Sie sind auf Seite 1von 7

INFERENTIAL DATA ANALYSIS

SAS Inferential Data Analysis Assignment


Edwin J Ocasio
COH 602 Biostatistics
Dr. Alan Smith
National University
May 28, 2014

INFERENTIAL DATA ANALYSIS


SAS Inferential Data Analysis Assignment
For your draft paper you need to submit the following:
1.

Short description of your research question


The prevalence rates of childhood (under age 12) overweight and obesity in the U.S. have remained fairly

stable during the past decade, and sustained higher rates are observed in Hispanic and Black (specifically girls)
children in comparison to White children (primarily girls). The factors to most influence overweight and obesity
were race/ethnicity, poverty level, and physical activity.
The question is, what is the extent of the prevalence rates of adolescents (ages 12-17) overweight and
obesity in Hispanic and Black adolescents compared to their White adolescent counterparts?. The sample
population will include responders from San Diego Countys Central and South communities.
2.
a.

List of the SAS analyses you did to provide inferential statistics


Frequency distributions for RBMI, OVRWT, POVLL, RACEDO_P; and RBMI*RACEDO_P,

OVRWT*RACEDO_P, POVLL*RACEDO_P
b.

Univariate statistics, VAR RBMI BY RACEDO_P, VAR OVRWT BY RACEDO_P, and VAR POVLL,

RACEDO_P
c.

Student's t distribution TTEST for RBMI BY RACEDO_P and RACEDO_P BY OVRWT

d.

Data set filtered to display Sex = Female; and Race 1 (LATINO), 4 (NON-LATINO AFR. AMER.), and 5

(NON-LATINO WHITE).
3.

Discussion of the results of those analyses


This analysis will answer the question of what is the extent of the prevalence rates of adolescents (ages 12-

17) overweight and obesity in Hispanic and Black adolescents compared to their White adolescent counterparts. In
assessing the outcome, the analysis considered the possible confounding variables and any independent or dependent
relationship that may affect the results. The hypothesis for this analysis is that overweight and obesity are more
prevalent in Latino and Black females than in White females. Another factor that affects overweight and obesity is
the poverty level of the Latino and Black community.

This project the 2011-2012 California Health Interview Survey (CHIS 2011-2012) Adolescent Data File.
Using the SAS PROC FREQ function, the frequency distribution race-ethnicity (Latino, Black, and White), body fat

INFERENTIAL DATA ANALYSIS


index (BMI: 4 levels), obesity status, age (12 17) and poverty level (< 100%, 100 - 200%, 200 300%, and >
300% FPL) were processed. In addition, the SAS PROC FREQ function was used to compare the BMI by raceethnicity, obesity status by race-ethnicity, and poverty level by race-ethnicity distributions. The results showed a
disparate and tendency for higher BMI and obesity status especially in Latino females. Table 1 depicts the samples
age distribution depicted negligible differences from the 12 -17 years old. The remaining distributions provided the
basis for further analysis.
Univariate statistics using the SAS PROC UNIVARIATE procedure were computed for BMI by raceethnicity, obesity status by race-ethnicity, and poverty level by race-ethnicity. Each variable was run against the
race-ethnicity class to determine the mean, standard deviation, variance, Student's t test and corresponding p
Value. The results showed a significant difference in the BMI, obesity status and poverty level among Latino,
Black, and White females (Tables 6, Table 7.1, and Table 7.2).
In the final SAS analysis, all the variables were processed by the PROC CORR (Correlation Coefficients)
function which created a matrix of correlations. There were significant correlations shown in Table 5.1 and 5.2 of
the race-ethnicity, obesity status, BMI, and poverty level variables. The level displayed with the age and raceethnicity did not correlate with each other.
4.

Your code

PROC SQL;
CREATE TABLE Teen_Fem AS
SELECT OVRWT, Racedo_p,RBMI, POVLL, HGHTI_P, WGHTP_P, SRSEX, SRAGE_P
FROM sasuser.teen
Where SRSEX = 2;
QUIT;
PROC SQL;
CREATE TABLE Teen_FR15 AS
SELECT OVRWT, Racedo_p,RBMI, POVLL, HGHTI_P, WGHTP_P, SRSEX, SRAGE_P
FROM Teen_Fem
Where (Racedo_p = 1 or Racedo_p = 4 or Racedo_p = 5)
ORDER BY Racedo_p;
QUIT;
PROC SQL;
CREATE TABLE Teen_FR51 AS
SELECT OVRWT, Racedo_p,RBMI, POVLL, HGHTI_P, WGHTP_P, SRSEX, SRAGE_P
FROM Teen_Fem
Where (Racedo_p = 1 or Racedo_p = 5)
ORDER BY Racedo_p;
QUIT;
PROC FREQ DATA = Teen_FR15;
TABLES SRSEX SRAGE_P Racedo_p OVRWT RBMI POVLL ;
RUN;
PROC CORR DATA = Teen_FR15;

INFERENTIAL DATA ANALYSIS


var Racedo_p SRAGE_P OVRWT RBMI POVLL ;
RUN;
PROC FREQ DATA = Teen_FR15;
TABLES RBMI*RACEDO_P OVRWT*RACEDO_P POVLL*RACEDO_P / chisq;
RUN;
PROC UNIVARIATE DATA = Teen_FR15;
VAR RBMI;
BY RACEDO_P;
RUN;
PROC UNIVARIATE DATA = Teen_FR15;
VAR OVRWT;
BY RACEDO_P;
RUN;
PROC UNIVARIATE DATA = Teen_FR15;
VAR POVLL;
BY RACEDO_P;
RUN;
PROC UNIVARIATE DATA = Teen_FR15;
VAR SRAGE_P;
BY RACEDO_P;
RUN;
PROC TTEST DATA = Teen_FR51;
VAR RBMI;
CLASS RACEDO_P;
RUN;
PROC TTEST DATA = Teen_FR15;
VAR RACEDO_P;
CLASS OVRWT;
RUN;
5.

Your results
In the sample demographics (n = 1216; Sex = Females; Race-Ethnicity = Latino, Black, White) the

prevalence for overweight risk in Latino females is at a rate of 61.06% compared to 35.10% for White females. The
obesity status for both groups are similarly distributed as the overweight status at 62.24% for Latino females and
31.47% for White females. Uncharacteristic results are revealed in the distribution of underweight status of both
groups 85.83% (Latino) and 91.8% (White) this fact confirms that the disparity exists for the risk of overweight
and obese females only. Table 2 provides a comparison of the levels of BMI for each race-ethnicity group. The
disparity between Latino and White females is more apparent in Table 3 where the obesity (overweight) and risk for
overweight rates are shown side by side by race-ethnicity. Table 4 illustrates the poverty level for each group that
may also influence the tendency for a higher risk of overweight and obesity status among female Latinos. The
prevalence of overweight and obesity in Latino females is also dependent on race-ethnicity and poverty level.
Without any other confounding factors, it is clear that the prevalence of overweight and obesity among Latino
adolescent females parallel those of children, especially females compared to their White counterparts.

INFERENTIAL DATA ANALYSIS


Table 1
Frequency distribution of Age
Age
12
13
14
15
16
17

Frequency
194
202
193
214
206
207

Percent
15.95
16.61
15.87
17.60
16.94
17.02

Cumulative
Frequency
194
396
589
803
1009
1216

Cumulative
Percent
15.95
32.57
48.44
66.04
82.98
100.00

Table 2
Age and Gender Specific Body Mass Index (BMI 4 Levels) distribution by race-ethnicity
BMI/Weight Status
Underweight
Normal
At risk of overweight

#
32
833
208

%
2.63
68.5
17.11

Latino
71.88
46.70
61.06

Black
0.00
2.64
3.85

White
28.13
50.66
35.10

Not overweight (Cum)


Overweight

1073
143

88.24
11.76

85.83
62.24

79.92
6.29

91.8
31.47

Table 3
Risk of overweight and obesity by race-ethnicity
Race
Hispanic
African - American
White

#
628
39
549

%
51.64
3.21
45.15

OVRWT (%)
62.24
6.29
31.47

Table 4
Poverty level of Latino, Black, and White females
Poverty level
0-99% FPL
100-199% FPL
200-299% FPL
>300% FPL

#
259
287
155
515

%
21.3
23.6
12.75
42.35

Latino
%
85.71
71.08
51.61
23.69

Black
%
3.09
4.53
3.23
2.52

White
%
11.20
24.39
45.16
73.79

Risk of OVRWT (%)


61.06
3.85
35.10

INFERENTIAL DATA ANALYSIS


Table 5.1
Correlation of Race, Age, Obesity, BMI and Poverty Level
PearsonCorrelationCoefficients,N=1216
Prob>|r|underH0:Rho=0
RACEDO_P

SRAGE_P

OVRWT

RBMI

POVLL

RACEDO_P
RACEETHNICITY

1.00000

0.03899
0.1742

0.08410
0.0033

0.10644
0.0002

0.52230
<.0001

SRAGE_P
AGE

0.03899
0.1742

1.00000

0.06188
0.0310

0.06652
0.0204

0.07878
0.0060

OVRWT
OVERWEIGHTOROBESE

0.08410
0.0033

0.06188
0.0310

1.00000

0.81748
<.0001

0.10350
0.0003

0.10644
0.0002

0.06652
0.0204

0.81748
<.0001

1.00000

0.13931
<.0001

0.52230
<.0001

0.07878
0.0060

0.10350
0.0003

0.13931
<.0001

1.00000

RBMI
AGEANDGENDERSPECIFICBMI
POVLL
POVERTYLEVEL

Table 5.2
Simple Statistics or Correlation of Race, Age, Obesity, BMI and Poverty Level
SimpleStatistics
Variable

Mean

StdDev

Sum

Minimum

Maximum Label

RACEDO_P

1216

2.90214

1.97419

3529

1.00000

SRAGE_P

1216

14.54030

1.70307

17681

12.00000

OVRWT

1216

1.88240

0.32226

2289

1.00000

2.00000 OVERWEIGHTOROBESE

RBMI

1216

2.37993

0.72377

2894

1.00000

4.00000 AGEANDGENDER
SPECIFICBMI(4LVLS)

POVLL

1216

2.76151

1.20658

3358

1.00000

4.00000 POVERTYLEVEL

5.00000 RACEFORMERDOFSAS
(PUFRECODE)
17.00000 AGE(PUFRECODE)

Table 6
Chi-Square for BMI, Obesity and Poverty level by Race-Ethnicity
RBMI by RACEDO_P OVRWT by RACEDO_P POVLL by RACEDO_P
Statistic

DF

Value

Prob

DF

Value

Prob

DF

Value

Prob

Chi-Square

37.7704

<.0001 2

15.0516

0.0005 6

346.5305 <.0001

Likelihood Ratio Chi-Square

38.5628

<.0001 2

14.6477

0.0007 6

373.5074 <.0001

Mantel-Haenszel Chi-Square

13.7655

0.0002 1

8.5926

0.0034 1

331.4521 <.0001

INFERENTIAL DATA ANALYSIS

Table 7.1
Mean and Standard Deviation of BMI and Obesity by Race-Ethnicity
Variable

Mean

Std Dev

Sum Label

RBMI

1216

2.37993

0.72377

2894 AGE AND GENDER SPECIFIC BMI (4 LVLS)

OVRWT

1216

1.88240

0.32226

2289 OVERWEIGHT OR OBESE

Table 7.2
Mean, Standard Deviation, Skewness, Variance, Students t, and p Value of BMI and Obesity by each individual
Race-Ethnicity
RBMI (AGE AND GENDER SPECIFIC OVRWT (OVERWEIGHT OR OBESE)
BMI (4 LVLS))
Latino

Black

White

Latino

Black

White

628

39

549

628

39

549

Mean

2.4490

2.6667

2.2805

1.8583

1.7692

1.9180

Std Deviation

0.7779

0.8377

0.6319

0.3490

0.4268

0.2746

Skewness

0.8434

0.7130

1.6534

-2.0595

-1.3297

-3.0562

Variance

0.6051

0.7018

0.3993

0.1218

0.1822

0.0754

78.9000

19.8796

84.5636

133.4183

25.8857

163.6806

<.0001

<.0001

<.0001

<.0001

<.0001

<.0001

Student's t
p Value Pr>|t|

California Health Interview Survey, Public Use Data Files, 2011-2012 CHIS Data
http://healthpolicy.ucla.edu/chis/data/public-use-data-file/Documents/chis11-12_adolescent_sas.zip

Das könnte Ihnen auch gefallen