Hypothesis testing - which method to use?

#1
Hi everyone,

I am taking a introductory Statistics course for my business program. I have been given a case to work with and I am completely confused about it. A little help would be great! I am not looking for exact answers per say, but just some guidance on which methods to use. Here is the case:

Nutrition education programs, which teach their clients how to lose weight or reduce cholesterol levels through better eating patterns, have been growing in popularity. The nurse in charge of one such program at a local hospital wanted to know whether the programs actually work. A random sample of 33 clients who attended a nutrition education program for those with elevated cholesterol levels was drawn. The study recorded the weight, cholesterol level, total dietary fat intake per average day, total dietary cholesterol intake per average day, and percent of daily calories from fat. These data were gathered before and three months after the program. The researchers also determined the gender, age, and height of the clients.

The data are as follows:
Gender (1=female; 2=male), Age, Height (in meters), Weight, before and after (in kilograms), Cholesterol level, before and after, Total dietary fat intake per average day, before and after (in grams), Dietary cholesterol intake per average day, before and after (in milligrams), Percent daily calories from fat, before and after.

Answer the following Questions:
1. In terms of each of weight, cholesterol level, fat intake, cholesterol intake, and calories from fat, is the program a success?

2. Does gender affect the amount of reduction in each of weight, cholesterol level, fat intake, cholesterol intake, and calories from fat?


Now, my questions are as follows:

- What type of descriptive measurement should I use for hypothesis testing?

- Is it independent samples (equal variances or unequal variances?) or matched pairs experiment (test statistic) under comparing 2 populations? OR is it central location (t- distribution test) under describing a SINGLE population?

- They are using 33 people as the population. Amongst those 33 people, they are asking to compare males and females. So, does that equate to a single population or double (2) populations? I mean technically, it's just one group of people. I think it might be matched pairs for 2 populations (find difference between two populations then use the difference in the test stat calculation) but i am very uncertain about it.

Some help would really be appreciated. This is kind of urgent. Thank you very much! :)
 
Last edited:

CB

Super Moderator
#4
I'd say that the first question could be addressed with a paired samples t-test. (paired because you're dealing with repeated measurements from the same people). The terminology you're using isn't what I'm used to but hopefully that makes sense to you.

The second question could be addressed by calculating the reduction in each of the dependent variables (i.e. difference between pre- and -post program for each person), and then completing an independent samples t-test for each dependent variable (i.e cholesterol etc). (Independent samples in that males and females can be seen as independent populations). This isn't an ideal solution though, so it depends what level of sophistication they're expecting in your answers. Repeated measures ANOVA would be better (calculating "difference scores" is problematic for a few reasons that I won't bang on about for the moment).