klaas Sijtsma and Rob R. Meijer Vrije Universiteit For a set of k items having nonintersecting

More information

Number of analysis cases (objects) n n w Weighted number of analysis cases: matrix, with wi

Number of analysis cases (objects) n n w Weighted number of analysis cases: matrix, with wi CATREG Notation CATREG (Categorical regression with optimal scaling using alternating least squares) quantifies categorical variables using optimal scaling, resulting in an optimal linear regression equation

More information

PIRLS 2016 Achievement Scaling Methodology 1

PIRLS 2016 Achievement Scaling Methodology 1 CHAPTER 11 PIRLS 2016 Achievement Scaling Methodology 1 The PIRLS approach to scaling the achievement data, based on item response theory (IRT) scaling with marginal estimation, was developed originally

More information

Package mokken. February 15, 2013

Package mokken. February 15, 2013 Package mokken February 15, 2013 Version 2.7.5 Date 2013-01-27 Title Mokken Scale Analysis in R Author L. Andries van der Ark Maintainer L. Andries van der Ark Depends

More information

Comparison of parametric and nonparametric item response techniques in determining differential item functioning in polytomous scale

Comparison of parametric and nonparametric item response techniques in determining differential item functioning in polytomous scale American Journal of Theoretical and Applied Statistics 2014; 3(2): 31-38 Published online March 20, 2014 (http://www.sciencepublishinggroup.com/j/ajtas) doi: 10.11648/j.ajtas.20140302.11 Comparison of

More information

Overview. Multidimensional Item Response Theory. Lecture #12 ICPSR Item Response Theory Workshop. Basics of MIRT Assumptions Models Applications

Overview. Multidimensional Item Response Theory. Lecture #12 ICPSR Item Response Theory Workshop. Basics of MIRT Assumptions Models Applications Multidimensional Item Response Theory Lecture #12 ICPSR Item Response Theory Workshop Lecture #12: 1of 33 Overview Basics of MIRT Assumptions Models Applications Guidance about estimating MIRT Lecture

More information

Monte Carlo Simulations for Rasch Model Tests

Monte Carlo Simulations for Rasch Model Tests Monte Carlo Simulations for Rasch Model Tests Patrick Mair Vienna University of Economics Thomas Ledl University of Vienna Abstract: Sources of deviation from model fit in Rasch models can be lack of unidimensionality,

More information

Contents. Acknowledgments. xix

Contents. Acknowledgments. xix Table of Preface Acknowledgments page xv xix 1 Introduction 1 The Role of the Computer in Data Analysis 1 Statistics: Descriptive and Inferential 2 Variables and Constants 3 The Measurement of Variables

More information

On the Construction of Adjacent Categories Latent Trait Models from Binary Variables, Motivating Processes and the Interpretation of Parameters

On the Construction of Adjacent Categories Latent Trait Models from Binary Variables, Motivating Processes and the Interpretation of Parameters Gerhard Tutz On the Construction of Adjacent Categories Latent Trait Models from Binary Variables, Motivating Processes and the Interpretation of Parameters Technical Report Number 218, 2018 Department

More information

- 1 - Items related to expected use of technology appear in bold italics.

- 1 - Items related to expected use of technology appear in bold italics. - 1 - Items related to expected use of technology appear in bold italics. Operating with Geometric and Cartesian Vectors Determining Intersections of Lines and Planes in Three- Space Similar content as

More information

Testing hypotheses about the person-response function in person-fit analysis Emons, Wilco; Sijtsma, K.; Meijer, R.R.

Testing hypotheses about the person-response function in person-fit analysis Emons, Wilco; Sijtsma, K.; Meijer, R.R. Tilburg University Testing hypotheses about the person-response function in person-fit analysis Emons, Wilco; Sijtsma, K.; Meijer, R.R. Published in: Multivariate Behavioral Research Document version:

More information

A Simulation Study to Compare CAT Strategies for Cognitive Diagnosis

A Simulation Study to Compare CAT Strategies for Cognitive Diagnosis A Simulation Study to Compare CAT Strategies for Cognitive Diagnosis Xueli Xu Department of Statistics,University of Illinois Hua-Hua Chang Department of Educational Psychology,University of Texas Jeff

More information

Local response dependence and the Rasch factor model

Local response dependence and the Rasch factor model Local response dependence and the Rasch factor model Dept. of Biostatistics, Univ. of Copenhagen Rasch6 Cape Town Uni-dimensional latent variable model X 1 TREATMENT δ T X 2 AGE δ A Θ X 3 X 4 Latent variable

More information

Experimental Design and Data Analysis for Biologists

Experimental Design and Data Analysis for Biologists Experimental Design and Data Analysis for Biologists Gerry P. Quinn Monash University Michael J. Keough University of Melbourne CAMBRIDGE UNIVERSITY PRESS Contents Preface page xv I I Introduction 1 1.1

More information

The Rasch Model, Additive Conjoint Measurement, and New Models of Probabilistic Measurement Theory

The Rasch Model, Additive Conjoint Measurement, and New Models of Probabilistic Measurement Theory JOURNAL OF APPLIED MEASUREMENT, 2(4), 389 423 Copyright 2001 The Rasch Model, Additive Conjoint Measurement, and New Models of Probabilistic Measurement Theory George Karabatsos LSU Health Sciences Center

More information

CHAPTER 3. THE IMPERFECT CUMULATIVE SCALE

CHAPTER 3. THE IMPERFECT CUMULATIVE SCALE CHAPTER 3. THE IMPERFECT CUMULATIVE SCALE 3.1 Model Violations If a set of items does not form a perfect Guttman scale but contains a few wrong responses, we do not necessarily need to discard it. A wrong

More information

UCLA Department of Statistics Papers

UCLA Department of Statistics Papers UCLA Department of Statistics Papers Title Can Interval-level Scores be Obtained from Binary Responses? Permalink https://escholarship.org/uc/item/6vg0z0m0 Author Peter M. Bentler Publication Date 2011-10-25

More information

36-720: The Rasch Model

36-720: The Rasch Model 36-720: The Rasch Model Brian Junker October 15, 2007 Multivariate Binary Response Data Rasch Model Rasch Marginal Likelihood as a GLMM Rasch Marginal Likelihood as a Log-Linear Model Example For more

More information

Computerized Adaptive Testing With Equated Number-Correct Scoring

Computerized Adaptive Testing With Equated Number-Correct Scoring Computerized Adaptive Testing With Equated Number-Correct Scoring Wim J. van der Linden University of Twente A constrained computerized adaptive testing (CAT) algorithm is presented that can be used to

More information

Item Response Theory (IRT) Analysis of Item Sets

Item Response Theory (IRT) Analysis of Item Sets University of Connecticut DigitalCommons@UConn NERA Conference Proceedings 2011 Northeastern Educational Research Association (NERA) Annual Conference Fall 10-21-2011 Item Response Theory (IRT) Analysis

More information

Freeman (2005) - Graphic Techniques for Exploring Social Network Data

Freeman (2005) - Graphic Techniques for Exploring Social Network Data Freeman (2005) - Graphic Techniques for Exploring Social Network Data The analysis of social network data has two main goals: 1. Identify cohesive groups 2. Identify social positions Moreno (1932) was

More information

Bayesian Nonparametric Rasch Modeling: Methods and Software

Bayesian Nonparametric Rasch Modeling: Methods and Software Bayesian Nonparametric Rasch Modeling: Methods and Software George Karabatsos University of Illinois-Chicago Keynote talk Friday May 2, 2014 (9:15-10am) Ohio River Valley Objective Measurement Seminar

More information

Lesson 7: Item response theory models (part 2)

Lesson 7: Item response theory models (part 2) Lesson 7: Item response theory models (part 2) Patrícia Martinková Department of Statistical Modelling Institute of Computer Science, Czech Academy of Sciences Institute for Research and Development of

More information

Application Note. The Optimization of Injection Molding Processes Using Design of Experiments

Application Note. The Optimization of Injection Molding Processes Using Design of Experiments The Optimization of Injection Molding Processes Using Design of Experiments Problem Manufacturers have three primary goals: 1) produce goods that meet customer specifications; 2) improve process efficiency

More information

Quality of Life Analysis. Goodness of Fit Test and Latent Distribution Estimation in the Mixed Rasch Model

Quality of Life Analysis. Goodness of Fit Test and Latent Distribution Estimation in the Mixed Rasch Model Communications in Statistics Theory and Methods, 35: 921 935, 26 Copyright Taylor & Francis Group, LLC ISSN: 361-926 print/1532-415x online DOI: 1.18/36192551445 Quality of Life Analysis Goodness of Fit

More information

PREDICTING THE DISTRIBUTION OF A GOODNESS-OF-FIT STATISTIC APPROPRIATE FOR USE WITH PERFORMANCE-BASED ASSESSMENTS. Mary A. Hansen

PREDICTING THE DISTRIBUTION OF A GOODNESS-OF-FIT STATISTIC APPROPRIATE FOR USE WITH PERFORMANCE-BASED ASSESSMENTS. Mary A. Hansen PREDICTING THE DISTRIBUTION OF A GOODNESS-OF-FIT STATISTIC APPROPRIATE FOR USE WITH PERFORMANCE-BASED ASSESSMENTS by Mary A. Hansen B.S., Mathematics and Computer Science, California University of PA,

More information

Equating Tests Under The Nominal Response Model Frank B. Baker

Equating Tests Under The Nominal Response Model Frank B. Baker Equating Tests Under The Nominal Response Model Frank B. Baker University of Wisconsin Under item response theory, test equating involves finding the coefficients of a linear transformation of the metric

More information

Principal Components Analysis. Sargur Srihari University at Buffalo

Principal Components Analysis. Sargur Srihari University at Buffalo Principal Components Analysis Sargur Srihari University at Buffalo 1 Topics Projection Pursuit Methods Principal Components Examples of using PCA Graphical use of PCA Multidimensional Scaling Srihari 2

More information

New Developments for Extended Rasch Modeling in R

New Developments for Extended Rasch Modeling in R New Developments for Extended Rasch Modeling in R Patrick Mair, Reinhold Hatzinger Institute for Statistics and Mathematics WU Vienna University of Economics and Business Content Rasch models: Theory,

More information

Evaluating Goodness of Fit in

Evaluating Goodness of Fit in Evaluating Goodness of Fit in Nonmetric Multidimensional Scaling by ALSCAL Robert MacCallum The Ohio State University Two types of information are provided to aid users of ALSCAL in evaluating goodness

More information

Discrete Applied Mathematics

Discrete Applied Mathematics Discrete Applied Mathematics 194 (015) 37 59 Contents lists available at ScienceDirect Discrete Applied Mathematics journal homepage: wwwelseviercom/locate/dam Loopy, Hankel, and combinatorially skew-hankel

More information

Principal Component Analysis, A Powerful Scoring Technique

Principal Component Analysis, A Powerful Scoring Technique Principal Component Analysis, A Powerful Scoring Technique George C. J. Fernandez, University of Nevada - Reno, Reno NV 89557 ABSTRACT Data mining is a collection of analytical techniques to uncover new

More information

One-Way ANOVA. Some examples of when ANOVA would be appropriate include:

One-Way ANOVA. Some examples of when ANOVA would be appropriate include: One-Way ANOVA 1. Purpose Analysis of variance (ANOVA) is used when one wishes to determine whether two or more groups (e.g., classes A, B, and C) differ on some outcome of interest (e.g., an achievement

More information

Likelihood and Fairness in Multidimensional Item Response Theory

Likelihood and Fairness in Multidimensional Item Response Theory Likelihood and Fairness in Multidimensional Item Response Theory or What I Thought About On My Holidays Giles Hooker and Matthew Finkelman Cornell University, February 27, 2008 Item Response Theory Educational

More information

Multidimensional Joint Graphical Display of Symmetric Analysis: Back to the Fundamentals

Multidimensional Joint Graphical Display of Symmetric Analysis: Back to the Fundamentals Multidimensional Joint Graphical Display of Symmetric Analysis: Back to the Fundamentals Shizuhiko Nishisato Abstract The basic premise of dual scaling/correspondence analysis lies in the simultaneous

More information

A COEFFICIENT OF DETERMINATION FOR LOGISTIC REGRESSION MODELS

A COEFFICIENT OF DETERMINATION FOR LOGISTIC REGRESSION MODELS A COEFFICIENT OF DETEMINATION FO LOGISTIC EGESSION MODELS ENATO MICELI UNIVESITY OF TOINO After a brief presentation of the main extensions of the classical coefficient of determination ( ), a new index

More information

Basic IRT Concepts, Models, and Assumptions

Basic IRT Concepts, Models, and Assumptions Basic IRT Concepts, Models, and Assumptions Lecture #2 ICPSR Item Response Theory Workshop Lecture #2: 1of 64 Lecture #2 Overview Background of IRT and how it differs from CFA Creating a scale An introduction

More information

Whats beyond Concerto: An introduction to the R package catr. Session 4: Overview of polytomous IRT models

Whats beyond Concerto: An introduction to the R package catr. Session 4: Overview of polytomous IRT models Whats beyond Concerto: An introduction to the R package catr Session 4: Overview of polytomous IRT models The Psychometrics Centre, Cambridge, June 10th, 2014 2 Outline: 1. Introduction 2. General notations

More information

Three-group ROC predictive analysis for ordinal outcomes

Three-group ROC predictive analysis for ordinal outcomes Three-group ROC predictive analysis for ordinal outcomes Tahani Coolen-Maturi Durham University Business School Durham University, UK tahani.maturi@durham.ac.uk June 26, 2016 Abstract Measuring the accuracy

More information

Advising on Research Methods: A consultant's companion. Herman J. Ader Gideon J. Mellenbergh with contributions by David J. Hand

Advising on Research Methods: A consultant's companion. Herman J. Ader Gideon J. Mellenbergh with contributions by David J. Hand Advising on Research Methods: A consultant's companion Herman J. Ader Gideon J. Mellenbergh with contributions by David J. Hand Contents Preface 13 I Preliminaries 19 1 Giving advice on research methods

More information

Paradoxical Results in Multidimensional Item Response Theory

Paradoxical Results in Multidimensional Item Response Theory UNC, December 6, 2010 Paradoxical Results in Multidimensional Item Response Theory Giles Hooker and Matthew Finkelman UNC, December 6, 2010 1 / 49 Item Response Theory Educational Testing Traditional model

More information

An Equivalency Test for Model Fit. Craig S. Wells. University of Massachusetts Amherst. James. A. Wollack. Ronald C. Serlin

An Equivalency Test for Model Fit. Craig S. Wells. University of Massachusetts Amherst. James. A. Wollack. Ronald C. Serlin Equivalency Test for Model Fit 1 Running head: EQUIVALENCY TEST FOR MODEL FIT An Equivalency Test for Model Fit Craig S. Wells University of Massachusetts Amherst James. A. Wollack Ronald C. Serlin University

More information

The Discriminating Power of Items That Measure More Than One Dimension

The Discriminating Power of Items That Measure More Than One Dimension The Discriminating Power of Items That Measure More Than One Dimension Mark D. Reckase, American College Testing Robert L. McKinley, Educational Testing Service Determining a correct response to many test

More information

Multidimensional Scaling in R: SMACOF

Multidimensional Scaling in R: SMACOF Multidimensional Scaling in R: SMACOF Patrick Mair Institute for Statistics and Mathematics WU Vienna University of Economics and Business Jan de Leeuw Department of Statistics University of California,

More information

Bayesian Methods for Testing Axioms of Measurement

Bayesian Methods for Testing Axioms of Measurement Bayesian Methods for Testing Axioms of Measurement George Karabatsos University of Illinois-Chicago University of Minnesota Quantitative/Psychometric Methods Area Department of Psychology April 3, 2015,

More information

The application and empirical comparison of item. parameters of Classical Test Theory and Partial Credit. Model of Rasch in performance assessments

The application and empirical comparison of item. parameters of Classical Test Theory and Partial Credit. Model of Rasch in performance assessments The application and empirical comparison of item parameters of Classical Test Theory and Partial Credit Model of Rasch in performance assessments by Paul Moloantoa Mokilane Student no: 31388248 Dissertation

More information

Likelihood-ratio tests for order-restricted log-linear models Galindo-Garre, F.; Vermunt, Jeroen; Croon, M.A.

Likelihood-ratio tests for order-restricted log-linear models Galindo-Garre, F.; Vermunt, Jeroen; Croon, M.A. Tilburg University Likelihood-ratio tests for order-restricted log-linear models Galindo-Garre, F.; Vermunt, Jeroen; Croon, M.A. Published in: Metodología de las Ciencias del Comportamiento Publication

More information

Score-Based Tests of Measurement Invariance with Respect to Continuous and Ordinal Variables

Score-Based Tests of Measurement Invariance with Respect to Continuous and Ordinal Variables Score-Based Tests of Measurement Invariance with Respect to Continuous and Ordinal Variables Achim Zeileis, Edgar C. Merkle, Ting Wang http://eeecon.uibk.ac.at/~zeileis/ Overview Motivation Framework Score-based

More information

Latent Class Analysis for Models with Error of Measurement Using Log-Linear Models and An Application to Women s Liberation Data

Latent Class Analysis for Models with Error of Measurement Using Log-Linear Models and An Application to Women s Liberation Data Journal of Data Science 9(2011), 43-54 Latent Class Analysis for Models with Error of Measurement Using Log-Linear Models and An Application to Women s Liberation Data Haydar Demirhan Hacettepe University

More information

Algebra II Vocabulary Alphabetical Listing. Absolute Maximum: The highest point over the entire domain of a function or relation.

Algebra II Vocabulary Alphabetical Listing. Absolute Maximum: The highest point over the entire domain of a function or relation. Algebra II Vocabulary Alphabetical Listing Absolute Maximum: The highest point over the entire domain of a function or relation. Absolute Minimum: The lowest point over the entire domain of a function

More information

Nominal Data. Parametric Statistics. Nonparametric Statistics. Parametric vs Nonparametric Tests. Greg C Elvers

Nominal Data. Parametric Statistics. Nonparametric Statistics. Parametric vs Nonparametric Tests. Greg C Elvers Nominal Data Greg C Elvers 1 Parametric Statistics The inferential statistics that we have discussed, such as t and ANOVA, are parametric statistics A parametric statistic is a statistic that makes certain

More information

Workshop Research Methods and Statistical Analysis

Workshop Research Methods and Statistical Analysis Workshop Research Methods and Statistical Analysis Session 2 Data Analysis Sandra Poeschl 08.04.2013 Page 1 Research process Research Question State of Research / Theoretical Background Design Data Collection

More information

Principal Component Analysis for Mixed Quantitative and Qualitative Data

Principal Component Analysis for Mixed Quantitative and Qualitative Data 1 Principal Component Analysis for Mixed Quantitative and Qualitative Data Susana Agudelo-Jaramillo sagudel9@eafit.edu.co Manuela Ochoa-Muñoz mochoam@eafit.edu.co Francisco Iván Zuluaga-Díaz fzuluag2@eafit.edu.co

More information

The Model Building Process Part I: Checking Model Assumptions Best Practice

The Model Building Process Part I: Checking Model Assumptions Best Practice The Model Building Process Part I: Checking Model Assumptions Best Practice Authored by: Sarah Burke, PhD 31 July 2017 The goal of the STAT T&E COE is to assist in developing rigorous, defensible test

More information

ESTIMATION OF IRT PARAMETERS OVER A SMALL SAMPLE. BOOTSTRAPPING OF THE ITEM RESPONSES. Dimitar Atanasov

ESTIMATION OF IRT PARAMETERS OVER A SMALL SAMPLE. BOOTSTRAPPING OF THE ITEM RESPONSES. Dimitar Atanasov Pliska Stud. Math. Bulgar. 19 (2009), 59 68 STUDIA MATHEMATICA BULGARICA ESTIMATION OF IRT PARAMETERS OVER A SMALL SAMPLE. BOOTSTRAPPING OF THE ITEM RESPONSES Dimitar Atanasov Estimation of the parameters

More information

A Note on Item Restscore Association in Rasch Models

A Note on Item Restscore Association in Rasch Models Brief Report A Note on Item Restscore Association in Rasch Models Applied Psychological Measurement 35(7) 557 561 ª The Author(s) 2011 Reprints and permission: sagepub.com/journalspermissions.nav DOI:

More information

Regression. Oscar García

Regression. Oscar García Regression Oscar García Regression methods are fundamental in Forest Mensuration For a more concise and general presentation, we shall first review some matrix concepts 1 Matrices An order n m matrix is

More information

CHAPTER 17 CHI-SQUARE AND OTHER NONPARAMETRIC TESTS FROM: PAGANO, R. R. (2007)

CHAPTER 17 CHI-SQUARE AND OTHER NONPARAMETRIC TESTS FROM: PAGANO, R. R. (2007) FROM: PAGANO, R. R. (007) I. INTRODUCTION: DISTINCTION BETWEEN PARAMETRIC AND NON-PARAMETRIC TESTS Statistical inference tests are often classified as to whether they are parametric or nonparametric Parameter

More information

Assessing Factorial Invariance in Ordered-Categorical Measures

Assessing Factorial Invariance in Ordered-Categorical Measures Multivariate Behavioral Research, 39 (3), 479-515 Copyright 2004, Lawrence Erlbaum Associates, Inc. Assessing Factorial Invariance in Ordered-Categorical Measures Roger E. Millsap and Jenn Yun-Tein Arizona

More information

Equivalency of the DINA Model and a Constrained General Diagnostic Model

Equivalency of the DINA Model and a Constrained General Diagnostic Model Research Report ETS RR 11-37 Equivalency of the DINA Model and a Constrained General Diagnostic Model Matthias von Davier September 2011 Equivalency of the DINA Model and a Constrained General Diagnostic

More information

Nonparametric tests for the Rasch model: explanation, development, and application of quasi-exact tests for small samples

Nonparametric tests for the Rasch model: explanation, development, and application of quasi-exact tests for small samples Nonparametric tests for the Rasch model: explanation, development, and application of quasi-exact tests for small samples Ingrid Koller 1* & Reinhold Hatzinger 2 1 Department of Psychological Basic Research

More information

I L L I N O I S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN

I L L I N O I S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN Introduction Edps/Psych/Stat/ 584 Applied Multivariate Statistics Carolyn J Anderson Department of Educational Psychology I L L I N O I S UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN c Board of Trustees,

More information

Introduction to Basic Statistics Version 2

Introduction to Basic Statistics Version 2 Introduction to Basic Statistics Version 2 Pat Hammett, Ph.D. University of Michigan 2014 Instructor Comments: This document contains a brief overview of basic statistics and core terminology/concepts

More information

Because it might not make a big DIF: Assessing differential test functioning

Because it might not make a big DIF: Assessing differential test functioning Because it might not make a big DIF: Assessing differential test functioning David B. Flora R. Philip Chalmers Alyssa Counsell Department of Psychology, Quantitative Methods Area Differential item functioning

More information

Two Correlated Proportions Non- Inferiority, Superiority, and Equivalence Tests

Two Correlated Proportions Non- Inferiority, Superiority, and Equivalence Tests Chapter 59 Two Correlated Proportions on- Inferiority, Superiority, and Equivalence Tests Introduction This chapter documents three closely related procedures: non-inferiority tests, superiority (by a

More information

Ch. 16: Correlation and Regression

Ch. 16: Correlation and Regression Ch. 1: Correlation and Regression With the shift to correlational analyses, we change the very nature of the question we are asking of our data. Heretofore, we were asking if a difference was likely to

More information

Quasi-Exact Tests for the dichotomous Rasch Model

Quasi-Exact Tests for the dichotomous Rasch Model Introduction Rasch model Properties Quasi-Exact Tests for the dichotomous Rasch Model Y unidimensionality/homogenous items Y conditional independence (local independence) Ingrid Koller, Vienna University

More information

More Powerful Tests for Homogeneity of Multivariate Normal Mean Vectors under an Order Restriction

More Powerful Tests for Homogeneity of Multivariate Normal Mean Vectors under an Order Restriction Sankhyā : The Indian Journal of Statistics 2007, Volume 69, Part 4, pp. 700-716 c 2007, Indian Statistical Institute More Powerful Tests for Homogeneity of Multivariate Normal Mean Vectors under an Order

More information

Development and Calibration of an Item Response Model. that Incorporates Response Time

Development and Calibration of an Item Response Model. that Incorporates Response Time Development and Calibration of an Item Response Model that Incorporates Response Time Tianyou Wang and Bradley A. Hanson ACT, Inc. Send correspondence to: Tianyou Wang ACT, Inc P.O. Box 168 Iowa City,

More information

Comparing IRT with Other Models

Comparing IRT with Other Models Comparing IRT with Other Models Lecture #14 ICPSR Item Response Theory Workshop Lecture #14: 1of 45 Lecture Overview The final set of slides will describe a parallel between IRT and another commonly used

More information

Studies on the effect of violations of local independence on scale in Rasch models: The Dichotomous Rasch model

Studies on the effect of violations of local independence on scale in Rasch models: The Dichotomous Rasch model Studies on the effect of violations of local independence on scale in Rasch models Studies on the effect of violations of local independence on scale in Rasch models: The Dichotomous Rasch model Ida Marais

More information

examples of how different aspects of test information can be displayed graphically to form a profile of a test

examples of how different aspects of test information can be displayed graphically to form a profile of a test Creating a Test Information Profile for a Two-Dimensional Latent Space Terry A. Ackerman University of Illinois In some cognitive testing situations it is believed, despite reporting only a single score,

More information

Anders Skrondal. Norwegian Institute of Public Health London School of Hygiene and Tropical Medicine. Based on joint work with Sophia Rabe-Hesketh

Anders Skrondal. Norwegian Institute of Public Health London School of Hygiene and Tropical Medicine. Based on joint work with Sophia Rabe-Hesketh Constructing Latent Variable Models using Composite Links Anders Skrondal Norwegian Institute of Public Health London School of Hygiene and Tropical Medicine Based on joint work with Sophia Rabe-Hesketh

More information

Selection of Variable Selecting the right variable for a control chart means understanding the difference between discrete and continuous data.

Selection of Variable Selecting the right variable for a control chart means understanding the difference between discrete and continuous data. Statistical Process Control, or SPC, is a collection of tools that allow a Quality Engineer to ensure that their process is in control, using statistics. Benefit of SPC The primary benefit of a control

More information

Decoupled Collaborative Ranking

Decoupled Collaborative Ranking Decoupled Collaborative Ranking Jun Hu, Ping Li April 24, 2017 Jun Hu, Ping Li WWW2017 April 24, 2017 1 / 36 Recommender Systems Recommendation system is an information filtering technique, which provides

More information

Chapter 1. Modeling Basics

Chapter 1. Modeling Basics Chapter 1. Modeling Basics What is a model? Model equation and probability distribution Types of model effects Writing models in matrix form Summary 1 What is a statistical model? A model is a mathematical

More information

Polytomous Item Explanatory IRT Models with Random Item Effects: An Application to Carbon Cycle Assessment Data

Polytomous Item Explanatory IRT Models with Random Item Effects: An Application to Carbon Cycle Assessment Data Polytomous Item Explanatory IRT Models with Random Item Effects: An Application to Carbon Cycle Assessment Data Jinho Kim and Mark Wilson University of California, Berkeley Presented on April 11, 2018

More information

Logistic Regression and Item Response Theory: Estimation Item and Ability Parameters by Using Logistic Regression in IRT.

Logistic Regression and Item Response Theory: Estimation Item and Ability Parameters by Using Logistic Regression in IRT. Louisiana State University LSU Digital Commons LSU Historical Dissertations and Theses Graduate School 1998 Logistic Regression and Item Response Theory: Estimation Item and Ability Parameters by Using

More information

MATHEMATICS. Units Topics Marks I Relations and Functions 10

MATHEMATICS. Units Topics Marks I Relations and Functions 10 MATHEMATICS Course Structure Units Topics Marks I Relations and Functions 10 II Algebra 13 III Calculus 44 IV Vectors and 3-D Geometry 17 V Linear Programming 6 VI Probability 10 Total 100 Course Syllabus

More information

On Kusuoka Representation of Law Invariant Risk Measures

On Kusuoka Representation of Law Invariant Risk Measures MATHEMATICS OF OPERATIONS RESEARCH Vol. 38, No. 1, February 213, pp. 142 152 ISSN 364-765X (print) ISSN 1526-5471 (online) http://dx.doi.org/1.1287/moor.112.563 213 INFORMS On Kusuoka Representation of

More information

NFC ACADEMY COURSE OVERVIEW

NFC ACADEMY COURSE OVERVIEW NFC ACADEMY COURSE OVERVIEW Algebra II Honors is a full-year, high school math course intended for the student who has successfully completed the prerequisite course Algebra I. This course focuses on algebraic

More information

Regression Graphics. 1 Introduction. 2 The Central Subspace. R. D. Cook Department of Applied Statistics University of Minnesota St.

Regression Graphics. 1 Introduction. 2 The Central Subspace. R. D. Cook Department of Applied Statistics University of Minnesota St. Regression Graphics R. D. Cook Department of Applied Statistics University of Minnesota St. Paul, MN 55108 Abstract This article, which is based on an Interface tutorial, presents an overview of regression

More information

Item Reliability Analysis

Item Reliability Analysis Item Reliability Analysis Revised: 10/11/2017 Summary... 1 Data Input... 4 Analysis Options... 5 Tables and Graphs... 5 Analysis Summary... 6 Matrix Plot... 8 Alpha Plot... 10 Correlation Matrix... 11

More information

Types of Statistical Tests DR. MIKE MARRAPODI

Types of Statistical Tests DR. MIKE MARRAPODI Types of Statistical Tests DR. MIKE MARRAPODI Tests t tests ANOVA Correlation Regression Multivariate Techniques Non-parametric t tests One sample t test Independent t test Paired sample t test One sample

More information

Online Appendix to Yes, But What s the Mechanism? (Don t Expect an Easy Answer) John G. Bullock, Donald P. Green, and Shang E. Ha

Online Appendix to Yes, But What s the Mechanism? (Don t Expect an Easy Answer) John G. Bullock, Donald P. Green, and Shang E. Ha Online Appendix to Yes, But What s the Mechanism? (Don t Expect an Easy Answer) John G. Bullock, Donald P. Green, and Shang E. Ha January 18, 2010 A2 This appendix has six parts: 1. Proof that ab = c d

More information

Similarity and Dissimilarity

Similarity and Dissimilarity 1//015 Similarity and Dissimilarity COMP 465 Data Mining Similarity of Data Data Preprocessing Slides Adapted From : Jiawei Han, Micheline Kamber & Jian Pei Data Mining: Concepts and Techniques, 3 rd ed.

More information

The Model Building Process Part I: Checking Model Assumptions Best Practice (Version 1.1)

The Model Building Process Part I: Checking Model Assumptions Best Practice (Version 1.1) The Model Building Process Part I: Checking Model Assumptions Best Practice (Version 1.1) Authored by: Sarah Burke, PhD Version 1: 31 July 2017 Version 1.1: 24 October 2017 The goal of the STAT T&E COE

More information

Problems with parallel analysis in data sets with oblique simple structure

Problems with parallel analysis in data sets with oblique simple structure Methods of Psychological Research Online 2001, Vol.6, No.2 Internet: http://www.mpr-online.de Institute for Science Education 2001 IPN Kiel Problems with parallel analysis in data sets with oblique simple

More information

Log-linear multidimensional Rasch model for capture-recapture

Log-linear multidimensional Rasch model for capture-recapture Log-linear multidimensional Rasch model for capture-recapture Elvira Pelle, University of Milano-Bicocca, e.pelle@campus.unimib.it David J. Hessen, Utrecht University, D.J.Hessen@uu.nl Peter G.M. Van der

More information

2. TRIGONOMETRY 3. COORDINATEGEOMETRY: TWO DIMENSIONS

2. TRIGONOMETRY 3. COORDINATEGEOMETRY: TWO DIMENSIONS 1 TEACHERS RECRUITMENT BOARD, TRIPURA (TRBT) EDUCATION (SCHOOL) DEPARTMENT, GOVT. OF TRIPURA SYLLABUS: MATHEMATICS (MCQs OF 150 MARKS) SELECTION TEST FOR POST GRADUATE TEACHER(STPGT): 2016 1. ALGEBRA Sets:

More information