#employrate

[fantastatistic]

Assignment 6: Analysis of ANOVA

When performing an ANOVA, we use an categorical explanatory variable, while the response variavle is quantitative. I have previously grouped my employment variable (according to quartiles), and therefore I shall use this as my explanatory variable. And use Breast Cancer as my response variable. 

Employment and Breast Cancer

H(0) = There is no association between the Breast Cancer Incidence among the different Employment groups. 

H(alt) = There is an association between the Breast Cancer Incidence among the different Employment groups. 

The obtained results are seen below:

We see from here that the F-value is 4,69, and p = 0,0036, which means that we can reject H(0) and accept H(alt), which states that there is an association between the Breast Cancer Incidene among the different Employment groups. If I chose to accept the H(0), I would be wrong 0,36 times out of 100 times or 3,6 times out of 1000 times.

From the table we see that the mean for women from Employment group 4 is much smaller then the remaining groups, which may indicate in a way that countries with high employmentrate are subject to lower incidence of Breast Cancer. To be completely sure, we need to make a post hoc paired comparison Test. So I a Duncan test, and find out that I was right.

The mean in Employment group 4 (B) is significantly smaller then the three groups with the letters A. 

[fantastatistic]

Assignment 5: Scatterplot for Employment and morbidity-related variables

So managed to finish Week 3 today and do both quiz and of course submit the Assignments.

For Assignment 5 I find that my explanatory variable is employment and my response variables are ) Breast Cancer incidence, 2) HIV prevalence and 3) suicide rate. As my response variable is quantitative, I have made scatterplots:

In the Scatterplot with Employment rate on the X-axis and Breast Cancer incidebce on Y-axis, I am not sure, if I can say if there is a relationship yet. I didnt figure out, how to make a regression line, which would allow to decide, how strong the relationship is in the case of one. I had predictided that there would be a positive relationship because studies have showed that incidence of the disease is highest in industrial countries (except Japan). I make an assumption that employment rate is higher in industrial countries then non-industrial (dont know if I can say that :oI ) countries. I will need to wait till I am tought how to examine for relationship. 

This Scatterplot with Employment rate on X-axis and HIV Prevalence on the Y-axis is a bit difficult to conclude something for. I therefore decided to do a bar chart, where I group employment rate. I will use the previous groups I have used according to percentiles (Q1, median, Q3). With this I had following results.

Surprise! I expected that there would be a negative relationship, as I believe that a country with high unemployment would have citizens with more risky behaviour. So I expected that the number would steadily decrease with higher rate of employment. But we cant really conclude anything before we make further analysis and calculate a p value that lets us know if the results are results of coincidence :)!

Yeah, I am eager to continue with lectures so that I can make some further analysis and find out, weather there is an association.

Hope you all are doing fine...

[fantastatistic]

Assignment 1

Hi there,

Finally back and writing another post. The Gapminder includes a lot of data on variables related to development and MDG – just my cup of tea. But I of course had difficulties making a choice on what variables to work with. I have now decided to focus on employmentrate and morbidity (that is breatscancer, HIV and suicide). 

Employment is essential not only to maintain an income but also to have appropriate living conditions. Unemployment contributes to social health inequalities and is found to be an independent predictor of morbidity (Bartley M et.al., 2004) and mortality from suicide and external causes (Moser KA, 1984; Sorlie PD, 1990).

So I decided that my first research questions should be:

1. Is employment status associated with morbidity (breatscancer, HIV and suicide) 

My second research question is 

2. Is the Internetusage associated with morbidity (HIV and suicide)?

So I will now start working on my data analysis with SAS.