#### DMCA

## Making the most of statistical analyses: Improving interpretation and presentation (2000)

### Cached

### Download Links

Venue: | American Journal of Political Science |

Citations: | 549 - 24 self |

### Citations

2114 | Bayesian data analysis - Gelman, Carlin, et al. - 2003 |

1338 |
Regression models for categorical and limited dependent variables
- Long
- 1997
(Show Context)
Citation Context ...llary parameters, researchers should calculate and present quantities of direct substantive interest. Some researchers go a step farther by computing derivatives, fitted values and first differences (=-=Long 1997-=-; King, 1989: x5.2) which do convey numerically precise estimates of interesting quantities and require little specialized knowledge to understand. Even these approaches are inadequate, however, becau... |

463 |
Bayes and Empirical Bayes Methods for Data Analysis
- Carlin, Louis
- 2000
(Show Context)
Citation Context ...ul than our algorithms because they allow researchers to draw from the exact finite-sample distribution, instead of relying on the central limit theorem to justify an asymptotic normal approximation (=-=Carlin and Louis 1996-=-). Unfortunately these methods are difficult to use, particularly since statisticians still disagree about criteria for determining when a Markov chain has converged in distribution to the true poster... |

364 | Markov chain Monte Carlo convergence diagnostics: a comprehensive review
- Cowles, Carlin
- 1996
(Show Context)
Citation Context ...fortunately these methods are difficult to use, particularly since statisticians still disagree about criteria for determining when a Markov chain has converged in distribution to the true posterior (=-=Cowles and Carlin, 1996-=-; Kass et al., 1997). We welcome research that implements these advanced techniques, but in this paper we recommend a simpler method that should prove more accessible to a wide range of social scienti... |

225 |
Tools for Statistical Inference: Methods for Exploration of Posterior Distributions and Likelihood Functions
- Tanner
- 1993
(Show Context)
Citation Context ...ainties. Simulation can also help researchers understand the entire statistical model, take full advantage of the parameter estimates, and convey findings in a reader-friendly manner (see Fair, 1980; =-=Tanner, 1996-=-; Stern, 1997). 3.1 What is statistical simulation? Statistical simulation uses the logic of survey sampling to approximate complicated mathematical calculations. In survey research, we learn about a ... |

221 |
Bootstrapping: A nonparametric approach to statistical inference
- Mooney, Duval
- 1993
(Show Context)
Citation Context ...ible to a wide range of social scientists. Another useful alternative is bootstrapping, a non-parametric approach that relies on the logic of resampling to approximate the distribution of parameters (=-=Mooney and Duval, 1993-=-; Mooney 1996). In theory, the sampling distribution ofsfl can be viewed as a smoothed histogram of an infinite number ofsfl's, each estimated from a different sample of size n from a given population... |

163 |
Computer-intensive Methods for Testing Hypotheses: An Introduction
- Noreen
- 1989
(Show Context)
Citation Context ...vor of simulation. First, there is a computerintensive alternative to nearly every analytical method of computing quantities of interest and conducting statistical tests, but the reverse is not true (=-=Noreen 1989-=-). Thus, simulation can provide accurate answers even when no analytical solutions exist. Second, computerintensive techniques like simulation enjoy a pedagogical advantage. Studies have shown that, n... |

149 |
Unifying Political Methodology: The Likelihood Theory of Statistical Inference
- King
- 1989
(Show Context)
Citation Context ...eters, researchers should calculate and present quantities of direct substantive interest. Some researchers go a step farther by computing derivatives, fitted values and first differences (Long 1997; =-=King, 1989-=-: x5.2) which do convey numerically precise estimates of interesting quantities and require little specialized knowledge to understand. Even these approaches are inadequate, however, because they igno... |

120 |
Markov chain Monte Carlo in practice: A roundtable discussion
- KASS, CARLIN, et al.
- 1998
(Show Context)
Citation Context ...ult to use, particularly since statisticians still disagree about appropriate criteria for determining when a Markov chain has converged in distribution to the true posterior (Cowles and Carlin 1996; =-=Kass et al. 1998-=-). Nonetheless, this field has shown remarkable progress over the last decade and is well worth monitoring by political scientists. Another useful alternative is bootstrapping, a nonparametric approac... |

94 |
Simulation-based estimation
- Stern
- 1997
(Show Context)
Citation Context ...ulation can also help researchers understand the entire statistical model, take full advantage of the parameter estimates, and convey findings in a reader-friendly manner (see Fair 1980; Tanner 1996; =-=Stern 1997-=-). What Is Statistical Simulation? Statistical simulation uses the logic of survey sampling to approximate complicated mathematical calculations. In survey research, we learn about a population by tak... |

88 | War and the Survival of Political Leaders: A Comparative Study of Regime Types and Political Accountability”. American Political Science Review - Mesquita, Bruce, et al. - 1995 |

81 |
Comparative Statistical Inference
- BARNETT
- 1982
(Show Context)
Citation Context ...eories (such as whether θ or ˆθ is the random variable) are important but need not concern us here, as our approach can usually be employed with any of these and most other theories of inference (see =-=Barnett 1982-=-). these is equivalent to the expected value ( ˆ Y ) in a linear regression, which we discuss in the following subsection. To simulate one predicted value, follow these steps: 1. Using the algorithm i... |

59 | A Statistical Model for Multiparty Electoral Data.” American Political Science Review 93:15–32
- Katz, King
- 1999
(Show Context)
Citation Context ...the bottom left) or weakening (for the "."'s in the middle), with other variables held constant at their means. Figure 3 displays our results. The figure is called a "ternary plot"=-= (see Miller, 1977; Katz and King, 1999-=-), and coordinates in the figure represent predicted fractions of the vote received by each candidate under a different simulated election outcome. Roughly speaking, the closer a point appears to one ... |

32 |
Problems in making policy inferences from the Coleman report
- Cain, Watts
- 1970
(Show Context)
Citation Context ...eria for meaningful statistical communication since, for many nonlinear models,sfi andsff are difficult to interpret and only indirectly related to the substantive issues that motivated the research (=-=Cain and Watts, 1970-=-; Blalock, 1967). Instead of publishing the effect coefficients and ancillary parameters, researchers should calculate and present quantities of direct substantive interest. Some researchers go a step... |

28 | Democratizing Mexico: Public Opinion and Electoral Choices - Dominguez, McCann - 1996 |

24 |
Resampling: The new statistics
- Simon
- 1992
(Show Context)
Citation Context ...inson and Shevokas, 1976). One scholar has even offered a $5,000 reward for anyone who can demonstrate the superiority of teaching analytical methods, but so far no one has managed to earn the prize (=-=Simon 1992-=-). Of course, there are advantages to understanding the mathematics underlying the delta method and 4 When g is linear there is obviously no need for a linearizing approximation; an exact analytical s... |

20 | Data Analysis for Politics and Policy - Tufte - 1974 |

13 |
Estimating the Expected Predictive Accuracy of Econometric Models
- Fair
- 1980
(Show Context)
Citation Context ...their uncertainties. Simulation can also help researchers understand the entire statistical model, take full advantage of the parameter estimates, and convey findings in a reader-friendly manner (see =-=Fair, 1980-=-; Tanner, 1996; Stern, 1997). 3.1 What is statistical simulation? Statistical simulation uses the logic of survey sampling to approximate complicated mathematical calculations. In survey research, we ... |

12 |
Bootstrap statistical inference: Examples and evaluations for political science
- Mooney
- 1996
(Show Context)
Citation Context ...social scientists. Another useful alternative is bootstrapping, a non-parametric approach that relies on the logic of resampling to approximate the distribution of parameters (Mooney and Duval, 1993; =-=Mooney 1996-=-). In theory, the sampling distribution ofsfl can be viewed as a smoothed histogram of an infinite number ofsfl's, each estimated from a different sample of size n from a given population. Bootstrappi... |

10 |
Electoral Dynamics in Britain since
- Miller
- 1978
(Show Context)
Citation Context ... the "o"'s in the bottom left) or weakening (for the "."'s in the middle), with other variables held constant at their means. Figure 3 displays our results. The figure is called a =-="ternary plot" (see Miller, 1977-=-; Katz and King, 1999), and coordinates in the figure represent predicted fractions of the vote received by each candidate under a different simulated election outcome. Roughly speaking, the closer a ... |

8 |
Causal inferences, closed populations, and measures of association
- Blalock
(Show Context)
Citation Context ...atistical communication since, for many nonlinear models,sfi andsff are difficult to interpret and only indirectly related to the substantive issues that motivated the research (Cain and Watts, 1970; =-=Blalock, 1967-=-). Instead of publishing the effect coefficients and ancillary parameters, researchers should calculate and present quantities of direct substantive interest. Some researchers go a step farther by com... |

7 | Probability and statistics: experimental results of a radically different teaching method - Simon, Atkinson, et al. - 1976 |

4 |
Comparative statistical inference. 2nd ed
- Barnett
- 1982
(Show Context)
Citation Context ...heories (such as whether ` ors` is the random variable) are important but need not concern us here, as our approach can usually be employed with any of these and most other theories of inference (see =-=Barnett, 1982-=-). Repeat the second step, say, M = 1000 times to obtain 1000 draws of the main and ancillary parameters. Appendix A explains how to make these random draws using a popular software package. If we kne... |

3 | Logistic regression in rare events data. Unpublished manuscript - King, Zeng - 1999 |

2 |
PCSE.G: A Gauss Procedure to Implement Panel-Corrected Standard-Errors in Non-Rectangular Data Sets
- Franzese
- 1996
(Show Context)
Citation Context ...t (1998, pp. 80-81) by only 0.3 percent, on average. Standard errors diverged by 6.8 percent, on average, apparently due to discrepancies in the method of calculating panel-corrected standard errors (=-=Franzese, 1996-=-). None of the differences made any substantive difference in the conclusions. Figure 2: Simulated Levels of Government Spending: These panels contain density estimates (smooth versions of histograms)... |

2 | Electoral Surprise and the Midterm Loss - Scheve, Tomz - 1999 |

1 |
Simulation-Based Esimation
- Stern
- 1997
(Show Context)
Citation Context ...ation can also help researchers understand the entire statistical model, take full advantage of the parameter estimates, and convey findings in a reader-friendly manner (see Fair, 1980; Tanner, 1996; =-=Stern, 1997-=-). 3.1 What is statistical simulation? Statistical simulation uses the logic of survey sampling to approximate complicated mathematical calculations. In survey research, we learn about a population by... |

1 |
Wiley. statistical interpretation and presentation 361
- Blalock
- 1967
(Show Context)
Citation Context ...atistical communication since, for many nonlinear models, ˆ β and ˆα are difficult to interpret and only indirectly related to the substantive issues that motivated the research (Cain and Watts 1970; =-=Blalock 1967-=-). Instead of publishing the effect coefficients and ancillary parameters, researchers should calculate and present quantities of direct substantive interest. Some researchers go a step farther by com... |