Soc 712 Homework #4

Paul Kim

Soc 712

Prof. Song

Research Question: Do drug users have a lower income controlling for sex, education and race?

Hypothesis: Drug users will have a lower income compared to those who do not do drugs.

Dataset: This dataset is from Brett’s data which he posted on data.world. This dataset covers drug use and demographic information

Exploring the Data

Importing the Dataset

This R chunk is simply using the readr package to read or import the data into R.

library(readr)
drugData <- read_csv("/Users/paulkim/Downloads/balexturner-drug-use-employment-work-absence-income-race-education/data/nsduh_workforce_adults.csv")

This R chunk is used to check if the data was imported correctly and to see how the data looks.

head(drugData)

This R chunk used the names function to look at names of the variables to get a better understanding of the all the variables in the dataset.

names(drugData)

 [1] "column_a"               "irpinc3"                "irfamin3"               "marij_ever"            
 [5] "marij_month"            "marij_year"             "cocaine_ever"           "cocaine_month"         
 [9] "cocaine_year"           "crack_ever"             "crack_month"            "crack_year"            
[13] "heroin_ever"            "heroin_month"           "heroin_year"            "hallucinogen_ever"     
[17] "hallucinogen_month"     "hallucinogen_year"      "inhalant_ever"          "inhalant_month"        
[21] "inhalant_year"          "meth_ever"              "meth_month"             "meth_year"             
[25] "painrelieve_ever"       "painrelieve_month"      "painrelieve_year"       "tranq_ever"            
[29] "tranq_month"            "tranq_year"             "stimulant_ever"         "stimulant_month"       
[33] "stimulant_year"         "sedative_ever"          "sedative_month"         "sedative_year"         
[37] "anydrugever"            "pharmamonth"            "illicitmonth"           "illicitmonth_nomj"     
[41] "anydrugmonth"           "anydrugyear"            "anydrugever_nomj"       "anydrugmonth_nomj"     
[45] "anydrugyear_nomj"       "countofdrugs_ever"      "countofdrugs_month"     "countofdrugs_year"     
[49] "personalincome"         "familyincome"           "questid2"               "employmentstatus"      
[53] "preemploymentdrugtest"  "randomdrugtest"         "everdrugtest"           "everdrugtest2"         
[57] "race_str"               "race_num"               "education"              "wouldworkfordrugtester"
[61] "selectiveleave"         "skipsick"               "sex"                    "druglist"              

Descriptive Analysis

library(ggplot2)

This R chunk is looking into the distribution of the income.

ggplot(drugData, aes(x = personalincome)) + geom_histogram()

The distribution of income is positively skewed which indicates that the majority of the people are in the lower income and that there is a small number in the higher level of income.

The Personalincome variable is categorized in this data set by: 1 = Less than $10,000 (Including Loss) 2 = $10,000 - $19,999 3 = $20,000 - $29,999 4 = $30,000 - $39,999 5 = $40,000 - $49,999 6 = $50,000 - $74,999 7 = $75,000 or more

This R chunk is looking at the distribution of Sex.

ggplot(drugData, aes(x = sex)) + geom_histogram()

This chart shows that there there are more women than there are men. Men = 1 and Women = 2.

This R chunk is looking into the distribution of education.

ggplot(drugData, aes(x = education)) + geom_histogram()

The education variable is coded by 1 = Less high school 2 = High school grad 3 = Some coll/Assoc Dg 4 = College graduate. The chart is showing that the majority of the people had some college or had their Associates degree.

This R chunk is looking at the amount of people who did any drugs within the last year.

ggplot(drugData, aes(x = anydrugyear)) + geom_bar()

The anydruyear variable is coded as true which means the person did any drug within the year and false means that that person did not do any drugs within the year.

Drug use and its affect on income

m1 <- lm(personalincome ~ factor(anydrugyear), data = drugData)
summary(m1)

Call:
lm(formula = personalincome ~ factor(anydrugyear), data = drugData)

Residuals:
    Min      1Q  Median      3Q     Max 
-2.6215 -1.6215 -0.6215  1.3785  4.1536 

Coefficients:
                        Estimate Std. Error t value Pr(>|t|)    
(Intercept)              3.62151    0.01319  274.65   <2e-16 ***
factor(anydrugyear)true -0.77509    0.02605  -29.76   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 2.035 on 32038 degrees of freedom
Multiple R-squared:  0.0269,    Adjusted R-squared:  0.02686 
F-statistic: 885.5 on 1 and 32038 DF,  p-value: < 2.2e-16

This is a binary variable which means that there are only two options. In this case are true and false regarding any drug use within the year. This analysis shows thats if a person did any drugs within in the year they have -0.77509 times less income than those who did not do drugs. The 3.62151 value is the value for the dependent variable or the reference group which in this case is no drug use within in the year.

Drug user’s racial difference on income

m2 <- lm(personalincome ~ anydrugyear + race_str, data = drugData)
summary(m2)

Call:
lm(formula = personalincome ~ anydrugyear + race_str, data = drugData)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.0739 -1.9440 -0.1978  1.8101  4.9110 

Coefficients:
                                       Estimate Std. Error t value Pr(>|t|)    
(Intercept)                             4.07388    0.05183  78.594  < 2e-16 ***
anydrugyeartrue                        -0.75405    0.02543 -29.656  < 2e-16 ***
race_strBlack/African American         -1.11843    0.06048 -18.491  < 2e-16 ***
race_strHawaiian/Pacific Islander      -1.23078    0.16354  -7.526 5.37e-14 ***
race_strHispanic                       -1.12989    0.05825 -19.396  < 2e-16 ***
race_strMixed                          -0.82808    0.07942 -10.427  < 2e-16 ***
race_strNative American/Alaskan Native -1.00959    0.10663  -9.468  < 2e-16 ***
race_strWhite                          -0.12206    0.05374  -2.271   0.0231 *  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 1.98 on 32032 degrees of freedom
Multiple R-squared:  0.07909,   Adjusted R-squared:  0.07889 
F-statistic:   393 on 7 and 32032 DF,  p-value: < 2.2e-16

Controlling for race, a person who did drugs within the year has -0.75405 less income than those who did not do drugs. When race was out of the analysis in the m1 analysis, people who did drugs had -0.77509 less income. But when race is put into the analysis, the affect of drug use on income decreased.

Interactions

Interaction between Drug Use and Race

m3 <- lm(personalincome ~ anydrugyear*race_str, data = drugData)
summary(m3)

Call:
lm(formula = personalincome ~ anydrugyear * race_str, data = drugData)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.0812 -1.8771 -0.1328  1.8672  4.7955 

Coefficients:
                                                       Estimate Std. Error t value Pr(>|t|)    
(Intercept)                                             4.08123    0.05559  73.414  < 2e-16 ***
anydrugyeartrue                                        -0.80850    0.15127  -5.345 9.11e-08 ***
race_strBlack/African American                         -1.12696    0.06654 -16.936  < 2e-16 ***
race_strHawaiian/Pacific Islander                      -1.33956    0.18907  -7.085 1.42e-12 ***
race_strHispanic                                       -1.20417    0.06346 -18.974  < 2e-16 ***
race_strMixed                                          -0.89327    0.09405  -9.498  < 2e-16 ***
race_strNative American/Alaskan Native                 -1.01117    0.12478  -8.104 5.53e-16 ***
race_strWhite                                          -0.10683    0.05800  -1.842   0.0655 .  
anydrugyeartrue:race_strBlack/African American          0.05874    0.16658   0.353   0.7244    
anydrugyeartrue:race_strHawaiian/Pacific Islander       0.43893    0.38297   1.146   0.2518    
anydrugyeartrue:race_strHispanic                        0.33129    0.16358   2.025   0.0429 *  
anydrugyeartrue:race_strMixed                           0.20835    0.19543   1.066   0.2864    
anydrugyeartrue:race_strNative American/Alaskan Native  0.03554    0.25250   0.141   0.8881    
anydrugyeartrue:race_strWhite                          -0.03314    0.15473  -0.214   0.8304    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 1.98 on 32026 degrees of freedom
Multiple R-squared:  0.07995,   Adjusted R-squared:  0.07958 
F-statistic: 214.1 on 13 and 32026 DF,  p-value: < 2.2e-16

The only statistically significant interaction in this analysis is for Hispanics who did drugs within the year. According to this analysis, if a Hispanic person did drugs within the year they will have 0.33129 more income compared to those who do not do drugs.

Interaction Between Drug Use and Sex

m4 <- lm(personalincome ~ anydrugyear*sex, data = drugData)
summary(m4)

Call:
lm(formula = personalincome ~ anydrugyear * sex, data = drugData)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.0387 -1.5916 -0.2499  1.7501  4.4084 

Coefficients:
                    Estimate Std. Error t value Pr(>|t|)    
(Intercept)          4.82743    0.04181 115.449  < 2e-16 ***
anydrugyeartrue     -1.28181    0.08036 -15.950  < 2e-16 ***
sex                 -0.78878    0.02600 -30.338  < 2e-16 ***
anydrugyeartrue:sex  0.31178    0.05138   6.068 1.31e-09 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 2.003 on 32036 degrees of freedom
Multiple R-squared:  0.05738,   Adjusted R-squared:  0.0573 
F-statistic: 650.1 on 3 and 32036 DF,  p-value: < 2.2e-16

This shows that when drug use and sex are interacted together they led to an increase in income. According to this data, when someone has done drugs in the year and is a female they will have a 0.31178 increase in their income.

Interaction Between Drug Use and Education

m5 <- lm(personalincome ~ anydrugyear*education, data = drugData)
summary(m5)

Call:
lm(formula = personalincome ~ anydrugyear * education, data = drugData)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.6748 -1.6748 -0.0705  1.3252  5.4709 

Coefficients:
                          Estimate Std. Error t value Pr(>|t|)    
(Intercept)                1.06811    0.03649  29.271  < 2e-16 ***
anydrugyeartrue           -0.30974    0.07220  -4.290 1.79e-05 ***
education                  0.90168    0.01217  74.085  < 2e-16 ***
anydrugyeartrue:education -0.13097    0.02490  -5.259 1.45e-07 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 1.85 on 32036 degrees of freedom
Multiple R-squared:  0.1962,    Adjusted R-squared:  0.1961 
F-statistic:  2606 on 3 and 32036 DF,  p-value: < 2.2e-16

This analysis shows that the interaction between drug use and education on income will lessen the negative effect on income. According to this data, someone who has done drugs within the year and is educated will have -0.13097 less income compared to someone who did not do drugs. If only the drug use variable is looked at then the affect on income is more compared to when drug use and edcuation are interacted together.

library(texreg)
screenreg(list(m1, m2, m3, m4, m5))

============================================================================================================================
                                                        Model 1       Model 2       Model 3       Model 4       Model 5     
----------------------------------------------------------------------------------------------------------------------------
(Intercept)                                                 3.62 ***      4.07 ***      4.08 ***      4.83 ***      1.07 ***
                                                           (0.01)        (0.05)        (0.06)        (0.04)        (0.04)   
factor(anydrugyear)true                                    -0.78 ***                                                        
                                                           (0.03)                                                           
anydrugyeartrue                                                          -0.75 ***     -0.81 ***     -1.28 ***     -0.31 ***
                                                                         (0.03)        (0.15)        (0.08)        (0.07)   
race_strBlack/African American                                           -1.12 ***     -1.13 ***                            
                                                                         (0.06)        (0.07)                               
race_strHawaiian/Pacific Islander                                        -1.23 ***     -1.34 ***                            
                                                                         (0.16)        (0.19)                               
race_strHispanic                                                         -1.13 ***     -1.20 ***                            
                                                                         (0.06)        (0.06)                               
race_strMixed                                                            -0.83 ***     -0.89 ***                            
                                                                         (0.08)        (0.09)                               
race_strNative American/Alaskan Native                                   -1.01 ***     -1.01 ***                            
                                                                         (0.11)        (0.12)                               
race_strWhite                                                            -0.12 *       -0.11                                
                                                                         (0.05)        (0.06)                               
anydrugyeartrue:race_strBlack/African American                                          0.06                                
                                                                                       (0.17)                               
anydrugyeartrue:race_strHawaiian/Pacific Islander                                       0.44                                
                                                                                       (0.38)                               
anydrugyeartrue:race_strHispanic                                                        0.33 *                              
                                                                                       (0.16)                               
anydrugyeartrue:race_strMixed                                                           0.21                                
                                                                                       (0.20)                               
anydrugyeartrue:race_strNative American/Alaskan Native                                  0.04                                
                                                                                       (0.25)                               
anydrugyeartrue:race_strWhite                                                          -0.03                                
                                                                                       (0.15)                               
sex                                                                                                  -0.79 ***              
                                                                                                     (0.03)                 
anydrugyeartrue:sex                                                                                   0.31 ***              
                                                                                                     (0.05)                 
education                                                                                                           0.90 ***
                                                                                                                   (0.01)   
anydrugyeartrue:education                                                                                          -0.13 ***
                                                                                                                   (0.02)   
----------------------------------------------------------------------------------------------------------------------------
R^2                                                         0.03          0.08          0.08          0.06          0.20    
Adj. R^2                                                    0.03          0.08          0.08          0.06          0.20    
Num. obs.                                               32040         32040         32040         32040         32040       
RMSE                                                        2.04          1.98          1.98          2.00          1.85    
============================================================================================================================
*** p < 0.001, ** p < 0.01, * p < 0.05

According to this analysis, part of the affect of drug use on income can be explained by education levels. When education and drug use are interacted they lessen the affect on income. Model 5 has the highest r-squared value and this shows that it is the most desireable and this indiciates the strength of the variables and fits the data better than the others.

Conclusion

The hypothesis for this study was that “Drug users will have a lower income compared to those who do not do drugs.” This is true when only the drug use variable was looked at. If someone had done drugs in the year they had a lower income compared to those who did not do drugs. But in the case of Hispanics and drug use, the data showed that they would have a 0.33129 increase income. Another finding was that when durg use and education were interacted that would lead to a lower affect on income. The findings for this study were interesting and did not match the proposed hypothesis.