Robin Let\'s See if We

¶ … Robin

Let's see if we can formulate a plan that will take care of the data you will receive form the questionnaire developed earlier. As you recall I recommended that you use a correlation and multiple regression analysis to statistically analyze the raw data. Before there can be a statistical analysis of the data you must first identify those variables that will be known as the "predictor" variables and the one variable that will be known as the "outcome" variable. In your situation the predictor variables can be age, birth order (change question to first born and later born), length of time smoking (change to under 5 years, over 5 years), father smoking, mother smoking, the number of cigarettes smoked per day, alcohol consumption (change the questionnaire question to a yes or no answer), household income (change the questionnaire question to under $49,000 and over $50,000). On the other hand the outcome variables are grade point average, college major, participation in sports activities, smoking as injurious to one's health, and do you plan on quitting smoking. OKAY now ... follow me closely here Robin. We can set up an independent correlation situation for each group of the female student (high school and college females) individual predictor variables with respect to a possible relationship with an outcome variable. For example we can ask: Is there a statistically significant relationship (correlation) between birth order and grade point average for high school females who smoke? Now watch

Number of High School Females = 50

Predictor Variable #1: Birth Order (1=first born, 2=later born)

Predictor Variable #2: Grade Point Average (1=C and below, 2= above C)

Outcome Variable: Females High school students who smoke.

Null Hypothesis: There is no statistically significant relationship or correlation at the .05 alpha level of female high school students who smoke on the basis of grade point average and birth order.

Raw Data: (X and Y each have a 1 or 2 numeric value: 1 for first born and 2 for later born; and 1 for C. grade and below and 2 for above C. grade). NOTE: I am only going to use a sample of 15 here but you should use at least 30.

X Variable (Birth Order)

Y Value (Grade Point Average)

1

1

1

2

1

2

1

2

1

2

1

2

2

1

1

2

2

1

2

1

1

1

1

2

1

1

2

2

2

2

Answer: R = -0.2887, p

Conclusion: On the basis of the correlation coefficient received we can conclude that there is no statistically significant relationship between birth order and grade point for female high school smokers at the .05 alpha level. Robin ... you must do the same analysis for all other predictor variable combinations and do the same for the college female smokers. Once all correlations are calculated we can submit the data to a multiple regression analysis. Let us suppose, for example, we are looking at four predictor variables fro the list I gave earlier and we have calculated the correlation values for all combinations with these four independent or predictor values and there will be 10 combinations. Let us say these correlation values are as follows: -.28, -.16, +.34, +.28, +.19, -32, -15, +22, +.32, -21.