# Hi everyone!

#### Alicia Wong

##### New Member
Hi,

My name's Alicia, and I'm a psych student who's just finished college. I'm planning to get into an MSc program in clin psych, but also have an itch towards statistics.

I've to admit that I'm only a beginner in stats. Right now, I'm attempting to learn maximum likelihood estimation (because my college supervisor wants me to look over my senior thesis again), but top searches on that famous search engine gets me things that make me want to headdesk, and a digging at a local bookshop didn't pay off (although I did find a SAGE paper on interaction effects in logistic linear regression, I think. It was interesting).

Anyway, HI! :wave:

#### trinker

##### ggplot2orBust
Welcome aboard Alicia Wong!

You may find this article on MLE to be a helpful guide. It may be helpful to look at MLE in context of statistical tests/analysis that rely on MLE: Search Generalized Linear Models (many use MLE) and within GLM search logistic regression, Poisson and negative binomial.

Last edited:

#### Link

##### Ninja say what!?!
Something else that might help as well is to try and learn the difference between method of moments estimation and MLE. It'll give you a better understanding of what MLE does.

#### Alicia Wong

##### New Member
@Trinker: thanks for the file ^^ it's helping a lot (still in the midst of reading it, though). And just Alicia would do Alicia Wong sounds so formal!

@Link: I think I read about this somewhere - it's concerning the mean, variance, standard deviation and then estimating from there right?

#### spunky

##### Doesn't actually exist
darn! you missed me by a month! if you look at this thread you'll see i did a quick, one-week workshop on intro to maximum likelihood estimation for psychology/education/social science students... anyways, the book i suggested was Eliason, S. (1993). Maximum Likelihood Estimation: Logic and Practice by Sage University Press and, although i had to work on the material a little bit to make it more approachable and social-science-students friendly, i found it to be an invaluable source...

#### Alicia Wong

##### New Member
Oh man! I'm going to read that now - hopefully don't get confused halfway. I'm not familiar with the U.S. college system, so I was wondering, isn't statistics a required class to do psychology/social sciences?

Oh yes, I finished that short tute on MLE provided by @Trinker (thanks again!) and came across this as well, which said just about the same thing.

#### spunky

##### Doesn't actually exist
so I was wondering, isn't statistics a required class to do psychology/social sciences?
well, yes but it is taught in a very applied context. most of the statistics courses for psychology or education or social sciences in general only provides a brief, simple overview of the statistical theory and focuses more on the applications of the theory through examples and a lot of analysis using computer programs (SPSS mostly...) as i'm sure you've found already, to fully understand maximum likelihood estimation one needs, at the very least, some background knowledge on calculus, partial derivatives, probability density functions, basic optimisation etc. which is something most (although not all) students in the social/behavioural sciences know nothing about... however, ML are used all the time to estimate parameters of complicated models so that's why a group of students got together and asked me to try and give them a quick intro about how this thing work so they could at least look at the computer output and say " oh! i think i more or less understand how we got those numbers..."

i think it is very good for you to start looking into this before you begin your MSc, you'll certainly be ahead of a lot of people when you take your advanced methodology courses...

#### Alicia Wong

##### New Member
Oh! My classes also did use SPSS, but only in the beginning. Once we got the hang of where the buttons were, we abandoned the lab lessons and devoted time to understanding/calculating by hand chi-square, ANOVA, MANOVA and basics of linear regression, so in some ways we were more theoretical than applied.