Help with basic definitions

#1
What is the basic difference between Kaplan-Meier and Cox Regression? How do I say it in as few words as possible. I am really limited by word count.

Furthermore if possible, could someone some up, again in as few words as possible the very basic usage/definition of the intraclass correlation, fishers exact test, Cronbachs alpha and SPSS?

Thank you very much.

Jono
 
#2
Would these paragraphs be correct? : The Chi Squared test is a nonparametric alternative to the t-test to be used when the variables of interest are dichotomous in nature. Furthermore the Chi squared test along with the Fisher’s exact test can be used as a nonparametric alternative to the correlation coefficient and is primarily utilised when the two variables of interest are categorical in nature. Fisher’s exact test is used instead of the Chi Squared test when expected frequencies in each cell being analyzed are less than 6 and when samples are small.


The Kaplan-Meier survival analysis like Cox proportional-hazards regression is used to compare survival curves in two or more group but unlike Coxs proportional-hazards regression model it is unable to analyze the effect of several risk factors on survival. It is used when analysing survival variables in cases where time is the most prominent variable.


I could just do with a simplified explanation/definition of the intraclass correlation and the cronbach's alpha. If anyone could help that'd be much appreciated.
 
#3
' and the Mann-Whitney-U test a nonparametric test used to compare ordinal data of independent groups.'

Cronbachs alpha is a mathematical, statistical procedure that determines the internal consistency reliability within a measure.

Also are these correct?

All I have left to define is the intraclass correlation, but different sources are telling me different things?

This is quoted from one of the essays i am critiquing: 'We also calculated the intraclass correlation to take account of the nonindependence of data from members of the same group.'

I have come across conflicting definitions one gives the definition of:

'A measure of agreement between observers that can be used when your observations are scaled on an interval or ratio scale of measurement.'

and the other gives the definition:

'The Intraclass correlation is used to analyze the strength and direction of resemblance between two variables.'

Do these mean the same thing? Are they both correct? In the example I have cited which definition applies?

This is all for my literature review for my psychiatric nursing degree.

I have to/am aiming to give brief descriptions of the purpose of the various tests used in the studies I am critiquing. I have paraphrased from different sources, but some sources seem to conflict, perhaps because some go into more depth than others.

Could really do with some advice ASAP.

If my questions are unclear suggestions on clarification would also be appreciated/any help will be/would be much appreciated.

Thank you.
 

Link

Ninja say what!?!
#4
The Chi Squared test is a nonparametric alternative to the t-test to be used when the variables of interest are dichotomous in nature.
No. The Chi Square test is an independence test used to determine if two variables are independent. The variables do not need to be dichotomous. Only discrete/categorical.

Furthermore the Chi squared test along with the Fisher’s exact test can be used as a nonparametric alternative to the correlation coefficient and is primarily utilised when the two variables of interest are categorical in nature. Fisher’s exact test is used instead of the Chi Squared test when expected frequencies in each cell being analyzed are less than 6 and when samples are small.
I'm not sure where you're getting this stuff. I've never heard of the Chi-square test + Fisher's exact being a non-parametric to the correlation coefficient. If you are referring to categorical variables, then these are thought of as correlation coefficients. (see here: http://www.statsoft.com/textbook/nonparametric-statistics/)

The Kaplan-Meier survival analysis like Cox proportional-hazards regression is used to compare survival curves in two or more group but unlike Coxs proportional-hazards regression model it is unable to analyze the effect of several risk factors on survival. It is used when analysing survival variables in cases where time is the most prominent variable.
There's a lot more to this than I think you would care to list. The statement's not accurate either. I would change it to this if you want to keep it brief:
"Kaplan-Meier survival analysis, like Cox proportional-hazards regression, is used to compare survival in two or more groups. Unlike Cox proportional-hazards regression model it cannot easily analyze several risk factors simultaneously."

A lot of researchers specializing in survival analysis do not like Cox-proportional hazards modeling. This is because it models hazards proportionally. (note: here, you can think of hazards as mortality rates, i.e. the rate at which deaths occur). This makes Cox models more parametric than Kaplan Meier curves. It also makes Cox modeling more complicated (i.e. easier for someone to mess up if they don't know what they're doing). If we were predicting survival, I'm pretty confident I can do better with a Kaplan Meier curve than you can with a Cox model. Just something to keep in mind.


I could just do with a simplified explanation/definition of the intraclass correlation and the cronbach's alpha. If anyone could help that'd be much appreciated.
A popular way of describing the ICC is the proportion of total variance that is "between groups" (when you are using random effects modelling). It gets more complicated when you use more complicated forms of random effects modelling (such as random slopes + random intercepts).

Cronbach's alpha, I think you can look up on Wiki or google.
 
#5
Thank you for your response/help and sorry it has taken me so long to thank you. I have now completed(ish) the statistics part of my dissertation.

Just to let you know I actually got the info that you queried/are querying from the very page that you put a link to under 'Brief Overview of Nonparametric Methods'.

Thank you for clarifying the issues I was having. I find understanding statistics quite hard especially when tests have multiple uses.