Introducing LISREL

Citation

Palmer, A. (2003), "Introducing LISREL", European Journal of Marketing, Vol. 37 No. 5/6, pp. 937-939. https://doi.org/10.1108/03090560310465206

I came away from this introductory guide to LISREL with feelings of both happiness and annoyance. I was happy that in this concise guidebook, students now have one less reason for being frightened of LISREL. But I couldn't help feeling annoyed that recent refinements to LISREL, and guides such as this one, were not available to me when I struggled to master LISREL.

LISREL has emerged as a pre‐eminent quantitative data analysis tool, combining regression techniques with an analysis of variance associated with the use of multiple indicators of underlying phenomena. In achieving its pre‐eminence, the language and techniques of LISREL have remained mysterious to many, resulting in a learning curve too steep to contemplate. This short guide to version 8 of LISREL is aimed clearly at the novice user, for whom many existing manuals have probably led many students to develop their own rationale for pursuing a qualitative research strategy. The authors empathise with the fright felt by many such novice users and apply accessible language and helpful examples to introduce the reader to the basic principles of LISREL.

The authors continually remind us that LISREL is a powerful tool, and like any powerful tool in the wrong hands, it can be dangerous. Throughout the book there is therefore continual stress on the need for model building to be theory driven and that no amount of analysis will justify a poorly specified model. Model development takes us through path diagrams and the still seemingly arcane Greek notation used to identify the components of the structural and measurement models. The SIMPLIS programme has greatly simplified the process of parameter specification, and the authors make only passing reference to the matrices which were previously an essential, but tedious element of model development.

A lot of attention is paid to the assessment of models for goodness of fit, and a useful guide is provided through the range of indicators available to the researcher. From this, model modification is discussed. Structural equations modelling is often brought into disrepute by those who mechanically modify models to fit the available data. This has led many more qualitative‐orientated researchers to dismiss the value of the technique as one that merely establishes sometimes spurious associations between data, devoid of a holistic view of an issue which qualitative techniques can uncover. Stephen Brown once famously dismissed users of structural equations modelling as narrow‐minded “Lisrelites”. The authors go to great lengths to stress that any modification to a model is only as good as the theoretical underpinning of such modifications.

Great use is made of appendices and footnotes in a way which satisfies the needs of the curious, while not causing the novice user to become overwhelmed with what may appear at first to be very confusing detail. It may not be strictly necessary, for example to know the mathematical derivation of a covariance matrix, but should the reader wish for an explanation, one is readily provided in an appendix. Footnotes provide further paths down which the curious can wander and it is in these footnotes that the authors’ mastery of the subject really shines through. One such footnote, for example, brings the reader up to date with recent discussion about the effects of the number of indicators per latent variable on convergence, parameter stability and construct reliability. For readers seeking further justification of the guidance contained in this book, there is an extensive bibliography, including references to many recent developments in the application of LISREL.

As a statistical package, LISREL has uses beyond marketing, but the examples presented in this book have particular relevance to marketing problems. If a student wishes to explore whether they should be using LISREL as part of their research methodology, this book will provide a concise and useful introduction. The reader may well take the authors’ advice and get no further than the first chapter, where warnings are given about situations in which LISREL should not be used.

My only slight criticism of this introductory book is that it might have been useful to say a little more in the introduction about the relative merits of LISREL and other more recent structural equations analysis programmes. There is a great danger of LISREL becoming self‐perpetuating among researchers who may not be inclined to venture beyond the programme which they grew up with, despite many favourable reviews of AMOS, for example. But this is a minor issue, and above all else the book does what it's title promises – it introduces LISREL in a very user‐friendly way, while retaining academic rigour. For students whose research may benefit from structural equations analysis, this is a very good book, firstly to identify whether structural equations modelling is for them, and secondly to provide clear guidance on the issues involved in model building and testing.

If you encounter a research student for whom the very word LISREL induces feelings of fear, quietly recommend that they read this book. They will thank you for it. With increasingly user‐friendly versions of LISREL being released and guide books such as this published, LISREL really should be accessible to all.