What do you think?
Rate this book
Using Multivariate Statistics
Using Multivariate Statistics provides practical guidelines for conducting numerous types of multivariate statistical analyses. It gives syntax and output for accomplishing many analyses through the most recent releases of SAS, SPSS, and SYSTAT, some not available in software manuals. The book maintains its practical approach, still focusing on the benefits and limitations of applications of a technique to a data set — when, why, and how to do it. Overall, it provides advanced students with a timely and comprehensive introduction to today's most commonly encountered statistical and multivariate techniques, while assuming only a limited knowledge of higher-level mathematics.
980 pages, Hardcover
First published January 1, 1983
18 people are currently reading
187 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 18 of 18 reviews
August 30, 2009
I sat in the Chinese restaurant next to the Eagle super market in Normal, Illinois, during the spring semester of 1999 while I was completing a master's degree in clinical psychology. I was talking to Charu Thakral about how I wanted to take more statistics classes, especially multivariate statistics, but could not do so because I had not taken the Experimental Design class yet (which was an entire class basically devoted to the Analysis of Variance). Just then, Matthew Hesson-McInnis, the head of the quantitative psychology program at Illinois State came in with Glenn Reeder, a social psychologist also on the faculty. As Charu knew Matthew and Glenn, she said hello and introduced me to Matthew, somewhat embarrassingly. She told him I was interested in taking his multivariate class, but that I hadn't taken Experimental Design yet. He said that they were changing pre-requisites, and that Experimental Design would no longer be a prerequisite for Multivariate Analysis.
I read the third edition of this book when Matthew was my teacher. Matthew had a Ph.D. in quantitative psychology from the University of Illinois, and knew more about statistics than most people I've met, even today. Unfortunately, he wasn't particularly disciplined, and would often get off topic. Still, with Tabachnick and Fidell's text, one gets a good, readable introduction to the various statistical techniques. They even try to be humorous, using numerous examples of belly dancers and the like. I wanted to look up simple effect and found "Simple Minded," to which they said "See Statistics." When I looked up "Statistics," it said "Pages 1 - 860." As Tabachnick and Fidell are themselves psychologists, they do a good job at illustrating how psychologists would use multivariate statistics.
Unfortunately, the book is very light when it comes to the mathematics underlying the analyses. In fact, the only book that is significantly lighter is Meyers, Gamst, and Guarino's "Applied Multivariate Research: Design and Interpretation." That book is truly bad, with so little rigor and without the nice write ups of results or examples in numerous statistics packages that Tabachnick and Fidell present. If one wanted to get the mathematical rigor of the techniques without the examples of how to write up results for a psychological journal, I'd suggest Johnson and Wichern's "Applied Multivariate Statistical Analysis." If one wants a level even more rigorous, including formal proofs but almost complete lack of application, I'd recommend T. W. Anderson's "An Introduction to Multivariate Statistical Techniques." There are certainly other multivariate texts that come to mind (i.e., Stevens or Lattin, Carroll, and Green), but none of them so clearly fill niches as 1) Tabachnick and Fidell, 2) Johnson and Wichern, and 3) T. W. Anderson.
On one of my papers for Matthew's class, he suggested I consider graduate school in quantitative psychology. I didn't really understand what that was, and having wanted to be a clinical psychologist since I had my sometimes suicidal, sometimes hallucinating ex-girlfriend from high school, I couldn't conceive of being anything other than a clinical psychologist. I wish I had listened to Matthew, as I couldn't really consider wanting to be a clinical psychologist now. Currently, I feel like I have a good, working knowledge of statistics and quantitative methods, but I feel like I'm always teaching myself along the way, hopefully filling the lacunae in my knowledge but doubting whether I've done so successfully. In fact, a few years later, Matthew gave me his phone number and I called him to talk about leaving my Ph.D. program in clinical psychology to pursuit a quant Ph.D. Unfortunately, I wasn't accepted to the few programs to which I applied; therefore, I completed a second master's degree in statistics and then transferred to the social psychology program at Loyola.
Interestingly enough, in the year 2009, Matthew's major professor at U of I received an outstanding lifetime contribution award to the field from the American Psychological Association. When I interviewed for my assistant professor job at Cal State, Fullerton, where I later would work for two years, my interviewer, a social psychologist who faked knowing statistics asked me about my statistics training and I told him I took multivariate statistics with one of Larry Hubert's students. That individual didn't know who Larry Hubert was, despite the fact that Larry Hubert edited Psychometrika just a few years earlier, the premiere quantitative psychology journal. I bring this all up because I'm bitter for my experiences with this individual and with living in California in general. I should have taken this as a sign of things to come, but was a bit blinded about going out to CA in the first place.
I read the third edition of this book when Matthew was my teacher. Matthew had a Ph.D. in quantitative psychology from the University of Illinois, and knew more about statistics than most people I've met, even today. Unfortunately, he wasn't particularly disciplined, and would often get off topic. Still, with Tabachnick and Fidell's text, one gets a good, readable introduction to the various statistical techniques. They even try to be humorous, using numerous examples of belly dancers and the like. I wanted to look up simple effect and found "Simple Minded," to which they said "See Statistics." When I looked up "Statistics," it said "Pages 1 - 860." As Tabachnick and Fidell are themselves psychologists, they do a good job at illustrating how psychologists would use multivariate statistics.
Unfortunately, the book is very light when it comes to the mathematics underlying the analyses. In fact, the only book that is significantly lighter is Meyers, Gamst, and Guarino's "Applied Multivariate Research: Design and Interpretation." That book is truly bad, with so little rigor and without the nice write ups of results or examples in numerous statistics packages that Tabachnick and Fidell present. If one wanted to get the mathematical rigor of the techniques without the examples of how to write up results for a psychological journal, I'd suggest Johnson and Wichern's "Applied Multivariate Statistical Analysis." If one wants a level even more rigorous, including formal proofs but almost complete lack of application, I'd recommend T. W. Anderson's "An Introduction to Multivariate Statistical Techniques." There are certainly other multivariate texts that come to mind (i.e., Stevens or Lattin, Carroll, and Green), but none of them so clearly fill niches as 1) Tabachnick and Fidell, 2) Johnson and Wichern, and 3) T. W. Anderson.
On one of my papers for Matthew's class, he suggested I consider graduate school in quantitative psychology. I didn't really understand what that was, and having wanted to be a clinical psychologist since I had my sometimes suicidal, sometimes hallucinating ex-girlfriend from high school, I couldn't conceive of being anything other than a clinical psychologist. I wish I had listened to Matthew, as I couldn't really consider wanting to be a clinical psychologist now. Currently, I feel like I have a good, working knowledge of statistics and quantitative methods, but I feel like I'm always teaching myself along the way, hopefully filling the lacunae in my knowledge but doubting whether I've done so successfully. In fact, a few years later, Matthew gave me his phone number and I called him to talk about leaving my Ph.D. program in clinical psychology to pursuit a quant Ph.D. Unfortunately, I wasn't accepted to the few programs to which I applied; therefore, I completed a second master's degree in statistics and then transferred to the social psychology program at Loyola.
Interestingly enough, in the year 2009, Matthew's major professor at U of I received an outstanding lifetime contribution award to the field from the American Psychological Association. When I interviewed for my assistant professor job at Cal State, Fullerton, where I later would work for two years, my interviewer, a social psychologist who faked knowing statistics asked me about my statistics training and I told him I took multivariate statistics with one of Larry Hubert's students. That individual didn't know who Larry Hubert was, despite the fact that Larry Hubert edited Psychometrika just a few years earlier, the premiere quantitative psychology journal. I bring this all up because I'm bitter for my experiences with this individual and with living in California in general. I should have taken this as a sign of things to come, but was a bit blinded about going out to CA in the first place.
April 1, 2008
This is the best stats reference ever--both comprehensive and accessible. I've read various chapters of this book at different times, as needed. When I was navigating my dissertation without any help, as a stats-phobe, this book was my savior. It's because of this book that I am confident enough to coach/tutor other students as they navigate their own dissertations. And I refer to this book all the time in that context. So in the end, this text has made me back far more than the hundred bucks I spent on it. In fact, I've destroyed the pathetic not-designed-to-last textbook binding and I need to get it rebound before I lose an important section or something.
December 21, 2021
Great book for the multivariate statistics, it is useful for postgraduate students.
September 11, 2023
Would rather die
November 9, 2024
“Een Studieboek Is Ook Een Boek” Deel 1:
Hier heb ik een 7.5 voor gehaald! #slay
Niet tegen 3-jaar-geleden-ik zeggen dat ik dit vak vrijwillig heb gevolgd.
Hier heb ik een 7.5 voor gehaald! #slay
Niet tegen 3-jaar-geleden-ik zeggen dat ik dit vak vrijwillig heb gevolgd.
November 16, 2020
I have just spent a couple of hours to be able to have a quick browse through the 7th Edition of the book (published in 2019). As a practicing statistician I would like only giving this book a 3-stars rating for the following reasons.
On the positive side of the book, I learned something new to me about the wide range of topics of the multivariate statistics. I like the logic flow of the book and it is a pleasant reading experience about the technical details of the statistics somehow treated as a cookbook recipe. This, on the other hand, causes my great concern about what statistics, as a major quantitative analysis tool in disciplines such as psychology and many other social and natural science fields today, is all about and what statistics can or cannot do for us in our scientific research practice.
First and foremost, Null hypothesis significance test (NHST) does not equate statistics. Unfortunately, this book is essentially equating NHST to statistics. In fact, NHST, in my view, is logically indefensible, technically flawed, and practically damaging to scientific research. For an most authoritative assessment about NHST, please refer to American Statistical Association 2019's official statement here: https://www.tandfonline.com/doi/full/....
Essentially, the concept of 'statistical significance' is logically indefensible because, the testing statistic (be it a p-value, or a confidence interval, or a Bayes factor) used for determining statistical significance is a continuous variable and any attempt to dichotomize (or categorize) a continuous variable is logically indefensible. Technically, p-value is a conditional probability which assumes the null hypothesis is true. Therefore, a p-value cannot tell us the probability of the null hypothesis being true. Actually, this book is a collection of typical inappropriate applications of what originally proposed method of 'test of significance' by Sir R.A. Fisher or the method of 'hypothesis test' by G. Neyman & E. Pearson. For more information about how NHST came about as a result of the confusion from the early stage of the modern statistics development in the 20th century. For a brief introduction of why statisticians are against NHST, you may refer to this short article: "Why statisticians are abandoning statistical significance" https://onlinelibrary.wiley.com/doi/1....
In addition to the above mentioned "NHST ritual", a more fundamental concern about this book is the so-called "one ritual". When the authors introducing various statistical models/tests, they did not remind the readers the very fact "for any single set of sample data, the statistical data analysis results cannot tell if an observed 'treatment effect' on the dependent variable (DV) is actually due to the impact of the treatment or due to the random chance (sampling variation). As the famous statistician Yates told us in 1951" Research workers, therefore, have to accustom themselves to the fact that in many branches of research the really critical experiment is rare, and that it is frequently necessary to combine the results of numbers of experiments dealing with the same issue in order to form a satisfactory picture of the true situation. This is particularly true of agricultural field trials, where in general the effects of the treatments are found to vary with soil and meteorological conditions. In consequence it is absolutely essential to repeat the experiment at different places and in different years if results of any general validity or interest are to be obtained. In such circumstances a number of experiments of moderate accuracy are of far greater value than a single experiment of very high accuracy. "
It is a little bit relief to find that the authors did contribute roughly half of a page to the topic "Controversy Surrounding Significance Testing" (page 33).
A potential technical fault I picked out from the book is on page 753 referring to the model selection in time series analysis. The authors wrote "AIC and SBC criteria are only used to compare nested models." And the the authors also tried to mixed AIC with hypothesis test for model selection. To my best knowledge, it is wrong to say AIC only apply to nested models for model selection and AIC follows a fundamentally different approach from NHST. Interested readers are refer to these two books: (1) Burnham, K. and Anderson, D. (2002). Model Selection and Multimodel Inference: a practical information-theoretical approach, Springer; (2) Konishi, S. and Kitagawa, G. (2008). Information criteria and statistical modeling. Springer Science+Business Media, LLC.
Thank you to whoever finish reading this incomplete book review.
John
On the positive side of the book, I learned something new to me about the wide range of topics of the multivariate statistics. I like the logic flow of the book and it is a pleasant reading experience about the technical details of the statistics somehow treated as a cookbook recipe. This, on the other hand, causes my great concern about what statistics, as a major quantitative analysis tool in disciplines such as psychology and many other social and natural science fields today, is all about and what statistics can or cannot do for us in our scientific research practice.
First and foremost, Null hypothesis significance test (NHST) does not equate statistics. Unfortunately, this book is essentially equating NHST to statistics. In fact, NHST, in my view, is logically indefensible, technically flawed, and practically damaging to scientific research. For an most authoritative assessment about NHST, please refer to American Statistical Association 2019's official statement here: https://www.tandfonline.com/doi/full/....
Essentially, the concept of 'statistical significance' is logically indefensible because, the testing statistic (be it a p-value, or a confidence interval, or a Bayes factor) used for determining statistical significance is a continuous variable and any attempt to dichotomize (or categorize) a continuous variable is logically indefensible. Technically, p-value is a conditional probability which assumes the null hypothesis is true. Therefore, a p-value cannot tell us the probability of the null hypothesis being true. Actually, this book is a collection of typical inappropriate applications of what originally proposed method of 'test of significance' by Sir R.A. Fisher or the method of 'hypothesis test' by G. Neyman & E. Pearson. For more information about how NHST came about as a result of the confusion from the early stage of the modern statistics development in the 20th century. For a brief introduction of why statisticians are against NHST, you may refer to this short article: "Why statisticians are abandoning statistical significance" https://onlinelibrary.wiley.com/doi/1....
In addition to the above mentioned "NHST ritual", a more fundamental concern about this book is the so-called "one ritual". When the authors introducing various statistical models/tests, they did not remind the readers the very fact "for any single set of sample data, the statistical data analysis results cannot tell if an observed 'treatment effect' on the dependent variable (DV) is actually due to the impact of the treatment or due to the random chance (sampling variation). As the famous statistician Yates told us in 1951" Research workers, therefore, have to accustom themselves to the fact that in many branches of research the really critical experiment is rare, and that it is frequently necessary to combine the results of numbers of experiments dealing with the same issue in order to form a satisfactory picture of the true situation. This is particularly true of agricultural field trials, where in general the effects of the treatments are found to vary with soil and meteorological conditions. In consequence it is absolutely essential to repeat the experiment at different places and in different years if results of any general validity or interest are to be obtained. In such circumstances a number of experiments of moderate accuracy are of far greater value than a single experiment of very high accuracy. "
It is a little bit relief to find that the authors did contribute roughly half of a page to the topic "Controversy Surrounding Significance Testing" (page 33).
A potential technical fault I picked out from the book is on page 753 referring to the model selection in time series analysis. The authors wrote "AIC and SBC criteria are only used to compare nested models." And the the authors also tried to mixed AIC with hypothesis test for model selection. To my best knowledge, it is wrong to say AIC only apply to nested models for model selection and AIC follows a fundamentally different approach from NHST. Interested readers are refer to these two books: (1) Burnham, K. and Anderson, D. (2002). Model Selection and Multimodel Inference: a practical information-theoretical approach, Springer; (2) Konishi, S. and Kitagawa, G. (2008). Information criteria and statistical modeling. Springer Science+Business Media, LLC.
Thank you to whoever finish reading this incomplete book review.
John
May 19, 2020
If you want a career in statistics and or advanced analytics you MUST read this book from cover to cover as your introduction.
March 5, 2021
An excellent introduction and reference. If you are a researcher, KEEP THIS BOOK!!
March 3, 2018
This book has too much about the SPSS software if you're just looking for a book on Multivariate Statistics.
July 29, 2014
This is not a starters' book in stats. You need some grounding in methodology, or with simpler books, first. My stats professor refers to this as "the bible," and I see why, but this book is not for the faint of heart. I appreciate it most for the succint examples of write-ups for each type of analysis. I have referred to those time and time again in my doctoral program.
May 14, 2016
A very useful reference point.
There is so much to be found within these pages. Whatever it is you are searching for within the book comes with very useful information, giving clear explanations before going into more depth to develop your understanding. Whether you wish to read from cover to cover, or simply use as a reference for once or two things, this book will help.
There is so much to be found within these pages. Whatever it is you are searching for within the book comes with very useful information, giving clear explanations before going into more depth to develop your understanding. Whether you wish to read from cover to cover, or simply use as a reference for once or two things, this book will help.
April 15, 2008
This book gives a great overview of the different multivariate analyses that are available. The book goes very in-depth with matrices and equations, yet doesn't lose the reader, even when they are not mathematically gifted.
August 31, 2007
helping me throughout the years..!! :)
Read
May 22, 2010i thought this was useful
March 15, 2011
Very readable for a math book. I'm liberal arts student and this has been a breath of fresh air compared to most math/stat books I've been assigned in the past.
reference-only
March 20, 2021This book is written for students in the social sciences. The authors provide many insightful details that go beyond the theory.
January 21, 2017
Almost everything you wanna know about linear analysis modelling in our days.
October 12, 2018
The book of multivariate statistics.
Displaying 1 - 18 of 18 reviews