Abstract
Recent versions of LISREL (Joreskög and Sorbom, 1983) contain procedures for estimating polyserial and polychoric correlations from crude rank category measures. In this paper, the accuracy of these procedures for estimating the relationship between constructs in both a single and a multiple indicator model is compared to that of using Pearsonian correlations based on equal distance scoring of the rank categories. In these comparisons, a variety of multivariate nonnormal distributions were simulated and the average bias and average absolute error of the estimated relationships between underlying constructs were calculated. These estimates were affected by the number of rank categories for each measure, the correlations among measures, their skewness and kurtosis, and the correlation between underlying constructs. The most important finding is that although the polychoric procedure can be helpful in estimating the correlation between unobserved variables in single indicator models, it does not improve estimates based on Pearsonian correlations in the multiple indicator model.
Similar content being viewed by others
References
K.Bollen & K.Barb. (1981). “Pearson's R and coarsely categorized measures”, American Sociological Review 46: 232–239.
C.L.Bliss. (1967). Statistics in Biology (Volume 1). New York: McGraw-Hill.
R.P.Gephart. (1983). “Multiple R, the ‘parametric strategy’, and measurement imprecision”, Sociological Perspectives 26: 473–500.
D.R.Johnson & J.C.Creech. (1983). “Ordinal measures in multiple indicator models: a simulation study of categorization error”, American Sociological Review 47: 398–407.
K.G.Joreskog & D.Sorbom. (1981). LISREL VI: Analysis of Linear Structural Relationships by Maximum Likelihood and Least Squares Methods. Chicago: International Educational Services.
D.R.Lehmann & J.Hulbert. (1982). “Are three-point scales always good enough?” Journal of Marketing Research 19: 444–445.
W.S.Martin. (1973). “The effects of scaling on the correlation coefficient: a test of validity”, Journal of Marketing Research 10: 316–318.
W.S.Martin. (1978). “Effects of scaling on the correlation coefficient: additional considerations”, Journal of Marketing Research 15: 304–308.
R.M.O'Brien. (1981a). “Reducing grouping distortion in rank category variables”, American Journal of Political Science 25: 605–616.
R.M.O'Brien. (1981b). “Using rank category variables to represent continuous variables: defects of common practice”, Social Forces 61: 144–155.
R.M.O'Brien. (1983). “Rank order versus rank category measures of continuous variables”, (comment on Bollen and Barb, ASR, April 1981, and Henry, ASR, April 1982). American Sociological Review 48: 284–286.
R.M.O'Brien & P.H.Homer. (1987). “Corrections for coarsely categorized measures: LIS-REL's polyserial and polychoric correlations”, Quality and Quantity 21: 349–360.
U.Olsson. (1979). “Maximum likelihood estimation of the polychoric correlation coefficient”, Psychometrika 44: 443–460.
U.Olsson, F.Drasgow & N.J.Dorars. (1982). “The polyserial correlation coefficient”, Psychometrika 47: 337–347.
C.Vale, C.David & V.A.Maurelli. (1983). “Simulating multivariate nonnormal distributions”, Psychometrika 48: 465–474.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Homer, P., O'Brien, R.M. Using LISREL models with crude rank category measures. Qual Quant 22, 191–201 (1988). https://doi.org/10.1007/BF00223041
Issue Date:
DOI: https://doi.org/10.1007/BF00223041