Ulf H. Olsson

Analyzing data sets with missing data: an empirical evaluation of imputation methods and likelihood-based methods

Missing data are often encountered in data sets used to construct software effort prediction models. Thus far, the common practice has been to ignore observations with missing data. This may result in biased prediction models. The authors evaluate four missing data techniques (MDTs) in the context of software cost modeling: listwise deletion (LD), mean imputation (MI), similar response pattern imputation (SRPI), and full information maximum likelihood (FIML). We apply the MDTs to an ERP data set, and thereafter construct regression-based prediction models using the resulting data sets. The evaluation suggests that only FIML is appropriate when the data are not missing completely at random (MCAR). Unlike FIML, prediction models constructed on LD, MI and SRPI data sets will be biased unless the data are MCAR. Furthermore, compared to LD, MI and SRPI seem appropriate only if the resulting LD data set is too small to enable the construction of a meaningful regression-based prediction model.

Two Equivalent Discrepancy Functions for Maximum Likelihood Estimation: Do Their Test Statistics Follow a Non-Central Chi-Square Distribution under Model Misspecification?

Over the years several discrepancy functions have been introduced both in the literature and in the software of Structural Equation Modeling (SEM). The test statistics for the discrepancy functions associated with Maximum Likelihood (ML), Generalized Least Squares (GLS), and Normal Theory Weighted Least Squares (NWLS) are all asymptotically equivalent. These test statistics are all approximately distributed as central chi-square under correct model specification and if the observed variables are multivariate normally distributed. However, it is known that the distribution of these test statistics will not approximate a central Chi-square distribution for models containing specification error, but is more likely to follow a non-central Chi-square distribution (Browne 1984). This study investigates the empirical distributions of the ML and NWLS discrepancy functions. The study includes 13 different factor models with different types and degrees of specification error. It is found, except for small samples, that the empirical distribution of the ML-test statistic outperforms the empirical distribution of the NWLS-test statistic in terms of approximation to the theoretical non-central Chi-square distribution. Furthermore, in some cases, it turned out that the non-central Chi-square approximation was not appropriate even for models that contained minor and moderate degrees of specification error.

Management orientation and export performance: the case of Norwegian ICT companies

Purpose – The purpose of this paper is to contrast three management orientations relevant for exporters: export, technology and customer orientations. The general hypothesis is that all orientations covariate positively with export performance. However, an alternative hypothesis regarding customer relations is propounded (negative impact on performance).Design/methodology/approach – Regression‐based techniques are used.Findings – The results support the hypotheses that export performance increases with export commitment. Technology orientation correlates positively with export performance. On the other hand, the much venerated customer orientation shows negative correlation with export performance.Originality/value – This paper argues that customer orientation may turn into what might be called customer obsession, without due attention to cost consequences and strategic orientation. Also, too much customer orientation may lead the firm away from its ability to innovate, leaving the company behind its comp...

Assessing the benefits of imputing ERP projects with missing data

Incomplete, or missing data is likely to be encountered in empirical software engineering data sets. The authors evaluate some methods for handling missing data. The methods are presented and discussed in general and thereafter applied to effort estimation of ERP projects. We found that two sampling based methods, mean imputation (MI) and similar response pattern imputation (SRPI), waste less information than listwise deletion (LD). However, MI may introduce more bias than the SRPI method. Compared to sampling based methods, likelihood based imputation methods require too large data sets to be realistic to use in empirical software engineering. None of the sampling based methods, such as MI and SRPI, seem able to correct bias. So, though imputation is an attractive idea, the available methods still have severe limitations.

Does the ADF fit function decrease when the kurtosis increases

In this study we demonstrate how the asymptotically distribution-free (ADF) fit function is affected by (excessive) kurtosis in the observed data. More specifically, we address how different levels of univariate kurtosis affect fit values (and therefore fit indices) for misspecified factor models. By using numerical calculation, we show (for 13 factor models) that the probability limit F(0) of F empty set for the ADF fit function decreases considerably as the kurtosis increases. We also give a formal proof that the value of F(0) decreases monotonically with the kurtosis for a whole class of structural equation models.

Effects of hosting a mega-sport event on country image.

Nations and cities compete to host international mega-sport events such as the Olympic Games even if very large costs are incurred. Country image may be changed by hosting such events, and country image dimensions are in turn related to product image and behavioral intentions regarding product purchase and tourism. In this article a model of these relationships is developed, based on several streams of literature. The empirical study reported relates to the Winter Olympics in Turin, Italy in 2006. A quasi-experimental design was employed based upon two samples of undergraduate students in Norway. Data was gathered both before and after the Olympic Games took place. The study indicates that dimensions of country image for those being very interested in sports may be changed by hosting a mega-sport event. However, there is no guarantee that the image of the host country will improve. it may actually deteriorate. This finding underscores the importance of managing international sport events properly.

A Simple Simulation Technique for Nonnormal Data with Prespecified Skewness, Kurtosis, and Covariance Matrix.

ABSTRACT We present and investigate a simple way to generate nonnormal data using linear combinations of independent generator (IG) variables. The simulated data have prespecified univariate skewness and kurtosis and a given covariance matrix. In contrast to the widely used Vale-Maurelli (VM) transform, the obtained data are shown to have a non-Gaussian copula. We analytically obtain asymptotic robustness conditions for the IG distribution. We show empirically that popular test statistics in covariance analysis tend to reject true models more often under the IG transform than under the VM transform. This implies that overly optimistic evaluations of estimators and fit statistics in covariance structure analysis may be tempered by including the IG transform for nonnormal data generation. We provide an implementation of the IG transform in the R environment.

Multivariate Analysis with LISREL

There are many books available on multivariate statistical analysis and many books have been written about structural equation modeling (SEM) and on LISREL. But this book is unique in the sense of being the only one that covers both the statistical theory and methodology and how to do the analysis with LISREL. It does not only cover the typical uses of LISREL such as confirmatory factor analysis (CFA) and structural equation models (SEM) but also several other topics of multivariate analysis such as regression (univariate, multivariate, censored, logistic, and probit), generalized linear models, multilevel analysis, and principal components analysis. There is no other book with such a full and detailed coverage of all the models, methods and procedures one can use with LISREL.

Correcting Too Much or Too Little? The Performance of Three Chi-Square Corrections

This simulation study investigates the performance of three test statistics, T1, T2, and T3, used to evaluate structural equation model fit under non normal data conditions. T1 is the well-known mean-adjusted statistic of Satorra and Bentler. T2 is the mean-and-variance adjusted statistic of Sattertwaithe type where the degrees of freedom is manipulated. T3 is a recently proposed version of T2 that does not manipulate degrees of freedom. Discrepancies between these statistics and their nominal chi-square distribution in terms of errors of Type I and Type II are investigated. All statistics are shown to be sensitive to increasing kurtosis in the data, with Type I error rates often far off the nominal level. Under excess kurtosis true models are generally over-rejected by T1 and under-rejected by T2 and T3, which have similar performance in all conditions. Under misspecification there is a loss of power with increasing kurtosis, especially for T2 and T3. The coefficient of variation of the nonzero eigenvalues of a certain matrix is shown to be a reliable indicator for the adequacy of these statistics.

The effect of kurtosis on the power of two test statistics in covariance structure analysis.

We study the effect of excess kurtosis on the non-centrality parameters of the rescaled and the residual-based test statistics for covariance structure models. The analysis is based on population matrices and parameters, which eliminates the sampling variability inherent in simulation studies. We show that the non-centrality parameters, and consequently the asymptotic power, decrease as kurtosis in the data increases. Examples are provided to compare this decrease for the two test statistics, and to illustrate how substantial it is.