CHAPTER 3 METHODOLOGY
Data and Sample
To acquire data to examine the effects of age, gender, or race on sentencing, primary data collection was conducted of district court records of two western North Carolina counties: Ashe and Watauga. The target population of the data to be collected was all disposed district court cases of driving while impaired offenses within the years 2011, 2012, 2013, and 2014. A list of all disposed cases was collected from each county via civil revocation files. Probability sampling design was implemented to ensure a random sample. A simple random sample of these cases was taken by assigning each case file number a number and choosing one hundred cases from each county to be included in the sample through the assistance of a random number generator. All cases within the target population had an equal chance of being chosen for the study.
Once the sample was selected for each county, data was collected from the case files chosen. This data included demographic information, specifically: age, gender, and race. Type of legal counsel, disposition, and sentence were collected as well. Identifying information such as offender name, address, specific date of birth, social security number, and driver’s license number were omitted from collection to insure confidentiality of the subjects included in the data and to reduce any bias that may occur based on the identify of the offender. The only identifying information collected from the case files was the county-assigned case number.
Measurement
The study included five dependent variables, each relating to the sentencing of the offender. The first dependent variable in the study was disposition type, which was gathered from the case files and measured as follows. Disposition had two attributes: guilty or other. The attribute of “other” included not guilty dispositions as well as cases that were dismissed by the court. These attributes were coded as 0 for other and 1 for guilty. The remaining dependent variables were different types of sentences that the offender could receive as a result of being found guilty of their offense, and were each coded 0 for no and 1 for yes. These variables were fines, unsupervised probation, supervised probation, and active time in a correctional facility.
The independent variables in the study include age, gender, and race of the offender. The type of legal counsel was also included, along with county, as a control measure and was measured using the data collected from the case files. Age was measured on a ratio level using the age of the offender at the time of disposition and sentencing. Gender was simply determined as either male or female. Race was measured using the information from the case files as white or other, which includes any other race reported in the file. Type of counsel or attorney that represented the offender in the case was coded as 0 for no attorney or for a court-appointed attorney and 1 for a retained attorney. County was measured by including which county the case was from: Watauga or Ashe.
Analytic Strategy
Statistical analysis was used to test the following hypotheses:
H1. Offenders of a younger age are more likely to be found guilty of their offense and receive fines, probation, or active time than offenders of older age.
H2. Male offenders are more likely to be found guilty and receive fines, probation, or active time than female offenders.
H3. Nonwhite offenders are more likely to be found guilty and receive fines, probation, or active time than white offenders.
In order to get the full picture of the independent variables effects on sentencing, several types of analyses were used. First, univariate analysis or descriptive statistics examining each of the variables was conducted. Next, cross-tabulations were used to describe the relationship between race and gender on sentencing decisions, separately. Then, independent sample t-tests were used to determine the average age of defendants for each dependent variable. Next, analysis of variance (ANOVA) was used to examine interactions between race and gender on sentencing decisions. Finally, binary logistic regression modeling was used for analysis in order to determine the effects of race, gender, and age on sentencing while controlling for other important factors. Because the dependent variables were measured at a binary or dichotomous level, binary logistic regression was appropriate (King, 2008). The models were estimated by the maximum likelihood estimate. A model was prepared for each dependent variable in relation to the independent variables.
CHAPTER 4 RESULTS
Data was collected from the two hundred selected cases and was analyzed both by the entire sample and by each county, one hundred cases being from Ashe County and one hundred from Watauga County. For the entire sample, 81 percent of the population was male, 84 percent of the population was white and 16 percent were nonwhite. As shown in Table 1, a vast majority of the defendants (79%) were found guilty. Fines were imposed in 78 percent of the cases. Over half (54%) of the guilty defendants were sentenced to unsupervised probation, while approximately 23 percent of respondents were sentenced to supervised probation, and another 23 percent to active time in custody of the corrections department. The average age of all two hundred defendants in the sample was 33 years. Regarding counsel, a majority of the sample (61.5%) had retained attorneys, while 39 percent had court appointed attorneys, had no attorney, or waived their right to counsel.
While demographics were similar for Ashe and Watauga counties as shown in Table 1, there were some differences. 86 percent of the defendants in Ashe County were male compared to 76 percent of the defendants in Watauga County. The average age of the defendants in Ashe County was 36 years, which was somewhat higher than the average of 33 years in Watauga County. The sample taken from Ashe County was found to be more racially diverse than that taken from Watauga County. While for both counties a large majority of the defendants were white, slightly less were white in Ashe County (78%) than Watauga County (89%). In Ashe County, 22 percent of the defendants were of nonwhite race, compared to only 11 percent of the Watauga County sample.
Table 1.
Descriptives for Dependent and Independent Variables
| Variable | Full Sample (n=200)
% or mean (SD) |
Ashe County (n=100)
% or mean (SD) |
Watauga County (n=100)
% or mean (SD) |
Description |
| Outcome Measures
Guilty |
79 |
80 |
78 |
0 = other; 1 = guilty |
| Fines | 78 | 79 | 76 | 0 = no; 1 = yes |
| Unsupervised | 54 | 40 | 68 | 0 = no; 1 = yes |
| probation Supervised probation |
23 |
39 |
7 |
0 = no; 1 = yes |
| Active time | 21 | 26 | 16 | 0 = no; 1 = yes |
| Independent and Control Measures
Male |
81 |
86 |
76 |
0 = female; 1 = male |
| Race
White |
83.50 |
78 |
89 |
0 = other; 1 = white |
| Other | 16.5 | 22 | 11 | 0 = other; 1 = white |
| Age | 32.98 (12.26) | 36.20 (12.95) | 32.98 (12.26) | Age in years (16-72) |
| Counsel
Retained |
61.50 |
37 |
86 |
0 = no; 1 = yes |
| Appointed/waived | 38.5 | 63 | 14 | 0 = no; 1 = yes |
In regards to disposition, as shown in Table 1, 80 percent of Ashe County defendants and 78 percent of Watauga County defendants were found guilty of driving under the influence. 79 percent of Ashe County defendants received fines as a result of their guilt, which was similar to the 76 percent in Watauga County. There were considerable differences in sentencing regarding probation between the two counties. Unsupervised probation was given in 40 percent of the Ashe County cases and 68 percent of the Watauga County cases. A higher percentage of the Ashe County defendants (39%) were sentenced to supervised probation than Watauga County defendants (7%). Active time in custody was given to a higher percentage of defendants in Ashe County (26%) compared to only 16 percent in Watauga County.
Bivariate descriptive analyses were conducted in order to assess the distribution of race, gender, and age across the dependent variables of guilt, fines, unsupervised probation, supervised probation, and active time. These analyses indicated that a higher percentage of whites (80.8%) were found guilty than defendants of nonwhite races (69.7%), as shown in Table 2. A higher percentage of white defendants (79.6%) were ordered to pay fines than nonwhite defendants (66.7%). White defendants were ordered to unsupervised probation at a higher rate (54.5%) than nonwhites (51.5%). Whites were also sentenced to supervised probation more (24.6%) than nonwhites (15.2%). A higher percentage of white defendants (21.6%) were sentenced to serving active time than nonwhites (18.2%).
Table 2.
Sentencing by Race, Gender, and Age
| White
(%) |
Other
(%) |
Male (%) | Female (%) | Mean (SD) | |
| Guilty | 80.8% | 69.7% | 77.2% | 86.8% | 33.39 (12.12) |
| Fines | 79.6% | 66.7% | 75.3% | 86.8% | 33.24 (12.00) |
| Unsupervised | 54.5% | 51.5% | 51.9% | 63.2% | 31.45 (12.05) |
| Probation | |||||
| Supervised Probation | 24.6% | 15.2% | 22.8% | 23.7% | 37.11 (10.96) |
| Active Time | 21.6% | 18.2% | 22.2% | 15.8% | 35.45 (10.75) |
| N=200 |
Race Gender Age
In regards to gender, a higher percentage of women (86.8%) were found guilty than men (77.2%), as shown in Table 2. More of the women in this sample paid fines (86.8%) than men (75.3%) as well. A higher percentage of women were sentenced to unsupervised probation (63.2%) than men (51.9%), as well as supervised probation (23.7% and 22.8%, respectively).
However, a higher percentage of men were sentenced to active time in custody (22.2%) than women (15.8%). Table 2 also shows the average age of defendants in each dependent variable category. The average age of both those found guilty and given fines was 33 years. The average age of those who were given unsupervised probation was 31 years of age, while the average was 37 years for supervised probation. The average age of defendants given active time in custody was 35 years.
Analysis of Variance (ANOVA)
Two-way ANOVA testing has been cited as the appropriate test to compare nominal or ordinal independent variables and their effects on the dependent variables (Szafran, 2012).
Therefore, an ANOVA test was used to assess the intersectional effects of race and gender on the dependent variables of guilty, fines, unsupervised probation, supervised probation, and active time. Defendant race and gender were not found to be significantly correlated with guilt (F(1)=.020, p=.89), fines (F(1)=.003, p=.96), unsupervised probation (F(1)=.262, p=.61), supervised probation (F(1)=271, p=.60), or active time (F(1)=.461, p=.50). In sum, no interactions were found to be statistically significant.
Binary Logistic Regression
In order to examine the effects of the independent and control variables on the five dichotomous dependent variables, binary logistic regression modeling was used, which is appropriate for dichotomous dependent variables (King, 2008). Using binary logistic regression, three statistical models were created for each of the dependent variables of guilt, fines, unsupervised probation, supervised probation, and active time in custody. The models reflected the entire sample and Ashe and Watauga counties separately. For the dependent variable of guilt, the full sample model (p=.35), the Ashe County model (p=.39), and the Watauga County model (p=.54) were not significant, and no significant relationships were found within any of the models, as seen in Table 3. However, it is important to note that the standard error for the variable of retained counsel for Watauga County was extremely high, as defendants who used an appointed attorney or waived their right to counsel in this county were all found guilty of the offense. There were also no significant relationships found in the binary logistic regression models of fines as shown in Table 4.
Table 3.
Binary Logistic Regression of Guilt
| Full Sample | Ashe County | Watauga County | ||||
| Variable | B (SE) | eB | B (SE) | eB | B (SE) | eB |
| Male | -.64 | .53 | -1.16 | .31 | -.31 (.64) | .73 |
| (.52) | (1.09) | |||||
| Race | ||||||
| White | .55 (.44) | 1.74 | .37 (.59) | 1.45 | .98 (.73) | 2.66 |
| Other | – | – | – | – | – | – |
| (Reference) | ||||||
| Age | .01 (.02) | 1.01 | .01 (.02) | 1.01 | -.01 (.03) | .99 |
| Counsel | ||||||
| Retained | -.19 | .82 | .65 (.57) | 1.91 | -20.21 | .00 |
| (.37) | (10613.67) | |||||
| Appointed/Waived | – | – | – | – | – | – |
| (Reference) Nagelkerke R2 |
.03 |
.06 |
.14 |
|||
| N (defendants) | 200 | 100 | 100 | |||
| p<.001*** p<.05** p<.10* | ||||||
Table 4.
Binary Logistic Regression of Fines
Full Sample Ashe County Watauga County
| Variable | B (SE) | eB | B (SE) | eB | B (SE) | eB |
| Male | -.73 (.52) | .48 | -1.17 (1.09) | .31 | -.43 (.63) | .65 |
| Race | ||||||
| White | .62 (.43) | 1.86 | .58 (.58) | 1.78 | .82 (.70) | 2.27 |
| Other (Reference) | – | – | – | – | – | – |
| Age | .01 (.02) | 1.01 | .01 (.02) | 1.01 | -.02 (.02) | .98 |
| Counsel | ||||||
| Retained | -.11 (.36) | .90 | .72 (.57) | 2.06 | -1.80 (1.12) | .17 |
| Appointed/Waived (Reference) | – | – | – | – | – | – |
| Nagelkerke R2 | .03 | .09 | .09 | |||
| N (defendants) | 200 | 100 | 100 |
p<.001*** p<.05** p<.10*
For unsupervised probation, retained counsel was found to have a positive significant relationship in both the full sample and Ashe County models (p=.00), as shown in Table 5. This finding indicates that those defendants who retained a private attorney had higher odds of receiving unsupervised probation than those who used a court-appointed attorney or had no attorney at all. In the Watauga County model, age had a slightly significant (p=.10) negative relationship with unsupervised probation, meaning that older defendants had lower odds of receiving unsupervised probation than younger defendants.
Table 5.
Binary Logistic Regression of Unsupervised Probation
| Full Sample | Ashe County | Watauga
County |
||||
| Variable | B (SE) eB | B (SE) eB | B (SE) eB | |||
| Male -.45 (.40) | .64 | .32 (.67) | 1.37 | -.60 (.57) | .55 | |
| Race | ||||||
| White -.05 (.41) | .95 | -.44 (.56) | .64 | .32 (.68) | 1.38 | |
| Other – | – | – | – | – | – | |
| (Reference) | ||||||
| Age -.02 (.01) | .99 | .00 (.02) | 1.00 | -.04 | .97 | |
| (.02)* | ||||||
| Counsel | ||||||
| Retained 1.32 (.32) | 3.73*** | 1.74 (.46) | 5.70*** | -.47 (.71) | .63 | |
| Appointed/Waived – | – | – | – | – | – | |
| (Reference) Nagelkerke R2 |
.15 |
.20 |
.07 |
|||
| N (defendants) | 200 | 100 | 100 | |||
| p<.001*** p<.05** p<.10* | ||||||
As shown in Table 6, the supervised probation binary logistic regression models found negatively significant relationships with retained counsel for both the full sample (p=.000) and the Ashe County sample (p=.004). These findings mean that defendants with privately retained attorneys had lower odds of receiving supervised probation than those with court appointed attorneys or no attorney at all. Race was found to have a slightly significant relationship with supervised probation in Ashe County, as whites had higher odds of receiving the sentence than nonwhites (p=.09). It is important to note that the white variable for Watauga County had a high standard error because all defendants in the supervised probation category for the Watauga model were white. None of the defendants in Watauga County who received supervised probation as a part of their sentence were of a nonwhite race.
Table 6.
Binary Logistic Regression of Supervised Probation
| Full Sample | Ashe County | Watauga County | |||||
| Variable | B (SE) | eB | B (SE) | eB | B (SE) | eB | |
| Male | -.09 (.47) | .91 | -.88 (.64) | .42 | .59 (1.15) | 1.80* | |
| Race | |||||||
| White | .86 (.56) | 2.37 | 1.01 (.60) | 2.74* | 18.69 | 130313828.00 | |
| (11867.55) | |||||||
| Other | – | – | – | – | – | – | |
| (Reference) | |||||||
| Age | .02 (.01) | 1.02 | .01 (.02) | 1.01 | .03 (.03) | 1.03 | |
| Counsel | |||||||
| Retained | -1.88 (.39) | .15*** | -1.44 (.50) | .24** | -.82 (.99) | .44 | |
| Appointed/Waived | – | – | – | – | – | – | |
| (Reference) Nagelkerke R2 |
.24 |
.18 |
.11 |
||||
| N (defendants) | 200 | 100 | 100 | ||||
p<.001*** p<.05** p<.10*
Regarding active time, the entire sample model found a slight, negative association with retained counsel (p=.03), as did the Watauga County model (p=.06), meaning that those with retained counsel had lower odds of receiving active time as a sentence as shown in Table 7. The Watauga County model also produced a slightly positive significance with gender, as males had higher odds of receiving active time than females (p=.09). No other significant relationships were found for active time.
Table 7.
Binary Logistic Regression of Active Time
| Full Sample | Ashe County | Watauga County | ||||
| Variable | B (SE) eB | B (SE) eB | B (SE) eB | |||
| Male .42 (.49) | 1.53 | -.60 (.64) | .55 | 1.89 (1.11) | 6.63* | |
| Race | ||||||
| White .31 (.51) | 1.36 | .24 (.60) | 1.27 | .72 (1.15) | 2.06 | |
| Other – | – | – | – | – | – | |
| (Reference) | ||||||
| Age .01 (.01) | 1.01 | -.01 (.02) | .99 | .03 (.03) | 1.03 | |
| Counsel | ||||||
| Retained -.81 (.36) | .44** | -.39 (.49) | .68 | -1.43 (.75) | .24* | |
| Appointed/Waived – | – | – | – | – | – | |
| (Reference) Nagelkerke R2 |
.06 |
.03 |
.18 |
|||
| N (defendants) | 200 | 100 | 100 | |||
| P<.001*** p<.05** p<.10* | ||||||