The stats book you wish you had as an undergrad senior?

[BeingThere]

I'd like some suggestions on a good stats book that will take me beyond simple correlation, chi square, and ANOVA.  I've finished the 2-course sequence of basic statistics required for my undergrad (which was combined with research methods).  I aced the courses, but I have two concerns about my current stats skills:  1) I still feel woefully unprepared for what I fear is going to be a tough first year in grad school stats, and 2) I want to use multiple regression for my senior research project and possibly for my honors project.  My professor is wonderful but he was unable to recommend a good next book.  No other advanced stats courses are offered at my school.  So, my question is, what is the best book out there that is the next logical step from intro-level stats?  Or at least a good book that covers multiple- and various other types of regression?  Thank you!

[DarwinAG]

http://www.amazon.com/Discovering-Statistics-Introducing-Statistical-Method/dp/1847879071/ref=pd_sim_b_1

 

Andy Field also has a stats book for R if you prefer that statistical packate.

This is a good basic stats book. quant_liz will be able to provide you with more books for more complicated statistics. Multiple regression is pretty basic as well. At least it is on par in terms of difficulty level with factorial and mixed-factorial ANOVA.

[BeingThere]

Thanks, DarwinAG.  While this book reviews what I've already learned, it covers quite a bit of material I've not learned yet.  And I am using SPSS, so this will be relevant in that regard.  And the price is nice too.  :-)  I really appreciate your help!

[boris]

I like this book a lot-- nice explanations, pretty basic overviews of concepts, very easy to approach.

http://www.amazon.com/SPSS-Survival-Manual-guide-analysis/dp/0335242391/ref=cm_cr_pr_sims_t

[boris]

I don't have this book but the reviews make it seem like an amazing introduction!

http://www.amazon.com/Understanding-Regression-Analysis-Introductory-Quantitative/dp/0803927584

[Soc Cog]

Another vote for the Andy Field book (Discovering Statistics Using SPSS) - this book is mostly an intro text but it really digs deep to explain what you are in fact looking at when using inferential statistics and how to ensure that you aren't going to report artifacts instead of meaningful differences!

 

 

Added: I think it is also critical to have at least one text that will help you with your analysis program of choice (SPSS, SAS, Stata, R, etc.) - perfect practice makes perfect

Edited December 24, 2012 by Soc Cog

[BeingThere]

Boris, I have the Survival Guide.  It's very handy!   I've gotten along fine with SPSS thanks to a great intro teacher and the Survival Guide, but I'm hoping to find something that gets more into what I guess you would call the math and theory (blood and guts??) behind the stats.

 

I'll be going into I/O psychololgy and from what I understand (and from the articles I've read) it's pretty stats heavy.  The more I can understand what is really going on statistically, the stronger foundation I think I will have. 

[esotericorigin]

it's pricy, but this book is a good next step in stats. he uses a lot of examples and it's not as bad to read as most math books.

 

http://www.amazon.com/gp/product/1111835489/ref=oh_details_o03_s00_i00

[GingerbreadLatte]

http://www.personality-project.org/r/book/

[Psychgrrrl]

The Andy Field book is pretty awesome. I'm starting to mess around with R, and considering buying that version of the book too (I already have it for SPSS).

[ohgoodness]

Moving beyond SPSS then we can use the STATA corp/Cleves book is pretty good: http://www.stata.com/bookstore/saus.html . You could also use the H-P Blossfeld http://www.stata.com/bookstore/event-history-analysis-stata/index.html which is pretty hard but still useful for going through all thedifferent models (cox, parametric, semi-parametric)

If you want to do multilevel regression then Stephen Jenkins is great: http://www.bristol.ac.uk/cmm/ especially the online course.

My feeling is that for practical work then the internet beat text books.

ps. I would also look into going to the quantitative methods summer schools - specifically the essex, michigan or slovenia. ds

ps2. you should really move beyond SPSS ds

Edited December 25, 2012 by cherub

[BeingThere]

That is interesting, Cherub.  Thanks for your suggestions.  SPSS is what I have available at school, so it is probably what I will be sticking with for now.  I think we have SAS and R as well, but I am not aware of any performance advantages of these over SPSS (other than R is free). 

 

However, I am intrigued that "statistics" is becoming synonymous with whatever software package one uses to perform analyses. 

 

I know very little, so forgive my ignorance, but I'd like some clarity on this.  How deep is the statistics education in other undergrad programs?  Are other schools teaching a lot of theory or just enough to be able to know which statistical test to apply and how to interpret the results?

 

Let me ask it this way:  does the Andy Field book go about as deep into the nuts and bolts (math- and theory-wise) as one needs to go?

 

I ordered the Andy Field book, by the way.   

[ohgoodness]

I know very little, so forgive my ignorance, but I'd like some clarity on this.  How deep is the statistics education in other undergrad programs?  Are other schools teaching a lot of theory or just enough to be able to know which statistical test to apply and how to interpret the results?

It really isnt but in social sciences where you do applied work then it usually becomes very tied to the packages you use. In demography STATA/R has become the way to go since it is very easy to deal with and has enough power to perform the main analyses, descriptive and regressions. Sadly I think that there is a wide divide between the theoretical part and the applied, my professors actually tell us not to take statistics since it is far too theoretical and not useful in social demography. It is enough if you know how to analyse output, control for various issues and know the major issues. I still think it would be very very useful to know more about the theoretical work but there is only so much time etc.

Side story - during my class in multilevel regression - my professor showed a summary of studies using a cox model and how many of them failed the major test of model fit.. Makes one question why people are knocking qualitative methods for being non-scientific.. (economists)!

forgot about this one - http://dss.princeton.edu/online_help/analysis/regression_intro.htm ds

Edited December 25, 2012 by cherub

[DarwinAG]

Level of depth is difficult. Do you mean to say does Andy Field derive the Central Limit Theorem for you and explain the foundational calculus of inferential statistics? Nope. But he does explain the the analysis and results enough more than the average undergraduate class. I think you will be happy with the level of difficulty.

The undergraduate psychological statistics courses at SFSU are pretty good, although I have no basis for comparison. The only experience that can speak of the quality is I attended a regional conference, and I noticed that many undergraduate researchers that I spoke to did not understand their results as well as I thought they should. This led me to conclude that my classes taught me a lot better than I thought. Of course, TAing statistics has helped me alot in terms of research interpretation and my advisor has allowed me to sit in his graduate statistics course that went in depth regarding ANOVA. He actually had the entire class calculate different types of ANOVAs by hand just so we can see how the variance is partitioned out.

Like you, I am interested in the "why" we conduct inferential statistics this way, but I do not think we will get anything satisfactory until later on in graduate school. Even then, I suspect we will be dissatisfied unless we take higher mathematics like calculus, which I have no intention of doing for funsies (just don't have the time for it).

[BeingThere]

Thanks Cherub and DarwinAG.  For practicality's sake, this eases my mind.  Time is definitely limited.  I'll be happy to learn as much as I can about stats theory in grad school.  Just want to make sure I'm not behind the curve going into it.  The Andy Field book will get here in time for me to have a week to digest the parts I need before the semester starts.  I think it will be a very useful guide for my project.

 

On a side note, DarwinAG, I attended SFSU (as a theatre major) a few years ago.  I'm from the Bay Area. Say hi to the ocean for me.   

 

Cherub, I don't know what a Cox model is.  Yet.

 

And Cherub, thank you for the links!

Edited December 26, 2012 by Bren2014

[Quant_Liz_Lemon]

Thanks Cherub and DarwinAG.  For practicality's sake, this eases my mind.  Time is definitely limited.  I'll be happy to learn as much as I can about stats theory in grad school.  Just want to make sure I'm not behind the curve going into it.  The Andy Field book will get here in time for me to have a week to digest the parts I need before the semester starts.  I think it will be a very useful guide for my project.   On a side note, DarwinAG, I attended SFSU (as a theatre major) a few years ago.  I'm from the Bay Area. Say hi to the ocean for me.      Cherub, I don't know what a Cox model is.  Yet.   And Cherub, thank you for the links!

Unless you take a biostats course or a course in survival analysis, you probably won't be exposed to a cox regression. I suspect that most grad psych stats classes will stick with parametric stats. At least that has been my experience anyway.

[lewin]

Cohen, Cohen, West, and Aiken is a classic but it's dense as crap. Almost anything you need to do, it's in there if you can figure it out.

http://www.amazon.com/Multiple-Regression-Correlation-Analysis-Behavioral/dp/0805822232

[TheFez]

We used this stats book in my program:

http://www.amazon.com/Applied-Linear-Statistical-Models-Michael/dp/007310874X/ref=sr_1_2?ie=UTF8&qid=1356667047&sr=8-2&keywords=linear+statistical+models

I thought it was a good text. Well written, good data sets used for problems throughout. The text uses SAS and SPSS, but any program will work fine. Don't get hung up on a particular software --- in my first year I have profs who used SAS, JMP, SPSS, STATA, R and MATLAB (uggh).

[ohgoodness]

I thought it was a good text. Well written, good data sets used for problems throughout. The text uses SAS and SPSS, but any program will work fine. Don't get hung up on a particular software --- in my first year I have profs who used SAS, JMP, SPSS, STATA, R and MATLAB (uggh).

Hey - how is JMP? When it looking at old syllabuses I always see it in there but then when I take the class - STATA/SAS it is. Recommended?

[Nausicaa]

http://www.amazon.com/Discovering-Statistics-Introducing-Statistical-Method/dp/1847879071/ref=pd_sim_b_1

 

Andy Field also has a stats book for R if you prefer that statistical packate.

This is a good basic stats book. quant_liz will be able to provide you with more books for more complicated statistics. Multiple regression is pretty basic as well. At least it is on par in terms of difficulty level with factorial and mixed-factorial ANOVA.

 

Thanks for this! I freaked out when I saw the price on amazon but luckily my uni library has a copy.

[sdt13]

Cohen, Cohen, West, and Aiken is a classic but it's dense as crap. Almost anything you need to do, it's in there if you can figure it out.

http://www.amazon.com/Multiple-Regression-Correlation-Analysis-Behavioral/dp/0805822232

This is the book used in the first year regression course at my school. It is dense, but really helpful if used properly along with a course.

 

I believe this one is very helpful as well, although it is no longer in print (I've pretty much been holding the library copy hostage lol):

 http://www.amazon.com/Essentials-Behavioral-Research-Methods-Analysis/dp/0073531960

Otherwise, I would suggest this, as it's a lot easier to read and understand (especially if you only have basic undergrad stats under your belt).

http://www.amazon.com/Research-Design-Statistical-Analysis-Edition/dp/0805864318/ref=pd_ys_iyr7

 

 

[3point14]

it's pricy, but this book is a good next step in stats. he uses a lot of examples and it's not as bad to read as most math books.

 

http://www.amazon.com/gp/product/1111835489/ref=oh_details_o03_s00_i00

Haha, this book was both my undergrad stats book, and now my first year grad stats book. Like you said, it's not nearly as dry or boring as math textbooks. I also like that he doesn't put his sole focus on SPSS (which I hate).

[BeingThere]

I bought the Andy Field book and have been pretty happy with it so far.    It's covering some of the same stuff I already know -- but some of it from a slightly different angle -- and it's filling in some small gaps.  And I know it will get to things I haven't learned yet.

 

The frustrating thing about trying to learn on one's own is not knowing really where one stands at any given time.  I'm asking around at my school to see if there's a statistics wonk that will advise me through a couple of credit hours of independent study in stats.

 

Sdt13, the third book you list looks like what I've learned already  It doesn't start looking foreign to me until Chapter 19 or so.  Looks like a good solid text, but ideally you'd have a suggestion for what you would choose after that one?

[DarwinAG]

The frustrating thing about trying to learn on one's own is not knowing really where one stands at any given time.  I'm asking around at my school to see if there's a statistics wonk that will advise me through a couple of credit hours of independent study in stats.

It can be frustrating learning on one's own. I just always work and teach myself on the assumption that there is waaaaay more to learn. Dissatisfaction is an incredibly strong motivator (if you can handle the negative affect associated with it) What really motivates me is identifying some expert in the field and telling myself, I want my knowledge to come close to that. Then if I do come close to that, I move on to another ideal state/place as a goal.

[sdt13]

Bren2014: Yeah, that book was recommended to me by my advisor the summer after I took basic statistics for psychology majors. After I took the basic classes I started studying more specific techniques, so I don't know about a general text that would extend beyond that (I've used this for an ANOVA class, but for me it wasn't really novel: http://www.amazon.com/Statistical-Methods-Psychology-David-Howell/dp/0495597848).

 

The cohen book is a nice text for regression, but otherwise I am starting to play with SEM, HLM, and moderated mediation (via Andrew Hayes' webpage).