US Public Attitudes Towards Welfare State | Research Study

Tim Mulligan

There's a specialist from your university waiting to help you with that essay.
Tell us what you need to have done now!

order now

Employment and Attitudes toward People on Welfare

Welfare is one of the United States most prominent political issues. Since the U.S welfare system was established in 1935, its fiscal structure, the source of its funding and the qualifications of its recipients have been continuous topics of debate. Because of America’s highly diverse population, a plethora of attitudes have developed regarding the way that people view welfare recipients, and this may be attributed to many different factors. I took it upon myself to look more specifically at the relationship between individuals who work (or do not work) and people who are on welfare. The question that I decided to research was, “does an individual’s employment status influence their attitude towards people who are on welfare?” I believe that this is an important question to address because people tend to generalize that individuals who work have harshly negative attitudes toward people receiving welfare checks because they do not have to work for the money. If this is in fact true, then I believe it would play a huge role in the outcomes of many elections as well as how states organize their welfare systems.

My hypothesis is Ha: in a comparison of individuals, those who are currently working will have more negative feelings towards people who are on welfare than individuals who are not working. My null hypothesis would be H0: there is no relationship between an individual’s employment status and their feelings toward people who are on welfare. I believe my hypothesis to be true because I think it would be very hard to find a person who works and exerts themselves to receive an income and is also tolerant of other individuals who are receiving money without having to work. Some people may feel that their work and efforts are belittled because individuals who do not put forth the same effort can still claim an “income”. There may also be individuals who had experienced financial hardship (like many of the people who utilize welfare) but worked their way back into financial stability without the aid of welfare. These people may have a more negative “if I could do it, then they should be able to do it” attitude towards people on welfare. I think this hypothesis is applicable to individuals in all types of occupations but even more so to individuals in the manual labor work force. People who work lower paying manual labor jobs could have extremely negative views towards people who are welfare because they are physically exerting themselves while welfare recipients may not have to do so themselves.

On the other end of my hypothesis, individuals who are not employed could have more positive feelings toward people on welfare for several reasons. The most notable reason is that there is probably a higher chance that individuals who are not employed may in fact be receiving welfare aid themselves. I do not believe that individuals who are currently on welfare will have negative feelings towards the very program that they are using. Another factor could be individuals who are not necessarily “out of the job” but are simply not actively looking to work. For example, housewives, non-working students and young adults may not have the same negative feelings as someone who is employed because they do not have a job or income to compare with those of people who are on welfare. These groups of people may not have the same “belittled” feeling that employed people may have and they may have more neutral or positive feelings towards people who are on welfare.

The data set that I used for my analysis is nes2008. This dataset is from an American National Election Time Series Study which took place in 2008. 4,424 total individuals were interviewed on a face-to-face basis, 2,322 individuals before the presidential election and 2,102 individuals after the presidential election. As can be assumed by the face-to-face polling the unit of analysis for this study was individuals. (ANES)

The integrity of this data set is strong in the way that the individuals were polled on a wide variety of topics such as their voting participation, values, familiarity with the media and their ideologies. This helps to insure that the individuals do not feel as though they are being interviewed for a specific topic or to answer a specific question which could swayed their answers in a less accurate direction. The large number of people who were sampled is also a positive aspect of the data set. Though four thousand people may not perfectly represent the opinions of the entire population of the U.S, the sample size is large enough to generate at least a sufficient representation. (ANES)

On the other hand, the nes2008 data set does have a few negative characteristics. The interviewing of individuals pre-and-post-election may have generated results that inconsistently represent the U.S population because of the effect that the election may have had on some people’s views or answers. Although the two waves of interviewees consisted of different people, the election may have influenced individuals to respond more positively or negatively to certain questions based on the outcome of the election. The population could have been represented far differently before the election than after the election. This may be an effect that the study was trying to induce, but for my research it does not generate the best representation of the population. Another issue with the nes2008 data set is that there was a designed oversampling of African-American and Latino respondents. This oversampling presents another issue in regards to the studies representation of the general population as it may not include as many answers from other races that could affect my testing outcomes. Luckily, the data set included a formula that would help to weigh the data in a way that would better represent the population. (ANES)

The dependent variable that I selected was welfare_therm. This is a continuous variable that asks for individuals to rate the warmth of their feelings toward people who are on welfare from 0? (coldest) to 100? (warmest). It is implied that warmer feelings are more positive than colder feelings. This was a good variable for me to use because the question that I am trying to answer pertains to individual’s feelings toward people who are on welfare. I believe rating their feelings in degrees rather than categories like “negative”, “slightly negative”, “neutral” and so on allows for individuals to be more specific when describing their feelings towards people on welfare. Although, I do believe that the wide range of the thermometer may bring about a less definitive description of what is considered a mildly positive or mildly negative feeling toward people who are on welfare. A graph depicting welfare_therm can be seen in figure 1.

My main independent variable was employ_status, which had individuals identify themselves within employment status categories. These categories were: working now, temporarily laid off, unemployed, retired, permanently disabled, homemaker, and student. At first, this variable did not present the most valid measurement of employment status that I would need for my research. To generate a better representation of the feelings generated by individuals who were working or not working, I had to refine the number of categories in the variable. I recoded the variable so that an individual’s response would either register as A. working or B. not working. This new variable was called working and would serve as a better variable for measuring a relationship with my dependent variable, feelings toward people on welfare. A graph depicting working can be seen in figure 2.

The first of my control variables was gender. This variable categorized individual respondents as either male or female. It is important to note that because of the way that this variable was coded in Stata (1=male, 2=female), I needed to recode it so that it would be more easily measured by my tests. I recoded the variable as 0=male and 1=female and I named the new variable female. I included this control variable because I believed that an individual’s gender would have a large impact on the feelings that they had towards people who are on welfare. Stereotypically women are assumed to be more emotional and sympathetic towards individuals who may be in need and I thought that this might have an effect on their attitude towards a person who is on welfare.

The second control variable that I included in my test was hh_kids, which is a categorical measure of the number of kids in the respondent’s household. 0=no kids 1=one kid and 2=two or more kids in the household. I believe that this variable would have served my research better if the categories represented the dynamic of households with few kids and households with many kids better. Perhaps categories such as 0 kids, 1-3 kids and 3 or more kids would have been better because I do not think that 2 kids represents a household with “many” kids, which was the dynamic I was aiming to measure. I do believe that this variable is sufficient, though. I believe that the number of kids that an individual has in their household effects their feelings toward people on welfare because individuals with many children may know what it’s like to be on a tight budget or to have to provide for children. People with many kids in their home could be sympathetic towards people on welfare because they might be under the impression that the people who are on welfare need it to support their children.

Income_r was my third control variable. This variable reports the income of the respondent within twenty five categories that range from “none or less than $2,999” to “$150,000 and over”. Unfortunately, the categories are not equally sized. For example, there is a category labeled “$15,000-16,999” and its subsequent category is labeled “$17,000-$19,999”. The former category has a range of $1,999 dollars and the latter a range of $2,999 and this difference in category size occurs throughout the variable. This may be a weak point of this control variable. None the less, I included this control variable because I believe that the lower an individual’s income, the more understanding or warm their feelings may be for people who are on welfare. In opposition people with high incomes who may work very hard for their money may have colder feelings for people on welfare and who may be out of work.

The final control variable that I included in my tests was relig_attendHi. This variable categorizes an individual’s level of religious attendance as either low or high. This variable may be weak because of the fact that there is no knowing exactly what amount of attendance fits the description of “high” or “low’ attendance. One respondents idea of high attendance could be once a week, where as another respondent could consider once every few months to be high attendance. There is also no telling where the dividing line between high and low may be. That being said, I believe that religious attendance has a significant impact on an individual’s feelings towards people on welfare because many religions are proponents of acts of charity or helping those who are in need. Individuals who are religious may be more inclined to see welfare as a means of helping those who need financial help rather than an unfair handout.

I used a multiple regression test to interpret the relationship between an individual’s employment status and their feelings toward people who are on welfare. I used this test because a multiple regression test is appropriate for my dependent variable which is continuous. It was also essential that I used a multiple regression test so that I would be able to control for my Z (control) variables when determining the relationship between employment status and feelings toward people who are on welfare.

After running my multiple-regression test on the responses of 1,922 individuals, I found that an individual’s employment status does matter when it comes to their feelings toward people who are on welfare. The employment status of an individual is statistically significant at the 95% confidence interval and my test showed that an individual that is working is likely to have feelings that are 2.49 degrees colder than an individual that is not working regarding people who are on welfare. (coefficient estimate) This provides me with enough evidence to state that there is support for my hypothesis that working individuals have more negative feelings toward individuals on welfare than people who are not working. These results can be seen in figure 3.

Gender, income, and religious attendance also tested to be statistically significant at the 95% confidence interval. These outcomes confirm my original predictions for these variables. Women are likely to have feelings that are 2.46 degrees warmer than males. This could in fact indicate that women are more sensitive and tolerant of people who are in need and may be using welfare for survival. Also, the higher category of income that an individual is in the colder their feelings get towards individuals who are on welfare. This is in line with my prediction that richer people may not be fond of people getting money from the government for no work, and poorer people being more understanding of people on welfare’s need for an income. People with higher religious attendance also show to have 3.56 degrees warmer feelings toward people who are on welfare than people with low religious attendance. Religious attendance appeared to be the control variable that was the most indicative of an individual’s feelings toward people who are on welfare. This also gave validity to my prior thought that people with higher religious attendance may be more open minded to acts of charity and aiding those in need.

The number of kids in an individual’s household did not turn out to be statistically significant. This disappointed me because I thought that people with more kids would have significantly different feelings toward people who are on welfare than people with few or no kids in their household. I believe that this may be because most people who have children have planned for their kids financially and may not have as high of a tolerance for people on welfare that may not have planned for the children they are supporting.

The R-squared value for my multiple regression test was only 0.0602. This means that the independent and control variables that I included in my test only accounted for six percent of the total variance in my dependent variable, feelings toward individuals on welfare. I thought that the control variables that I selected would have yielded higher variance accountability. It turns out that employment status, gender, income, number of kids in a household and religious attendance are only a few of the many factors that can influence an individual’s feelings toward welfare recipients.

In conclusion, I discovered that employment status is statistically significant when determining a person’s feelings towards people who are on welfare. I can reject my null hypothesis which is that there is not relationship between an individual’s employment status and their feelings toward people on welfare. In relation to the real world, one could say that people who work are more likely to view people who are on welfare more negatively. This could be a product of many things including an individual’s views on work ethic, morals, fairness, equality and much more. It would be interesting to research what exactly causes a working person to view welfare recipients more negatively. Although I have found support for my hypothesis, there are many more control variables that I could include in future research of this question. I believe that political ideology would influence the way a person feels about people on welfare because liberals and conservatives have specific views and welfare policies and who should be recipients. The state in which an individual lives in could also be a good control variable to include because although all states are a part of the federal welfare system, different states have different internal welfare systems that could spark different opinions. I also think it would be important to divide the term “welfare” into its different categories such as unemployment, healthcare, childcare etc. because I think that people tend to make the generalization that welfare means “unemployment checks”.


Figure 1

Figure 2

Figure 3

Works Cited

“ANES Data Center Study Pages ANES 2008 Time Series Study.” ANES Data Center Study Pages ANES 2008 Time Series Study. Accessed May 6, 2015.