(controlled by USCALE=, UMEAN=, UDECIM=, LOCAL=, TOTAL=)
PERSON STATISTICS: MEASURE ORDER
Above the Table are shown the "real" separation coefficent and reliability (separation index) coefficients from Table 3.
ENTRY NUMBER is the sequence number of the person, or item, in your data, and is the reference number used for deletion or anchoring.
"PERSONS" or "ITEMS", etc. is the item name or person-identifying label.
TOTAL SCORE or RAW SCORE is the raw score corresponding to the parameter, i.e., the raw score by a person on the test, or the sum of the scored responses to an item by the persons. Totalscore=Yes to include all responses. Totalscore=No to omit responses in extreme scores.
COUNT is the number of data points used to construct measures. Totalscore=Yes to include all responses. Totalscore=No to omit responses in extreme scores.
MEASURE is the estimate (or calibration) for the parameter, i.e., person ability (theta, B, beta, etc.), or the item difficulty (b, D, delta, etc.). Values are reported in logits with two decimal places, unless rescaled by USCALE=, UIMEAN=, UPMEAN=, UDECIM=. If the score is extreme, a value is estimated, but as MAXIMUM (perfect score) or MINIMUM (zero score). No measure is reported if the element is DROPPED (no valid observations remaining) or DELETED (you deleted the person or item). The difficulty of an item is defined to be the point on the latent variable at which its high and low categories are equally probable. SAFILE= can be used to alter this definition.
If unexpected results are reported, check whether TARGET= or CUTLO= or CUTHI= or ISGROUPS= are specified.
INESTIMABLE is reported if all observations are eliminated as forming part of extreme response strings.
ERROR is the standard error of the estimate. For anchored values, an "A" is shown on the listing and the error reported is that which would have been obtained if the value had been estimated. Values are reported in logits with two decimal places, unless rescaled by USCALE=, UDECIM=
INFIT is a t standardized information-weighted mean square statistic, which is more sensitive to unexpected behavior affecting responses to items near the person's measure level.
MNSQ is the mean-square infit statistic with expectation 1. Values substantially less than 1 indicate dependency in your data; values substantially greater than 1 indicate noise. See dichotomous and polytomous fit statistics.
Value Meaning
>2.0 Off-variable noise is greater than useful information. Degrades measurement.
>1.5 Noticeable off-variable noise. Neither constructs nor degrades measurement
0.5 - 1.5 Productive of measurement
<0.5 Overly predictable. Misleads us into thinking we are measuring better than we really are. (Attenuation paradox.)
Always remedy the large misfits first. Misfits <1.0 are only of concern when shortening a test.
ZSTD is the infit mean-square fit statistic t standardized to approximate a theoretical "unit normal", mean 0 and variance 1, distribution. ZSTD (standardized as a z-score) is used of a t-test result when either the t-test value has effectively infinite degrees of freedom (i.e., approximates a unit normal value) or the Student's t-statistic distribution value has been adjusted to a unit normal value. The standardization is shown on RSA, p.100-101. When LOCAL=Y, then EMP is shown, indicating a local {0,1} standardization. When LOCAL=LOG, then LOG is shown, and the natural logarithms of the mean-squares are reported. More exact values are shown in the Output Files.
Ben Wright advises: "ZSTD is only useful to salvage non-significant MNSQ>1.5, when sample size is small or test length is short."
OUTFIT is a t standardized outlier-sensitive mean square fit statistic, more sensitive to unexpected behavior by persons on items far from the person's measure level.
| Value | Meaning |
| >2.0 | Off-variable noise is greater than useful information. Degrades measurement. |
| >1.5 | Noticeable off-variable noise. Neither constructs nor degrades measurement |
| 0.5 - 1.5 | Productive of measurement |
| <0.5 | Overly predictable. Misleads us into thinking we are measuring better than we really are. (Attenuation paradox.) |
Always remedy the large misfits first. Misfits <1.0 are usually only of concern when shortening a test.
MNSQ is the mean-square outfit statistic with expectation 1. Values substantially less than 1 indicate dependency in your data; values substantially greater than 1 indicate the presence of unexpected outliers. See dichotomous and polytomous fit statistics.
ZSTD is the outfit mean-square fit statistic t standardized similarly to the INFIT ZSTD. ZSTD (standardized as a z-score) is used of a t-test result when either the t-test value has effectively infinite degrees of freedom (i.e., approximates a unit normal value) or the Student's t-statistic distribution value has been adjusted to a unit normal value.
Ben Wright advises: "ZSTD is only useful to salvage non-significant MNSQ>1.5, when sample size is small or test length is short."
PTBSE CORR (reported when PTBIS=Yes) or PTBSA CORR (reported when PTBIS=All) is the point-biserial correlation, rpbis, between the individual item (or person) response "scores" and the total person (or item) test score (less the individual response "scores"). Negative values for items often indicate mis-scoring, or rating (or partial credit) scale items with reversed direction. Letters indicating the identity of persons or items appearing on the fit plots appear under PTBSE. For adaptive tests, an rpbis near zero is expected. See Correlations.
PTMEA CORR. (reported when PTBISERIAL=N) is the point-measure correlation, rpm or RPM, between the observations on an item (as fractions with the range 0,1) and the corresponding person measures, or vice-versa. Since the point-biserial loses its meaning in the presence of missing data, specify PTBISERIAL=N when there are missing data or when CUTLO= or CUTHI= are specified. The point-measure correlation has a range of -1 to +1.
EXP. is the expected value of the correlation when the data fit the Rasch model with the estimated measures.
EXACT MATCH
OBS% Observed% is the percent of data points which are within 0.5 score points of their expected values, i.e., that match predictions.
EXP% Expected% is the percent of data points that are predicted to be within 0.5 score points of their expected values.
ESTIM DISCRIM is an estimate of the item discrimination, see DISCRIM=
ASYMPTOTE LOWER and UPPER are estimates of the upper and lower asymptotes for dichotomous items, see ASYMPTOTE=
WEIGH is the weight assigned by IWEIGHT= or PWEIGHT=
DISPLACE is the displacement of the reported MEASURE from its data-derived value. This should only be shown with anchored measures. If small displacements are being shown, try tightening the convergence criteria, LCONV=.
G is the grouping code assigned with ISGROUPS=
M is the model code assigned with MODELS=
Component is the name of the list of persons (rows) or items (columns) reported here
Label is the person or item label.
Above the Table are shown the "real" separation and reliability coefficients from Table 3.
If SUBSET: appears, see Subsets.