Metodi di ricerca relazioni internazionali Metodi di ricerca

  • Slides: 39
Download presentation
Metodi di ricerca: relazioni internazionali

Metodi di ricerca: relazioni internazionali

Metodi di ricerca Six Steps in the Research Process 1 The formation of a

Metodi di ricerca Six Steps in the Research Process 1 The formation of a theory (research question: basic or applied? ) Although some research questions can serve both purposes, we frequently have to choose which goal we want to pursue. For example, should we study the determinants of aggression under conditions of stress in order to develop a predictive model of human behavior, or should we instead focus on the reasons ethnic conflicts occur in a specific nation and investigate ways to prevent them?

 A Simple Example Let us imagine that in the middle of the desert

A Simple Example Let us imagine that in the middle of the desert there is a town called Little America that consists exclusively of several miles of service stations and restaurants. 1 One can do nothing in Little America except eat and buy gas. Now, suppose we have decided to study the voting behavior of Little Americans in presidential elections so that we may explain why one person votes Democratic while another votes Republican. In this greatly simplified example, the subjects of our analysis (residents of Little America) differ from one another in only two ways besides their voting preferences: each is either an owner or a worker, and each is associated with either a service station or a restaurant. Each of these factors, which political scientists term variables, represents a characteristic of a particular individual. One citizen of Little

America might be (1) an employee of (2) a restaurant (3) who votes Democratic,

America might be (1) an employee of (2) a restaurant (3) who votes Democratic, whereas another is (1) an owner of (2) a service station (3) who votes Republican. We wish to explain differences in voting behavior in terms of other differences among the voters. In this instance, we have only two to choose from: employee or owner status and service station or restaurant affiliation. We can refer to these, respectively, as socioeconomic status (SES)—with ownership representing higher status than employment—and business affiliation. Is there any reason to expect that knowing either characteristic of a particular person will help us predict his or her voting preference? To answer that question, we must do two things. First, we must ask ourselves: Is there any logical reason to expect either of these factors to influence voting behavior? Second, we must consult the political science literature: Is there in previous studies any empirical evidence that one or the other of these factors influences voting behavior?

Metodi di ricerca 2 The operationalization of that theory Operationalization is the conversion of

Metodi di ricerca 2 The operationalization of that theory Operationalization is the conversion of our relatively abstract theoretical ideas into concrete terms that will allow us actually to measure the concepts that are of interest to us. Operationalization involves moving from the conceptual level of thinking about a problem to the practical level of making the observations we need to solve it. It requires highly concrete and specific thinking.

Metodi di ricerca 3 The selection of research techniques and development of a research

Metodi di ricerca 3 The selection of research techniques and development of a research plan First, we must select appropriate measures and we must take these measurements in ways that are consistent with the concepts we are investigating. We cannot, for example, measure public opinion by analyzing newspaper coverage of a given election, because newspaper content may reflect the views of an editor or of those few people whose letters to that editor are published, without necessarily reflecting the views of most voters.

 Second, we must choose the most appropriate type of data to study our

Second, we must choose the most appropriate type of data to study our theory. Sometimes we obtain numerical data and other times we gather non-numerical data. Quantitative and qualitative data each have different uses in scientific research. Third, we must consider the feasibility of different options. Before we go out into the real world to take measurements, we must be sure that whatever method or technique we select can be employed properly under the particular set of conditions we are likely to face. For example, because there is no newspaper in Little America (only service stations and restaurants), we cannot use content analysis, even if we want to. To summarize, we must find a way to measure those variables we wish to measure that will be (1) consistent with our working definitions of the variables and (2) practicable.

Metodi di ricerca 4 The observation of behavior and collection of data Generalizability refers

Metodi di ricerca 4 The observation of behavior and collection of data Generalizability refers to the logical justification for extending our conclusions from the observed behavior of a few cases to the presumed behavior of an entire population. It is a concern we must take into account in selecting the particular cases (people, organizations, nations, etc. ) that we wish to study. The problem here is basically one of scale and resources. Reactivity. Once we have selected our cases for analysis, we must exercise great care in observing them. We must avoid measuring political phenomena or behavior in ways that encourage reactivity—a situation in which either the person who is doing a study or the methods used in the study somehow alter the way those under observation behave or think. In other words, a danger exists that the act of observation may itself cause those being observed to change their behavior or perceptions so that the results of the observation are misleading

Metodi di ricerca 5 The analysis of data First, is there some association between

Metodi di ricerca 5 The analysis of data First, is there some association between the behavior we are hoping to explain and the factors we think will help us to do so? Second, what is the nature of any relationship we discover? Third, how sure can we be that this relationship is real? We might find in examining our data, for instance, that less educated people tend to vote about as often as more educated people and that knowing a person’s level of education does not help us to predict or explain differences in the likelihood of voting. If this is the case, we say that there is no association between the two variables. If, on the other hand, we discover that seven times out of ten, knowing the level of education does allow us to predict accurately the likelihood of a person’s voting, this constitutes evidence supporting our expectation that the two variables are related. It tells us that more educated people are systematically different from less educated people when it comes to voting. The first thing to look for in assessing a hypothesis, then, is whether the two variables are statistically related.

Metodi di ricerca 6 The interpretation of the results What answer have we discovered?

Metodi di ricerca 6 The interpretation of the results What answer have we discovered? What is the substantive importance or practical utility of our findings? In essence, we have by this time reduced some aspect of political behavior to a set of detailed and descriptive qualitative data, or numerical quantitative data. Now, though, we must decide what our data tell us about our research question and the real world events they represent.

Metodi di ricerca Ricerca Descrittiva vs Esplicativa -La semplice raccolta di fatti significativi è

Metodi di ricerca Ricerca Descrittiva vs Esplicativa -La semplice raccolta di fatti significativi è ‘ricerca descrittiva’ -La spiegazione delle cause di fatti, test di teorie, cause/effetti nel mondo è ‘ricerca esplicativa’

Metodi di ricerca Teorie induttive: generalizzazione tra l’osservabile e il non osservabile - Teorie

Metodi di ricerca Teorie induttive: generalizzazione tra l’osservabile e il non osservabile - Teorie fondate empiricamente i. e. l’osservazione ripetuta che le porte dell’ascensore si aprono al premere del pulsante genera l’idea che la pressione del pulsante CAUSA l’apertura delle porte (meccanismi di causazione). Teorie deduttive: sono i processi che ci permettono di utilizzare teorie per speigare gli eventi del mondo reale. i. e. la generalizzazione induttiva del bottone/apertura porte ascensore è utilizzata per dedurre/predire la possibilità di ingresso nell’ascensore.

Metodi di ricerca

Metodi di ricerca

Metodi di ricerca Induzione

Metodi di ricerca Induzione

Metodi di ricerca Processi deduttivi

Metodi di ricerca Processi deduttivi

Metodi di ricerca Ricerca quantitativa e qualitativa

Metodi di ricerca Ricerca quantitativa e qualitativa

Metodi di ricerca

Metodi di ricerca

Metodi di ricerca Operazionalizzazione dei concetti: dall’astratto al concreto Si supponga che si voglia

Metodi di ricerca Operazionalizzazione dei concetti: dall’astratto al concreto Si supponga che si voglia testare l’efficacia di un concime sulla crescita del grano. Crescita - Altezza – misurazione in cm del grano (indicatore)

Metodi di ricerca

Metodi di ricerca

Metodi di ricerca A. accuratezza dell’ indicatore di operazionalizzazione rispetto all’ipotesi da misurare B.

Metodi di ricerca A. accuratezza dell’ indicatore di operazionalizzazione rispetto all’ipotesi da misurare B. inevitabile perdita parziale di senso dell’ipotesi Utilizzo di categorie/indicatori mutuamente esclusivi : i. e. Repubblicani vs Democratici ma non liberali, conservatori, Repubblicani , Democratici (si possono avere combinazioni di conservatori Democratici etc. ) C. e se vi fossero molteplici indicatori per la misurazione di una singola ipotesi?

Metodi di ricerca

Metodi di ricerca

Metodi di ricerca Oltre ad una teoria/ipotesi è necessario specificare anche perché i criteri

Metodi di ricerca Oltre ad una teoria/ipotesi è necessario specificare anche perché i criteri adottati sono rilevanti per provare la teoria:

Metodi di ricerca Internal Versus External Validity Two primary types of validity are associated

Metodi di ricerca Internal Versus External Validity Two primary types of validity are associated with empirical research: internal and external. Internal validity involves accurate measurement of our theoretical concepts. It requires asking if we are measuring what we think we are measuring. A later section in this chapter examines types of validation, which seek to answer this question. External validity pertains to the generalizability of our results. Can we reasonably expect to find the same causal influences at work in other settings? Does this study tell us anything about people, governments, and situations not included in it? A field experiment on the effects on driving habits of adding a dollar-a-gallon surcharge to the price of gasoline, for example, has little external validity if it is conducted in a community where the average family’s annual income is above $100, 000, because we cannot expect middle- and lowincome people to react to increased prices in the same way as upper-class people.

Metodi di ricerca

Metodi di ricerca

Metodi di ricerca Causazione: To begin, we need to recognize that we can almost

Metodi di ricerca Causazione: To begin, we need to recognize that we can almost never see causation directly. Rather, we have to infer its existence on the basis of observation and theory. A medical example will help explain this. In most circumstances, we can’t physically observe a specific chemical causing cancer in the human body. However, if properly structured research shows that people who are exposed to a given synthetic chemical are statistically significantly (a concept you will explore later in this book) more likely to get a specific kind of cancer than similar people who are never exposed to the chemical AND we have a theory that identifies a mechanism by which that chemical could alter human cells to cause cancer, we will have a sound logical basis for inferring that the chemical is causing cancer.

Metodi di ricerca Causazione come nozione probabilistica (non assoluta) In the social sciences we

Metodi di ricerca Causazione come nozione probabilistica (non assoluta) In the social sciences we almost never find cases of absolute causation. What we find instead is tendencies for relationships to exist. For example, although we can say (based on extensive observation) that people with higher levels of education are more likely to be politically active, we cannot know with certainty that sending a given individual to college will result in their being more politically active than they would have been otherwise. Therefore, social scientists have to talk about causation in probabilistic (not absolute) terms. (Natural scientists often have to do this as well, so the distinction between the two types of science is not absolute. )

Metodi di ricerca Causazione multipla I fenomeni sociali hanno normalmente causazioni multiple. Correlation is

Metodi di ricerca Causazione multipla I fenomeni sociali hanno normalmente causazioni multiple. Correlation is not causation! To understand the need for attention to logical warrants, it is essential to recognize that any observed relationship between two variables can be misleading. The fact that two variables are statistically correlated (meaning that they change together) is not evidence that one causes the other. Correlation does not, by itself, prove causation because changes in both variables could be caused by changes in a third variable or by some other process. To justify the conclusion that one thing causes another, we have to construct our research to give us some way to rule out other possible explanations of why the two variables are correlated.

Metodi di ricerca Esperimento (causazione) The experiment is the classic model of establishing causation

Metodi di ricerca Esperimento (causazione) The experiment is the classic model of establishing causation in science. It is based on a hypothesis that changes in the value of one variable cause changes in the value of another variable (for example, seeing pictures of the effects of auto accidents on the human body causes people to buckle their seat belts more frequently). The experiment allows us to test this hypothesis by exposing individuals—referred to as subjects in experimental research—who manifest the dependent variable (drivers, in the earlier example) to the independent variable or stimulus (photos, in the earlier example) under conditions that allow us to be relatively sure that any observed change in the dependent variable is a result of changes in the independent variable, and only of changes in the independent variable.

Metodi di ricerca The basic experimental design involves (1) an experimental group composed of

Metodi di ricerca The basic experimental design involves (1) an experimental group composed of subjects who will be exposed to the stimulus; (2) a control group of subjects, who will not be exposed to the stimulus; and (3) the random assignment of subjects to each group. The value of the dependent variable in each group is measured prior to introduction of the stimulus in what is called a pre-test, and is again measured after the experimental group has been exposed to the stimulus in a post-test. The impact of the stimulus (independent variable) is inferred from a comparison of the pretest and posttest scores for each group. The greater the difference in values between pretest and posttest for the two groups, the greater the effect attributed to the independent variable.

Metodi di ricerca Who, what, where and when: il problema del campione/sampling Sampling: Sampling

Metodi di ricerca Who, what, where and when: il problema del campione/sampling Sampling: Sampling is the act of selecting a small portion, or sample, of a larger group. If chosen carefully enough, the sample should reflect the characteristics of the population from which it is drawn and allow us to draw conclusions about the bigger group. A sample is any subgroup of a population of cases that is identified for analysis. If we want to study and reach conclusions about the decision-making behavior of state legislatures, for instance, we might do so by examining decision making in the legislatures of Oregon and Washington rather than in all 50 states, and from these we might generalize our findings to the larger population of U. S. state legislatures.

Metodi di ricerca Non tutti i campioni/sampling, sono rappresentativi. Esistono diverse forme di raccolta

Metodi di ricerca Non tutti i campioni/sampling, sono rappresentativi. Esistono diverse forme di raccolta di campioni: Probability sampling involves randomly selecting cases from a population so that the group selected as a sample is representative of the population from which it was drawn. There are four basic types of probability samples: (1) random samples, (2) systematic random samples, (3) cluster or multistage random area samples, and (4) stratified samples.

Metodi di ricerca 1 -Random Samples The guiding principle underlying probability sampling is randomization.

Metodi di ricerca 1 -Random Samples The guiding principle underlying probability sampling is randomization. A sample is said to be a random sample (sometimes referred to as simple random sample) if two conditions are met. First, the sample must be chosen in such a manner that each and every individual or case in the entire population has an equal opportunity to be selected for analysis. Second, the sample must be chosen in such a manner that each and every possible combination of n cases, where n is simply the number of cases in the sample, has an equal opportunity to be selected for analysis.

Metodi di ricerca 2 -Sampling sistemico One common variation called the systematic random sample

Metodi di ricerca 2 -Sampling sistemico One common variation called the systematic random sample is used when we want to study a relatively large population whose members are individually listed in some central location, such as a telephone book, a student directory, a list of registered voters, or a membership roster.

Metodi di ricerca 3 -The Multistage Process We begin with a map of the

Metodi di ricerca 3 -The Multistage Process We begin with a map of the United States, which we divide into a large number of equally populated areas (called primary sampling units, or PSUs). The federal government has already made this division in the form of 435 congressional districts, each populated with just more than half a million people. We assign a number between 1 and 435 to each district and, using a random number table, select several districts for analysis. The exact number selected is a function of both the ultimate size of the sample to be drawn and the available resources, but in general, the more districts selected, the more accurate the sample will be. This reveals the principal cost saving of the multistage random area technique, for rather than having to track down respondents all over the country, we can focus our attention (and money) on a relatively few geographically defined areas.

Metodi di ricerca Stratified Random Samples The technique is known as stratified sampling is

Metodi di ricerca Stratified Random Samples The technique is known as stratified sampling is less a selection procedure than a strategy to supplement other approaches. It is used primarily when we wish to study in detail a population subgroup that is so small that a random sample will include too few members of that subgroup to permit detailed analysis. Suppose, for example, that we begin with the hypothesis that presidents are more open with the news media during the first two months of their administration (often termed a honeymoon period) than at all later times and that we wish to test the hypothesis by analyzing the content of the transcripts of presidential news conferences. Suppose further that we are able to identify 500 such transcripts for a given period, of which only 25 (or 1 in 20) are news conferences from the honeymoon period, and that we wish to sample 100 of these conferences. In this instance either a simple or a systematic random sample would likely include approximately 5 honeymoon-period transcripts and approximately 95 posthoneymoon transcripts. The very small number of the former makes a meaningful comparison very difficult, because it may provide too few examples to reflect accurately the range of presidential responses to reporters’ questions.

Metodi di ricerca Quantitave methods Qualitative methods

Metodi di ricerca Quantitave methods Qualitative methods

Metodi di ricerca

Metodi di ricerca

Metodi di ricerca

Metodi di ricerca