Chapter 1 - INTRODUCTION

This course focuses on introductory quantitative methods that are useful to managers and public administrators. We will cover statistics, decision analysis, data collection and management, computer graphing and linear regression analysis. We will look at the quantitative procedures that are most useful to you in your careers and also methods that may be useful for designing and carrying out an applied research project.

Statistics and statistical analysis can be very important to public agencies and public administrators and various constituencies -

For example:

Police Departments - crime rates, trend, comparison to national/state average, problem areas, etc. Policy Issue: Budget.

Health Agencies - do we need more/fewer hospitals?

National economic issues; inflation, unemployment rates, consumer confidence. Vouchers/charter schools/school choice - what effects will these programs have. Do students with vouchers do better academically? Statistically significantly better?

Affirmative action issues - relies on statistics, etc.

U.S. Census, Polling, etc.

Point: Many policy issues/decisions, even court cases, are based on statistics. Statistics can be extremely important in many issues.

And many times, statistics and data gathering are very important: Pressure to collect data.

Examples -
1. How many miles of highway were built last year.
2. How many mental health clients were treated last year.
3. How many people in Michigan have the AIDS virus.
4. What is the graduation rate at Michigan high schools.

Specific examples in textbook of using Quantitative Methods:

1. United Way agency wants to evaluate its programs in terms of community approval and to see if there are unmet community needs.  It will contract out the evaluation study, but the managers/administrators need to be aware of the statistical and quantitative issues and techniques involved.  Method will be survey questionnaires.

2. Many states and cities set up economic development programs to provide tax incentives for businesses to relocate.  You need to do a Cost-Benefit analysis to determine the desirability of these programs.

3. Evaluation of a job training program. We will discuss later.

Understanding statistics and statistical analysis is a valuable skill for many positions, perhaps mandatory. There are various levels of skills that may be specific to various positions.  Some jobs require data collection skills, some jobs requires performing statistical analysis, some jobs require being able to explain and present statistical outcomes, some jobs require that you can understand statistical analysis, etc.

Even high level public jobs like the mayor or the governor or the president require some understanding of statistics.  Mayor, governor, CEO, may not know how to perform statistical analysis, but they need to know how to understand and interpret the results.

And even for people who are not public administrators, it is important for us in general as citizens and voters to be educated, sophisticated consumers of statistics. Being able to understand statistics, as presented in the newspapers and in policy debates, is important to being a good citizen.

Main point of class: Become familiar with statistics, so that you are an educated consumer of statistical information. You may or may not need to actually perform statistical analysis in the future, but you will always have a need to understand statistical results..... And you should develop a healthy skepticism of statistics.

Example: Saying that women earn 77% of what men earn on average, as the result of sex discrimination, violates the "ceteris paribus" assumption.

Policy Evaluation - book describes the older, outdated method of evaluating public programs by evaluating the PROCESS.  Statistics were gathered that focused on the process.

University accreditation would look at: size of library, percentage of faculty holding Ph.D., number of programs, student/faculty ratio, etc.

Now the states and federal governments, and private funders of social agencies generally require program/policy evaluations that focus on OUTCOMES.

The focus is not on how many resources are used, or how the program operates, but whether the program meets its goals.

1. Do stronger drunk driving laws result in fewer traffic fatalities?
2. Does the death penalty reduce the murder rate? (ceteris paribus)
3. Do higher cigarette taxes reduce smoking related health costs and death?
4. Does a drug program help clients become drug free?
5. Does AFDC affect the rate of teenage illegitimacy?

Congress now requires program evaluation based on outcome. And the appropriate approach to program/policy evaluation is to use the scientific method, as it is used for social sciences, which involves statistical methods.

To do program evaluation properly, we have to use a consistent, logical, scientific approach.  There may be various approaches, but we use a general framework of statistical methods to properly evaluate a policy or program.

Social sciences over the years have adapted the scientific method to the specific issues facing sociologists, economists, anthropologists, historians, etc.  It can't be haphazard, it has to be based on the scientific method.  However, we can't conduct controlled experiments like in chemistry or medicine/hard sciences, but we have developed standard procedures for performing scientific inquiry into the social sciences.

Social science method involves 4 steps:

1. Form a testable hypothesis
2. Collect relevant data to test hypothesis
3. Perform statistical analysis.
4. Interpret the results.

Example: You are trying to evaluate the success of a workfare program that has been in place for two years, it is designed to give job training to welfare recipients.

Did the program succeed in helping get off welfare and find a job?

The testable hypothesis would be something like:

"People in the workfare program find a job faster and keep a job longer than those who are not in the program."

Tricky issues are:

1. Most people on welfare find a job eventually, but we want to know if the workfare program made a statistically significant difference! Even if people in the program got jobs at a faster rate, we want to know if the difference is statistically significant.

2. We have to carefully compare two different groups - those who didn't go through the workfare program and found a job, and those who went through workfare and found a job. Ceteris paribus.

Two of the most important concepts in statistics:

1. Ceteris paribus - research design issue, holding all variables constant, or controlling for all other potential effects, to isolate the independent effects of X on Y.

Example:  After controlling for years of continuous work experience, age, race, education, union status, number of hours worked, etc., we find that women make 4% more/less than men.

2. Statistical significance - interpretation of results.
Example 1:
The 4% difference in salaries is not statistically significant.
Example 2
: The difference in salaries is statistically significant at the 1% level (or 5% or 10% level)