I need to learn statistics


Hello wonderful people of this forum!

I am fairly new to the forum and I am very glad I found it! It feels good to know there are people out going through the same difficulties.
My background is totally irrelevant, and through the master I only managed to get a very basic idea about statistics. I feel that I 've learn to do some basic stuff in Excel but I have no deeper understanding.
My university offers some basic courses. I plan to attend them but I understand that it takes a lot of personal effort too. I 've never hidden my weakness from my supervisors. They suggested that I should get some training on that too.
I want to start doing some preliminary studying by myself. Problem is I don't know where to begin from. During my PhD I will need to analyse amazing quantities of data, both numerical and responses from questionnaires. Every time I go through papers and I read something complicated I didn't even know it existed, I feel totally ignorant and inadequate!
Can anyone suggest where I start from? Any software worth learning? I would appreciate any information on online training or whatever you found useful yourself.

Thanks in advance!


======= Date Modified 30 Jun 2010 01:50:22 =======
Don't worry. I think stats is pretty overwhelming for most of us :) It definately takes a lot of effort, or did for me, but once you get the knack the sense of achievement is amazing.

Do you know what programs you'll be using? Or which ones the department has access to? If you're in the social sciences then I imagine you'll be using SPSS (Statistical Package for Social Sciences). There are a lot of online tutorials that are pretty good but you'll probably need to work through them step by step. SPSS can take some getting used to but it has a lot of features that can be very useful - it's just a matter of finding them.

There are a few books I've used that have helped. Two of them are psychology based so I don't know how much use they'll be. 'Research Methods and Statistics in Psychology - Hugh Coolican' and 'An Introduction to Statistics in Psychology' - Howitt and Cramer. There's a third that I've used for advanced stats and is more theorectical but I've left my copy in the lab and I can't remember it's name.

This link might be helpful too:


Andy Field's Statistics For SPSS is a great introduction, both in terms of teaching you the fundamentals of statitics and how to apply it to a very popular statistics package.


I think it depends on what level of expertise you want to achieve. Try Wikipedia if you want to become an expert at a later stage. There are also good links at the bottom of most statistical estimation theories etc..
At package level like SPSS(which I never used) I assume you will gain expertise at the user interface level, like how to enter the data and what estimate you are interested in etc..Again, I am not SPSS expert so my knowledge at that package level is limited..


I would second the recommendation for Andy Field's book. Also, Statistics Without Maths For Psychology by Dancey and Reidy is good. It is aimed at psychologists but does teach the basic concepts very well.


Thanks everybody for the replies! :-)

I am not in psychology, I am in a Science PhD. I ve read about SPSS in previous threads. Why is SPSS so popular? ( I am not a smartarse, honestly wondering). I found this list on wiki


but I kind of feel like someone trying to learn Autocad without knowing how to draw.

Wish I could tell more about my research but it would give my anonymity away! It could help you to give me more accurate advices.


I'd recommend that you email your to-be supervisor and seek their advice on relevant packages etc. Otherwise you might spend time learning something else, but end up not using it at all.


I'd definately speak to your supervisor. I had to learn basic statistics (albeit for the most part unsuccessfully - when my sup who was teaching the course banged his head repeatedly on the desk I realised that I wasn't quite cut out for it lmao) but I know at our uni the maths dept offer MA modules that you can do as a PhD student even if you are in a different dept - a friend has done that as his need was greater than mine - I didn't even go there - me and numbers just don't mix ;-) So see what is available - the postgrad administrator in your future dept will be able to help you possibly more than your sup in that. I also very much like statisitics for dummies (yes seriously) it was far more easy to understand that the official - learn statisitics books - and gave me a bit of a grounding to begin to understand the more complex stuff. As I say, I've never gone much beyond the basics, I don't need that level of understanding, but your first port of call is to find out what software you'd most likely be using and what your uni offers in terms of taught courses for PhD students.


Thanks one more time for your interest!
I will ask my super again, I am just trying to avoid the "banging-on-the-desk" part in our discussion:-) , and being a little prepared.
When am I going to be an independent researcher? Right now I feel like a 5-year old crying for my mummy!


Lol, I'm more like a two year old having a tantrum right now ;-)


Quote From DrJeckyll:

I am not in psychology, I am in a Science PhD. I ve read about SPSS in previous threads. Why is SPSS so popular? ( I am not a smartarse, honestly wondering). I found this list on wiki

I've seen it used in other areas as well, like biomedical sciences. So it might still be relevant.

Basically I think the reason it's so popular is that it can handle most quantatative data, and a range of statistical tests from the most basic of descriptive stats to mixed ANOVA's and regressions etc. The user interface is reasonably friendly as well. Say you want to run a correlation then you select the test menu, go down to correlation and select want you want there. You might need to play around to get exactly what you want but it's not so complicated that you have to sacrifice a goat every time you open it :) That said I'm sure the latest version will probably come with a chicken's blood add on to improve speed.

It normally plays well with excel too. So if you set out your data properly (rows for subjects/entries, columns for scores etc) it's just a matter of copying and pasting across rather than having to re-enter it all.

It can make plots for you to double check data and comes with lots of little extra's. Want to know the statistical power? Got a hankering for knowing the effect size? Want to eliminate all the extreme outliers before the analysis, or split it on sex? All doable with a few clicks.

Hmmm, what else :D Starting to feel like an SPSS salesman. It comes with AMOS as well usually which is used for factoral modelling (I think. I've yet to see anyone use AMOS). Also when you get more confident with SPSS you can bypass the user interface and use the syntax/code to save yourself hours.

The only problem I find with SPSS, and I know a few who feel the same, is that it hasn't improved much since the earlier versions. In fact the newer versions seem more unstable/awkward to me. I think it's on version 17/18 at the moment.

What stressed and bilbo say though. Check with your supervisor, you don't want to learn one package and then have to start again. Also, maybe check with the grad school for courses? There might be a few that are helpful and you can use the credits to go towards a post grade certificate or even diploma if you put enough effort in. I've got a PGC in research training, I wish I'd collected some more of my easier credits (you can get 10 for doing a presentation, 5 for a poster, 10 for going to a conference etc) and bumped it up to a diploma.


Thank you guys!