http://www.smartgirl.org/reports/7902020.html

Introduction What effect does religion have in the life of a young person? Does it lessen pain or worsen it? More and more middle-school, high school, and college-aged youth are committing suicide,...

This is my final report for the statistics course I am taking on Coursera. Peer review starts this afternoon (5 pm EST). Prayers please! =)

   Introduction

What effect does religion have in the life of a young person? Does it lessen pain or worsen it?

More and more middle-school, high school, and college-aged youth are committing suicide, often spurred by negative (or negligent) influences. Sometimes these same influences lead a young person to engage in sexual activity. Could ideas about sexual behavior be a matter of life or death?

Religious young people may have the tools to handle life circumstances without suicide. But Sexual behavior and suicide could be inversely related among Christian young people. That is, the more one accepts religious principles and the stronger he feels about his ability to handle sex, the more likely he is to end his life. But why?

Research Questions

(1) Is suicide associated with one's total belief in the Sacred Scriptures? (2) Does a young person's perceptions of sex impact that association?

Methods

Sample

Protestant and Catholic adolescents (n=6022) were taken from the public-access version of the first wave of the National Longitudinal Survey of Adolescent Health.

This survey is a nationally representative study of adolescents (grades seven to twelve) in the United States. This data was collected in the 1994-95 school year.

Measures

The question "During the past 12 months, did you actually attempt suicide?" is answered with the number of attempts.

The questions regarding suicide were self-administered, using audio CASI (computer-assisted self-interviewing) to reduce external interference or influence.

Questions concerning health-protective measures were asked only to those aged fifteen and older.

Results

Univariate

Among Christian adolescents, profession of Born-Again Christianity was split around 50-50 (51.81% said no; 48.19% said yes).

Nearly 80% of the sample professed total belief in the sacred Scriptures.

A little over 16% of Christian adolescents attempted suicide at least once in the past year; 72% never attempted suicide.

Nearly 51% of young people perceived themselves as capable of stopping to take birth control measures once highly aroused.

Bivariate

Looking at the relationship between the number of suicide attempts in the past year (quantitative response variable) and profession of total belief in the sacred Scriptures (categorical explanatory variable), an Analysis of Variance (ANOVA) revealed more suicidal attempts in the past year among those who professed full faith in the sacred writings (Mean=0.55, s.d. ±1.22) versus those who did not profess to believe the Scriptures fully (Mean=0.46, s.d. ±1.12). But the probability value is not statistically significant. F(1, 577)=0.67, p=0.4150. There is no solid relationship between the two variables, at least in this sample.

The association between number of suicide attempts and the perceived ability to stop for birth control measures once aroused has no statistical significance: F(4, 535)=0.27, p=0.8965.

Furthermore, the relationship between number of suicide attempts and professing to be a Born-Again Christian is also statistically insignificant as F(1, 408)=0.85, p=0.3584.

Moderation

Neither profession of Born-Again Christianity nor perceived ability to stop once aroused moderated or impacted the relationship between faith in the sacred Scriptures and number of attempted suicides in the past year (each had probability values greater than 0.05).

Instead, those who thought the most highly of their ability to stop sexual activity once aroused moderated the rate of suicide attempts among those with total belief in the sacred Scriptures:  p=0.0317. My p value (probability value) is less than 0.05, indicating statistical significance.

Those who both believed highly in their ability to stop in a state of arousal and had total belief in the sacred Scriptures were more likely to attempt suicide (Mean=0.7326) than those with the same level of confidence and no belief in the sacred Scriptures (Mean=0.3033).

ANOVA results for "number of suicide attempts" by "faith in sacred Scriptures" moderated by "stopping sexual activity once aroused" (with Duncan post hoc test results).

Visual representation of the above results

Discussion

Statistically, these results reveal no relationship between attempted suicide and two of faith's most fundamental principles  - profession and acceptance (Romans 10:8-9).

There appears to be some association between faith and suicide among young people who are confident of their ability to control themselves while in an aroused state; and those youth make up 50% of my sample.

This research is based on a nationally representative study of adolescents who are either Protestant or Catholic and who have attempted suicide.

The AddHealth survey (Wave I) was conducted in the 1994-95 school year.  The dynamics and the responses may be different now, over ten years later.

 The sample of young people who both are Christians and have attempted suicide is minimal (n=757): a larger sample would provide for better research.

Also, measuring the depth of religious experience would provide more sound data, and answer some questions about why Christian young people are choosing suicide.

My most prominent research question is this: Is there any statistically significant data which answers the cause (or at least the observational cause) for such a high rate of suicide among Bible-believing Christian young people?

SAS Program

Moderated by "Born-Again Christianity"

__________________

Moderated by "Perceived Ability to Stop Sexual Activity in an Aroused State"

These tests (Chi square and Pearson correlation, respectively) were for Assignments 7 and 8. The Chi square test of independence analyzes whether two variables are dependent or not. The Pearson correlation test looks for the strength (strong or weak) and direction (positive or negative) of a linear relationship between two variables. Chi square deals with categorical variables; Pearson deals with quantitative. (ANOVA handles a categorical explanatory variable and a quantitative response variable.)

Chi Square Test of Independence

This test was relatively simple for me because my research deals chiefly with categorical variables. Testing an explanatory variable that has more than two "buckets" requires a post hoc test (to see where the differences are); so I decided to begin with a simple two category variable. 

For this test, I explored the relationship between being a professed Born-Again Christian and happiness.

Below is the table produced by SAS. I labeled the table to see clearly what I was reading. As you can see, dummy code 1 for HAPPYWEEK is "was not happy" and dummy code 2 is "was happy". For H1RE5, dummy code 0 is "I do not consider myself a Born-Again" and dummy code 1 is "I do consider myself a Born-Again Christian".

For each category in the response variable, there are three parts of data: the percent, the row percent, and the column percent. For example:

11% of those interviewed did not profess to be Born-Again Christians. (Percent)

57% of those who were not happy this past week also did not profess to be Born-Again Christians. (Row Percent)

22% of those who did not profess to be Born-Again Christians were not happy in the past week. (Column Percent) 

In this test, the most important data is found in the "Column Percent" row. I want to compare the happiness rates of Born-Again Christians versus non-Born-Again Christians.

The value circled in red in the table below represents probability - the probability of me being wrong should I reject the null hypothesis.

Speaking of which...

Ho: Being a Born-Again Christian and being happy most of the time are not related. (They are independent.)

Ha: Being a Born-Again Christian and being happy most of the time are related. (They are dependent.)

My p value (probability value) must be less than or equal to 0.05 (or 5%) in order to be considered statistically significant. My p value is .001 or (0.1%). If I reject the null hypothesis and say that these two variables are dependent on each other, I will be wrong less than one percent of the time.

In conclusion:

When examining the relationship between being a Born-Again Christian (categorical response variable) and general happiness during the week (categorical explanatory variable), a chi-square test of independence exposed that among adolescent Protestants and Catholics (my sample), those who professed to be Born-Again Christians were more likely to have a happy week (81.5%) than those without such a profession (77.11%), X2 =10.83, 1 df, p=0.0010.

And here is a table for your visual benefit:

Pearson Correlation

As I mentioned above, this test looks for linear relationships: how close are the different opinions to the general line of thinking? Is that line of thinking positive or negative?

Pearson tells you this by a number, from -1 to +1. A negative 0.75 is a strongly negative or inverse trend (as one goes up, the other goes down); while a positive 0.75 is equally strong, but in a positive direction (direct variation, or as one goes up, so does the other).

My table below features the following statistics:

A correlation value of 0.01141 (r=0.011)

A probability value of 0.7755 (p= n.s. or not significant)

Sorry guys, but there is not much to this one. The r value is very very small, meaning that the positive linear relationship that does exist is negligible. And the p value states that if I reject the null hypothesis and say that there is a correlation between the two, I will be wrong more than 75% of the time!

In conclusion:

Among adolescent Protestant and Catholics (my sample), the correlation between frequency of prayer per month (quantitative explanatory variable) and number of suicide attempts in the last twelve months (quantitative response variable) was 0.011 (p=0.7755, or n.s.), suggesting that the varying number of suicide attempts among Christian young people cannot be explained by the strength of their prayer life.

I think I have it!

For while now, I have been trying to refine my research question to something more manageable. But I believe that Inspiration has given me an answer.

My first question: "What is the relationship between one's happiness and being a Born-Again Christian?"

Born-Again Christian is a categorical variable - it's either a "yes" or a "no". Each person's response can be placed in a bin. 

Happiness is also categorical. Is my response variable - it is the effect caused by being a Born-Again Christian or not. I put it on my y-axis to see if the height changes between the two groups. But in order to do this effectively, my response variable needs to have only two categories. So I went from this table...

...to this one...

...using this code:

If variable H1FS11 is less then 1, then variable HAPPYWEEK should equal 1 (Responses 0 and 1 were never and sometimes happy in H1FS11.)

Else if H1FS11 does not equal missing data, then HAPPYWEEK should equal 2 (Responses 6 and 8 were coded out as missing, leaving responses 2 and 3, happy most of the time or all of the time that week.)

By the way, a period always represents missing data in SAS. I used it to code out unwanted responses (such as "refused" to answer) by telling SAS the following:

If VARIABLE = # then SAMEVARIABLE = . ; 

Anyway, here is my graph answering question 1...

bi-variate: consisting of data from two variables (in contrast with univariate which has data from only one variable)

Did I mention that I love graphing? It is my absolute favorite algebraic topic. And I don't even like math.

I am going to make some graphs to focus on some questions that I have, such as the relationship between (1) happiness and being a Born-Again Christian; (2) the strength of prayer life and the ability to resist intercourse in the event that one's partner resisted birth control; and (3) level of youth involvement and the ability to stop to use birth control once highly aroused.

This week we are focusing on graphing! =) Now I get to see my data in action.

I am a little nervous though: Do I have too many variables? If I do, oh well. I will refine as I go along.