Biostatistics help?

Question

Is there anyone out there who knows anything about biostatistics. I have a homework assignment due Monday & I have no clue how to solve one of the problems. Here's what the paper says,

An animal physiologist studied the pituitary function of hens put through a standard forced molt regimen by using egg producers to bring the hen back into egg production. 25 hens were used for the study. 5 hens were used for the measurements at the premolt stage prior to the forced molt regimen and at the end of each of 4 stages of the forced molt regimen. The 5 stages of the regimen were (1) premolt (control), (2) fasting for 8 days, (3) 60 grams of bran per day for 10 days, (4) 80 grams of bran per day for 10 days, and (5) laying mash for 42 days. The objective was to follow various physiological responses associated with the pituitary of the hen during the regimen to aid in explaining why the hens will come back into production after forced molt. one of the compounds measure was serum T3 concentration. The data in the table are the serum T3 measurements for each of the five hens scarified at the end of each stage of the regimen.

Treatment Serum T3, (ng/dl) x 10-1

Premolt 94.09; 90.45; 99.38; 73.56; 74.39

Fasting 98.81; 103.55; 115.23; 129.06; 117.61

60 g bran 197.18; 207.31; 177.50; 226.05; 222.74

80 g bran 102.93; 117.51; 119.92; 112.01; 101.10

Mash 83 .14; 89.59; 87.76; 96.43; 82.94

A) Write a linear statistical model for this study and explain the model components.

B) State the assumptions necessary for an analysis of variance of the data.

C) Compute the analysis of variance for the data and test the hypothesis of no difference among means of the five treatments at 0.05 level of significance.

D) Compute the 95% confidence interval estimates of the treatment means.

E) Determine how many chickens the biologist would need for each treatment to reject the null hypothesis at the 0.05 level of significance with a power of 0.90 if the difference between the control treatment and any new treatment was 30 units of T3 concentration.

I need someone to help point me in the right direction on getting started with this problem. Here's my issues,

I have no idea what part A is asking
Is part C asking for a single factor analysis of variance (one-way ANOVA) or am I supposed to use a different test (i.e. random effects analysis of variance, Bartlett's, etc)?
What is part E talking about?
Please note, I am looking for help on this question. By no means do I want someone to solve it for me, just point me in the right direction for solving it. Also, although I appreciate the offer, I do not have the money to pay a tutor so please don't offer to tutor me unless you can do it for free.

EDIT: By the way, I do have the SAS program to help with creating ANOVA Tables.

Answer 1

You have SAS. That's rad :)

Okay, I'm going to take this question over a span of time, so periodically check back for updates.

Part A wants you to fit a linear model to the data. that means rule out logistic regressions and dichotomous variable correlations. I'm thinking PROC GLM would work better than just doing a PROC REG linear regression because you're running multiple sets of matched data (I read that right, right? It's all matched to each of the chickens in each stage of the trial?).

Part C *corrected* Previously, I said two-way ANOVA because I was thinking "two-tailed" (silly me). But you only do two-WAY if you've got multiple independent variables against each other. Your independent variable is treatment regimen, but there's nothing else being measured (i.e., you didn't divide your chickens by color or something), so a ONE-WAY ANOVA should be sufficient.

Part E is asking you to determine the n necessary to generate sufficient statistical power to get a p-value<0.05 with beta set to 90%. There is an actual formula for this, but there is also a quickie program called CPower that generate the n based on your inputs for alpha, beta and whatever else (haven't used CPower nor made this calculation in a while since I don't do clinical trials).

//ADDENDUM//
About study modeling, that means you're trying to relate the variables to each other. In the case of part A, you are trying to do some kind of linear regression to find y=mx+b, R-squared. that is, what's the slope of the line of best fit, and how well does it fit your data points? You can fit models using correlations, but I'd not recommend it. It's better just for finding a R-squared value.

Answer 2

How do you do a bar graph for someone if there salary is 70,000 a year. How do you find the mean or mode or median

Answer 3

i had no idea what you just wrote back ther but i would wanna help but i can't srry