Do you think the variables are appropriately used?
Respond to this post and comment on the following:
- Do you think the variables are appropriately used? Why or why not?
- Does the addition of the control variables make sense to you? Why or why not?
- Does the analysis answer the research question? Be sure and provide constructive and helpful comments for possible improvement.
- If there was a significant effect, comments on the strength and its meaningfulness.
- As a lay reader, were you able to understand the results and their implications? Why or why not?
Post
and both respondent income and highest level of schooling.
Kim
References
Christ, S. L., Fleming, L. E., Lee, D. J., Muntaner, C., Muennig, P. A., & Caban-Martinez, A. J. (2012). The effects of a psychosocial dimension of socioeconomic position on survival: Occupational prestige and mortality among us working adults. Sociology of Health & Illness, 34(7), 1103-1117. http://dx.doi.org/10.1111/j.1467-9566.2012.01456.x
Data Set 2014 General Social Survey [Dataset File]. (2014). Retrieved from https://class.waldenu.edu
Frankfort-Nachmias, C., & Leon-Guerrero, A. (2015). Social statistics for a diverse society (7th ed.). Thousand Oaks, CA: Sage Publications, Inc.
Grove, S. K., Burns, N., & Gray, J. R. (2013). The practice of nursing research: Appraisal, synthesis, and generation of evidence(7th ed.). St. Louis: MO: Elsevier.
Kent State University. (2014). SPSS tutorials: Pearson correlations. Retrieved from http://libguides.library.kent.edu/SPSS/PearsonCorr
Laureate Education (Producer). (2016). Multiple regression [Video file]. Baltimore, MD: Author.
Miller, D. C., & Salkind, N. J. (2002). Scales assessing social status. In Handbook of research design & social measurement (6th ed., pp. 455-469). Thousand Oaks, CA: SAGE Publications, Inc. http://dx.doi.org/10.4135/9781412984386.n81
Segrin, C. (2010). Multiple regression. In N. J. Salkind (Ed.), Encyclopedia of Research Design (pp. 844-849). http://dx.doi.org/10.4135/9781412961288
U.S. Bureau of Labor Statistics. (n.d.). National longitudinal survey of older and young men: Occupations and occupational prestige indices. Retrieved from https://www.nlsinfo.org/content/cohorts/older-and-young-men/topical-guide/employment/occupations-and-occupational-prestige
Here is the post – It was not properly paste
Multiple Regression
For this discussion, I was interested in exploring the relationship between a dependent variable and two independent variables, all measured at the metric level. For this type of study, multiple regression is indicated and allows the researcher to determine the effects of two or more independent variables on a dependent variable (Frankfort-Nachmias & Leon-Guerrero, 2015). In this discussion, I will propose a research design requiring multiple regression, describe the variables, interpret the results, and comment on the strength of the effect. In addition, I will provide a brief explanation in layman’s terms to further explain the answer to the research question.
Research Question and Null Hypothesis
I chose to explore the relationship between three quantitative variables measured at the metric level from the Data Set 2014 General Social Survey (“Data Set”, 2014). Variables included the respondent’s occupational prestige score measured in 2010 (prestg10), the RESPONDENT INCOME IN CONSTANT DOLLARS (conrinc), and HIGHEST YEAR OF SCHOOL COMPLETED (educ). I propose the following research question: Does income measured in constant dollars and the highest level of schooling completed predict occupational prestige. An appropriate null hypothesis would read: There is no relationship between income and highest level of schooling and occupational prestige scores. Stated differently, I could hypothesize that income and highest level of education are not predictors of occupational prestige.
Research Design
This research question explores the relationship between three metric-level variables. Since the variables are not manipulated and are extracted from an existing data set, a non-experimental, correlational, quantitative research design is indicated (Grove, Burns, & Gray, 2013). Building on correlation coefficients used with bivariate testing, multiple regression is indicated given the number of variables of interest. Correlation coefficients are calculated to determine if there exists a linear relationship between variables (Kent State University [Kent State], 2014). Multiple regression allows the researcher to test to what extent (degree of the relationship) two or more independent variables explain variation in a dependent variable (Segrin, 2010). Multiple regression can be used for continuous and categorical variables and assumes a linear relationship between variables and normal distribution of the dependent variable (Segrin).
Description and Justification of Variables
The prestg10 variable describes the respondent’s occupational prestige score and serves as the dependent variable. Prestige scores speak to the perception of an individual’s social standing based on their professional position and are used in determining social status (Miller & Salkind, 2002). The conrinc variable describes the respondent’s income in constant dollars and the educ variable describes the respondent’s highest year of school completed. Income and education serve as the independent variables. All variables are measured at the quantitative, metric level.
I performed initial correlation and bivariate regression testing using SPSS Statistical Software between the prestg10 (dependent) and conrinc (independent) variable to determine if there was a statistically significant relationship between them. Output data are reported in Figure 1 below for ease of review.
Figure 1
Correlations
Rs occupational prestige score (2010)
RESPONDENT INCOME IN CONSTANT DOLLARS
Rs occupational prestige score (2010)
Pearson Correlation
1
.424**
Sig. (2-tailed)
.000
N
2427
1523
RESPONDENT INCOME IN CONSTANT DOLLARS
Pearson Correlation
.424**
1
Sig. (2-tailed)
.000
N
1523
1523
**. Correlation is significant at the 0.01 level (2-tailed).
Model Summary
Model
R
R Square
Adjusted R Square
Std. Error of the Estimate
1
.424a
.180
.179
12.376
a. Predictors: (Constant), RESPONDENT INCOME IN CONSTANT DOLLARS
ANOVAa
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
51107.526
1
51107.526
333.659
.000b
Residual
232976.310
1521
153.173
Total
284083.836
1522
a. Dependent Variable: Rs occupational prestige score (2010)
b. Predictors: (Constant), RESPONDENT INCOME IN CONSTANT DOLLARS
Results of the Pearson correlation revealed a statistically significant, moderately positive correlation between occupational prestige score and respondent income (R=0.424, p=.000; alpha = 0.05) suggesting that the prestige score increases as income increases. Bivariate regression revealed an R2 of 0.180. I concluded that 18% of the prestige score is explained by income. Analysis of variance output revealed that the regression model was statistically a better predictor of the relationship between variables than by just comparing variable means (F=333.659, p=0.000; alpha = 0.05). However, since income only accounted for 18% of variation in occupational prestige scores, I concluded that other variables may affect (control) prestige scores; therefore, it was necessary to add an additional variable to the regression model (Frankfort-Nachmias & Leon-Guerrero, 2015). A quick search of literature related to occupational prestige identified income and education as two factors commonly studied in relation to occupational prestige (Christ et al., 2012; U.S. Bureau of Labor Statistics, n.d.). Considering, I decided to conduct a multiple regression test, adding highest level of schooling as a second independent variable.
SPSS output for the multiple regression testing is provided in Figure 2 below.
Figure 2
Model Summary
Model
R
R Square
Adjusted R Square
Std. Error of the Estimate
1
.561a
.315
.314
11.318
a. Predictors: (Constant), HIGHEST YEAR OF SCHOOL COMPLETED, RESPONDENT INCOME IN CONSTANT DOLLARS
ANOVAa
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
89366.609
2
44683.305
348.806
.000b
Residual
194717.227
1520
128.103
Total
284083.836
1522
a. Dependent Variable: Rs occupational prestige score (2010)
b. Predictors: (Constant), HIGHEST YEAR OF SCHOOL COMPLETED, RESPONDENT INCOME IN CONSTANT DOLLARS
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
B
Std. Error
Beta
1
(Constant)
15.191
1.418
10.714
.000
RESPONDENT INCOME IN CONSTANT DOLLARS
.000
.000
.286
12.586
.000
HIGHEST YEAR OF SCHOOL COMPLETED
1.795
.104
.392
17.282
.000
a. Dependent Variable: Rs occupational prestige score (2010)
The Pearson’s multiple correlation coefficient demonstrated a moderate-to-strong relationship between the dependent variable, occupational prestige score, and both independent variables, income and education (R=0.561). Therefore, I reject the null hypothesis that income and highest year of schooling does not predict occupational prestige scores. There was a slight difference in the R2 (0.315) and adjusted R2 (0.314) scores. Since the analysis included multiple independent variable predictors, the adjusted R2 score was interpreted (Laureate Education, 2016). Based on the Model Summary, I conclude that approximately 31% of an occupational prestige score is explained by the combination of income and education. Based on analysis of variance (ANOVA) results, I conclude that the regression model is statistically significant and serves as a better predictor of the relationship between variables than just comparing their means (F=348.806, p=0.000, alpha = 0.05). Coefficient analysis reveals a Y intercept (a) of 15.191 indicating where the regression line would cross the y-axis.
Unstandardized coefficient results related to income to prestige score were interesting. According to the SPSS output, for every one unit increase in income, there is a 0.000 increase in occupational prestige score, controlling for the highest year of school completed. This model proved statistically significant (t=12.586, p=0.000, alpha=0.05). I initially questioned how the unstandardized coefficient between income and prestige score could be zero, yet statistically significant. I suspect that because the income variable is measured in constant dollars, a one unit increase equates to only one dollar, a very minute unit, which is too small to register as an unstandardized coefficient (the result is likely several additional places away from the decimal). In order to address this issue, I could rescale the income variable into greater units, such as income to nearest 100 or 1000. To gain more insight into this relationship, beta analysis will be helpful.
Examining the results related to income, I conclude the model between income and prestige scores was statistically significant suggesting that for every one year increase in school completed, the prestige score increases 1.795 (t=17.282, p=0.000, alpha = 0.05), controlling for respondent income. In summary, both independent variables (income and education) are predictors of the dependent variable, occupational prestige.
Standardized coefficients, termed beta, standardizes units across variables and suggests that for every one standard deviation unit increase for an independent variable, the dependent variable’s standard deviation will change by the beta amount (Laureate Education, 2016). Based on these data, I conclude that for every one standard deviation increase in education, prestige score standard deviations will increase by 0.392. For every one standard deviation increase in income, prestige scores standard deviations will increase by 0.286. Standardizing units did, indeed, help to validate the statistical significance between income and prestige scores, significance not appreciable or understandable just by analyzing the related unstandardized coefficient alone.
Interpretation for Lay Audiences
The research question asked whether income (measured in constant dollars) and highest level of schooling predict occupational prestige scores. A hypothesis was tested that stated income and highest level of education were not related to occupational prestige and were not predictors of prestige scores. Based on statistical analysis using computer software, I concluded there is a statistically significant, moderate-to-strong relationship between occupational prestige scores and both income and highest level of education. Approximately 31% of the prestige score is explained by income and education. To be more specific, I concluded that prestige scores increase as income and education increase. For every year of school completed, the prestige score increases by 1.795. Interpretation of the relationship between income and prestige scores is a bit more challenging due to the way income is measured; however, further statistical analysis did confirm the same type of positive relationship; when income increases, prestige scores increase. Cause-and-effect is not suggested by this data; only that a positively correlated relationship exists (Frankfort-Nachmias & Leon-Guerrero, 2015).
Conclusion
Correlation and multiple regression testing provide a flexible statistical method for exploring the relationships between two or more independent and one dependent variable (Segrin, 2010). For this discussion, I proposed a research question and associated null hypothesis that explored the relationship between occupational prestige scores, income, and education. Pearson’s multiple correlation coefficient and multiple regression testing determined a statistically significant, moderate-to-strong positive relationship between occupational prestige scores and both respondent income and highest level of schooling.
Kim
References
Christ, S. L., Fleming, L. E., Lee, D. J., Muntaner, C., Muennig, P. A., & Caban-Martinez, A. J. (2012). The effects of a psychosocial dimension of socioeconomic position on survival: Occupational prestige and mortality among us working adults. Sociology of Health & Illness, 34(7), 1103-1117. http://dx.doi.org/10.1111/j.1467-9566.2012.01456.x
Data Set 2014 General Social Survey [Dataset File]. (2014). Retrieved from https://class.waldenu.edu
Frankfort-Nachmias, C., & Leon-Guerrero, A. (2015). Social statistics for a diverse society (7th ed.). Thousand Oaks, CA: Sage Publications, Inc.
Grove, S. K., Burns, N., & Gray, J. R. (2013). The practice of nursing research: Appraisal, synthesis, and generation of evidence (7th ed.). St. Louis: MO: Elsevier.
Kent State University. (2014). SPSS tutorials: Pearson correlations. Retrieved from http://libguides.library.kent.edu/SPSS/PearsonCorr
Laureate Education (Producer). (2016). Multiple regression [Video file]. Baltimore, MD: Author.
Miller, D. C., & Salkind, N. J. (2002). Scales assessing social status. In Handbook of research design & social measurement (6th ed., pp. 455-469). Thousand Oaks, CA: SAGE Publications, Inc. http://dx.doi.org/10.4135/9781412984386.n81
Segrin, C. (2010). Multiple regression. In N. J. Salkind (Ed.), Encyclopedia of Research Design (pp. 844-849). http://dx.doi.org/10.4135/9781412961288
U.S. Bureau of Labor Statistics. (n.d.). National longitudinal survey of older and young men: Occupations and occupational prestige indices. Retrieved from https://www.nlsinfo.org/content/cohorts/older-and-young-men/topical-guide/employment/occupations-and-occupational-prestige
Respond to this post and comment on the following:
Do you think the variables are appropriately used? Why or why not?
Does the addition of the control variables make sense to you? Why or why not?
Does the analysis answer the research question? Be sure and provide
constructive and helpful comments for possible improvement.
If there was a significant effect, comments on the strength and its
meaningfulness.
As a lay reader, were you able to understand the results and their
implications? Why or why not?
