Voice For Liberty

Category: Politics

  • Homicide in the States

    Homicide in the States

    Predicting state homicide rates from two factors: Political sentiment and the state’s total population.

    We tried to predict state homicide rates from two factors: how much of the vote went to Donald Trump in 2024, and the state’s total population. The relationship with population is very weak. There’s a slight hint that states with higher Trump vote shares might have higher homicide rates, but this finding is right on the edge of typical “significance” and should be taken with a grain of salt. Overall, these two factors, by themselves, do not explain much of the difference in homicide rates across states.

    Click for larger

    Regression Statistics

    Below is a table of the regression coefficients, their standard errors, t-values, p-values, and 95% confidence intervals:

    TermEstimateStd. Errort valuep valueLower CIUpper CI
    (Intercept)~0.53~2.87~0.190.853~-5.24~6.31
    x (Trump %)~10.20~5.08~2.010.051~-0.02~20.42
    size_values5.98e-086.61e-08~0.900.371-7.34e-081.93e-07

    Model Fit Statistics:

    • R-squared = ~0.084
    • Adjusted R-squared = ~0.044
    • F-statistic (df=2,46) = ~2.10 (p = 0.134)
    • Number of Observations = 49

    Detailed Explanation of the Results

    Intercept: The estimated intercept is approximately 0.53, but with a large standard error and a high p-value (0.853). This indicates that if both predictors (Trump vote share and population) were theoretically zero, the predicted homicide rate would be around 0.53 murders per 100k. However, this is not statistically distinguishable from zero.

    Share of Trump Vote (x): The coefficient is roughly 10.20 (p = 0.051), which is borderline significant at the 5% level. The positive sign suggests that, when state population is held constant, higher projected Trump vote share is associated with a higher homicide rate. Specifically, for each 1.0 increase in Trump vote share (that is, going from 0% to 100%, which is not realistic but a linear extension), the model predicts an increase of about 10 homicides per 100k. However, it’s important to note that 0.051 is slightly above the traditional 0.05 significance cutoff, so this result is borderline and should be interpreted cautiously.

    State Population (size_values): The coefficient is very small (~6.0e-08) and not statistically significant (p = 0.371). This means that once the model takes the share of Trump vote into account, there’s no clear evidence that the total population size by itself helps predict the homicide rate in a linear sense.

    Model Fit:

    • The R-squared (0.084) and adjusted R-squared (0.044) are relatively low, indicating that only around 8% of the variation in homicide rates is explained by these two predictors.
    • The overall F-statistic is not significant at conventional thresholds (p = 0.134), which further suggests caution in drawing strong conclusions from this model.

    Interpretation and Caution:

    • The data set is relatively small (49 states), and we have an NA for one case.
    • High multicollinearity or other issues could be at play (the condition number in the background is quite large), possibly because population can vary widely and might interact with many other state-level factors.
    • The borderline p-value for the Trump vote share means the result could go either way with more data or slightly different model specifications.

    In summary, with this simple linear model, we do not see strong or definitive evidence that either state population or projected Trump vote share is a powerful linear predictor of homicide rates. There is a weak (borderline) indication that higher Trump share might relate to higher homicide rates, but the effect fails to reach the conventional threshold of statistical significance at the 5% level, and the overall explanatory power of the model is low.

  • Adult Literacy and Numeracy in Counties

    Adult Literacy and Numeracy in Counties

    Analyzing literacy and numeracy rates in counties alongside political sentiment.

    PIAAC (the Program for the International Assessment of Adult Competencies) defines different levels for adult skills in literacy and numeracy, ranging from below Level 1 up to Level 5. Level 3 is often considered a baseline or minimum desired level of proficiency for dealing with the demands of everyday life and work in a modern society. In more detail:

    Level 3 in Literacy: At Level 3, tasks typically require integrating, interpreting, or synthesizing information from dense or lengthy texts. Individuals can identify important details, make some inferences, and navigate moderate complexity in reading materials.

    Level 3 in Numeracy: At Level 3, tasks may require multiple steps and the interpretation of numerical information in various formats (e.g., tables, charts, or text). Individuals can handle somewhat more complex calculations, interpret data, and apply reasoning to solve problems in real-life contexts (budgeting, scheduling, analyzing simple graphs, etc.).

    When a chart shows the share of adults “at or above Level 3,” it means the percentage of people who reach at least this medium-range skill level. Individuals below Level 3 may struggle more with everyday tasks — like understanding multi-paragraph instructions or making sense of typical quantitative information — whereas those at or above Level 3 can manage those tasks more successfully.

    I gathered county-level data from PIAAC and plotted it along with the percent of voters in the county that chose Donald Trump in 2020. The PIAAC data is for 2017. I also asked ChatGPT to help me explain the chart. (Click chart for larger)

    These two scatterplots each show U.S. counties as bubbles, where:

    • Horizontal axis (x-axis): The percentage of votes cast for Donald Trump in 2020 for each county.
    • Vertical axis (y-axis): The percentage of adults in the county who score “at or above level 3” in either literacy (left chart) or numeracy (right chart).
    • Bubble size: Proportional to the county’s population. Larger circles represent more populous counties.

    In other words, each point corresponds to a county. The position on the x-axis shows how strongly that county voted for Trump, while the position on the y-axis shows the share of adults who have at least level-3 skills in literacy or numeracy (based on data from PIAAC). Some key takeaways:

    1. Negative relationship: In both charts, you can see that as the percentage of Trump votes increases (moving right on the x-axis), the average proportion of adults scoring at or above level 3 in literacy/numeracy tends to go down. In other words, there’s a negative association between Trump’s vote share and higher literacy/numeracy skills in these data.

    2. Wide spread but a clear pattern: While there is a cluster around the middle, many points in the lower-right quadrant indicate that counties with higher Trump vote shares often have a smaller share of adults at or above level 3. Meanwhile, counties with lower Trump vote shares (points farther left on the x-axis) tend to show higher percentages of adults meeting level-3 thresholds.

    3. Bubble sizes and population: Larger circles often appear to the left (that is, in counties where Trump’s vote share was lower) and frequently have higher literacy/numeracy. These big circles represent large, more populous counties. Conversely, many smaller circles — which represent less populous counties — are distributed across the rest of the charts, including in areas where a higher share of votes went to Trump.

    Overall, the charts illustrate that — within these data — counties that leaned more heavily toward Trump in the 2020 election are on average associated with somewhat lower proportions of adults at or above a level-3 threshold in literacy or numeracy, and vice versa.

  • Women, Infants, and Children (WIC) Spending by State

    Women, Infants, and Children (WIC) Spending by State

    Participation and spending on WIC by state political sentiment.

    The Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) is a federal assistance initiative that provides nutritious foods, education, and support to low-income pregnant and postpartum women, infants, and children up to age five. Funding for WIC is primarily allocated by the U.S. Department of Agriculture’s Food and Nutrition Service (USDA-FNS) to state agencies, which then administer the program at the local level.

    Data on the program is available at WIC Data Tables.

    I’ve plotted data by state, using the political sentiment of the state as a dependent variable. This is measured by the portion of the vote Donald J. Trump received in 2024. Both spending and participation vary little based on a state’s vote for Trump. (Click for larger charts)

  • Cost of Voting and State Politics

    Cost of Voting and State Politics

    Examining the relationship between state voting laws and political sentiment.

    The Cost of Voting Index is an ongoing analysis by Michael J. Pomante II, Ph.D., a political scientist. It “examines election laws and policies and calculates a single measure of the relative difficulty of voting for each state. States with smaller values make voting more accessible than states with larger values.” Data and accompanying peer-reviewed papers are available on the Cost of Voting Index website.

    Briefly, “The COVI measures the time and effort required to vote, encompassing criteria such as registration requirements, voter identification laws, early voting availability, and absentee voting procedures.” (1)Pomante II, Michael J. (2024). “The Cost of Voting in the American States: 2024.” Election Law Journal: Rules, Politics, and Policy. DOI: 10.1089/elj.2024.0037. Available from https://costofvotingindex.com/publications I’ve taken the data for 2024 and plotted it against the percent of Donald Trump voting in the states. Chart 1 shows this. (click for larger charts)

    I asked ChatGPT to perform a regression analysis and explain. The summary was: “This analysis examines how the difficulty of voting (measured by the “Cost of Voting Index”) varies with a state’s political leanings and population. The results suggest that states with a higher percentage of Trump voters in 2024 tend to have stricter voting laws (higher cost of voting). However, state population size does not seem to have a meaningful impact on voting difficulty.” In more detail:

    We performed a multiple linear regression to examine the relationship between:

    • Independent variables:
      • Share of Trump Vote in 2024 (X1)
      • State Population (X2)
    • Dependent variable:
      • Cost of Voting Index (Y)

    Key Findings from the Regression:
    – R-squared (0.425): This indicates that approximately 42.5% of the variation in the Cost of Voting Index can be explained by the Share of Trump Vote and State Population.
    – Share of Trump Vote (Coefficient = 8.5481, p-value < 0.001): A statistically significant predictor of the Cost of Voting Index, suggesting that as the share of Trump votes increases, the Cost of Voting Index also tends to increase.
    – State Population (Coefficient = 0.8143, p-value = 0.751): This variable is not statistically significant, meaning there is no strong evidence that population size affects the Cost of Voting Index.

    I also examined voter turnout rate as a function of the Cost of Voting Index. Chart 2 illustrates this:

    As before, I asked ChatGPT for a regression analysis: “Think of this as trying to understand what factors influence how many people show up to vote in each state. We looked at whether making voting harder (higher Cost of Voting Index) or having a larger population affects turnout. The results suggest that voter turnout may slightly drop when voting is harder, but the effect is weak. However, the population of a state does not seem to play a major role in voter turnout. The model itself doesn’t explain much of the variation, meaning there are likely other factors (like political engagement, campaign efforts, or demographics) that influence voter turnout more strongly.” In more detail:

    We performed a multiple linear regression analysis to determine the relationship between Voter Turnout (dependent variable) and two independent variables:

    1. Cost of Voting Index (measuring how easy or difficult it is to vote in a state)
    2. State Population (the number of residents in the state)

    Key Findings:
    – The R-squared value is 0.068, meaning that only 6.8% of the variation in voter turnout is explained by the Cost of Voting Index and State Population.
    – The Cost of Voting Index has a coefficient of -0.0123, suggesting that as voting becomes harder (higher cost index), voter turnout decreases slightly.
    – The State Population has a coefficient of -6.35e-10, indicating that population size has a negligible impact on voter turnout.
    – The P-value for Cost of Voting Index is 0.077, which means it’s not statistically significant at the typical 0.05 threshold but is close. This suggests a possible weak relationship between the difficulty of voting and turnout.
    – The P-value for State Population is 0.591, meaning there is no significant relationship between state population and voter turnout.

    References

    References
    ↑1Pomante II, Michael J. (2024). “The Cost of Voting in the American States: 2024.” Election Law Journal: Rules, Politics, and Policy. DOI: 10.1089/elj.2024.0037. Available from https://costofvotingindex.com/publications
  • Trump Transcripts

    Trump Transcripts

    A collection of Donald J. Trump transcripts. (more…)

  • Understanding Consumer Debt Service Payments as a Percent of Disposable Personal Income

    Understanding Consumer Debt Service Payments as a Percent of Disposable Personal Income

    Consumer Debt Service Payments as a Percent of Disposable Personal Income is a financial metric that measures the proportion of after-tax income that households are using to pay off their debts. (more…)

  • Adult Literacy and Numeracy in Counties

    Adult Literacy and Numeracy in Counties

    Examining literacy and numeracy based on political preference. (more…)

  • Senate Intelligence Committee Reports on Russian Active Measures Campaigns and Interference in the 2016 U.S. Election

    Senate Intelligence Committee Reports on Russian Active Measures Campaigns and Interference in the 2016 U.S. Election

    These documents have been processed with Adobe Acrobat to produce searchable text. A valuable summary is at A Collusion Reading Diary: What Did the Senate Intelligence Committee Find?
    (more…)

  • Gasoline Prices

    Gasoline Prices

    This is an interactive chart. (more…)