Unit 5: Quantitative Methods

Why Do We Use Quantitative Methods?

Quantitative research, or also called scientific research method, is a way of obtaining information that provides the researcher with accurate and reliable data that is numeric in nature and has been obtained through testing or comparing ideas in the public arena. Quantitative research is based on observation and the measurement of repeated incidences of a political phenomenon.

Quantitative research enables to describe populations and phenomena in a detailed tradition based on figures or data, and enables the political scientist to make propositions out of relationships between two or more variables. Observation and experimentation are the strategies that can be employed to test an assumption or a theory. Observation is the most used strategy by political scientists for international relations, or other social sciences while experimentation is not that commonly used due to its nature of testing. Consequently, experimental research earns to be explained thoroughly and rigorously due to its complex nature of working with humans and the many social, cultural, anthropological, economic, political or historical factors that need to be considered in designing experimental research for social sciences.

Descriptive Research

Descriptive studies are about describing a particular situation or case as detailed and elaborated as possible, and it enables the classification and organization of taxonomic categories as in survey methods. Political scientists want to analyze whole populations, because with large numbers the social scientists can make generalizations about the empirical world. The larger the cases or subjects are, the more accurately it will be possible for the scientists to make inferences about the phenomenon they observe.

Associational Research

Associational research enables for understanding or identification of relations between certain variables. For a political scientist, for example, the researcher may like to learn in a study related to gender issues, if compulsory pre-school education in villages may have an impact on girls’ higher participation in higher education. In such cases the researcher can make statistical analysis to predict the relationships through correlational or causalcomparative methods, or make use of available large datasets that are accessible.

Correlational Research

Correlational studies enable the researcher to identify how one or a set of variables are related to another or enable the researcher to make predictions. Especially in the case where it is difficult to conduct an experimental study, the researcher can make use of the available data to analyze the relationships among different variables. Correlational research contradicts with experimental research in the sense that no variable is manipulated in this method. Thus, correlation should not be confused with causal relation, it only provides an existing or non-existing relationship between variables.

Causal-Comparative Research

Causal comparative research attempts to determine the cause or consequences of differences that already exist between identified set of subjects.

Experimental Research

Experimental research is unique yet not so common in political research as the nature of experimental research demands manipulating of the independent variable in the experiment. More explicitly, it is a research type that explicitly attempts to influence a particular variable. It is considered to be the best method to test hypothesis related to cause-and-effect relationships. Nevertheless, policy studies mostly cover huge sample sizes and it is very difficult to isolate participants from their social or cultural contexts to and it becomes hard for a social researcher to interpret if the results are related vis-á-vis to the effect of the treatment or experiment or if there are other intruding aspects such as time or location on the outcomes of the study. In addition, ethics is of utmost consideration because any research that deals with humans as subjects, it is important to receive consent from an official research agency and thereafter from the individuals themselves.

Survey Research

Survey research provides opportunities to build theory through empirical data via understanding the attitudes, behavior, and dynamics in the society. It has become a common method of collecting descriptive data, and it also enables the social scientist to manipulate data and do more complex analysis to understand the relationship among variables through associational research.

A list of survey types that can be utilized in political science research are provided such as cross-national surveys, national time series, election surveys, and panel surveys.

Cross-Sectional Surveys

A popular strategy for doing survey research is crosssectional research, which means collecting data through a survey across a predetermined population at one point in time. Thus, the survey is administered to a representative sample at the same time.

Cross-National Surveys

Cross-national surveys provide opportunities to collect data that is administered in various countries so that the researchers can collect data regarding how knowledge, attitudes, and behavior many vary according to difference in democracy, culture, economic conditions through comparisons. A popular recent cross-national study is the 2016 International Civic and Citizenship Education Study (known for ICCS 2016).

National Time-Series

The national time series surveys generally are data collected over the years that are in the interest of political scientists. Indicators may vary from economic to social indicators, such as income levels, education level, employment or unemployment figures by education level, in or out migration trends within a nation. The measurements are taken in an organized manner such as in fixed periods over time: these can be days, weeks, months or multiyear cycles as a season or time cycle may bear history that may influence the respondents’ attitudes or behaviors.

Election Surveys

Elections data are useful not only for contemporary analyses but also for historical analyses. Election data provide opportunity to analyze voting behavior of individuals through combining this with ecological and contextual data to produce maps and graphs of political tendencies that explain individual behavior across geographical area and specific contexts. Face-to-face interviews have an advantage of producing high response rates but may be very expensive and time consuming. With the increase of phones, especially cellular phones, calls also have become popular trends in collecting data, yet the general preference is to send out a link to the online questionnaire by e-mail.

Panel Surveys

Political sciences, or international relations fields value policy decisions and the changing trends of individuals based on policy making decisions or applications, therefore through a panel study, the researcher can collect data from similar samples at different times. For instance, consider that a governing party is interested in understanding its citizens’ attitudes toward becoming a member of the European Union. After certain international trade decisions, and changes at political and economic level, the governing party may reach the same sample again to see if the attitudes of the citizens have changed over time. One major limitation of such a longitudinal study is that since there is a long period of time between the two survey administrations, the researcher is likely to lose some of his or her sample.

Preparing Good Survey Instruments

There are three distinguished features that identify surveys: a) they aim at reaching a random sample so that the researcher can generalize the findings to its population; b) they benefit from standardized closed-items so that it facilitates the measurement of attitudes and other characteristics of the respondents; and finally, c) they hold an ability to generate quantifiable data so that it can be exposed to statistical analysis. Also, it is underscored that the researcher needs to avoid four types of errors as cornerstones of good quality surveys.

• Coverage error, which refers to drawing samples form a list of irrelevant samples to represent the population;
• Sampling error, which exists any time when the researcher reaches some of the sample and not all members;
• Nonresponse error, which occurs when some particular units of analysis tend to answer some particular items which are likely to influence the estimate of the findings;
• Measurement error is described as the difference between the estimate produced and the true value in case the respondents may answer inaccurately to the survey questions.

Tips for Writing Survey Items

• Each question should have only one single objective. ‘I volunteer at governmental elections and NGOs’ Note, what if the participant only volunteers in one institution!
• Adopt the same definitions or concepts throughout the survey. For instance, if your key term is university students, keep it like that, and do not alternate it with college students in different parts of the survey.
• Avoid leading items that imply a preferred way to answer.
• Avoid rating within the questions such as ‘I always attend student protests.’

1) Never 2) Rarely 3) Sometimes 4) Frequently 5) Always

• Avoid loaded questions, especially during faceto-face interview survey, which may emotionally charge the respondent.
• Use questions that are simple and easy to understand, and cause no misunderstanding.
• Avoid using technical jargon. It may discourage the individuals to continue with the survey.
• Avoid negative or double negative expressions such as ‘I do not like the politicians,’ or ‘I do not dislike politics.’
• Be precise, avoid estimations such as ‘Estimated how many students are there in your class?’ ask ‘What size is your class.’

Data Collection Procedures

In addition to underscoring that the quality of the survey instrument itself is essential, there are different techniques to collect survey data. One of the earliest, yet most expensive approach could be the face-to-face interviews which could also be done through the phone. It is advantageous in that it increases the likelihood for participation. In addition to the interview technique, survey data may be collected through questionnaires or standardized interviews through mail, or online platforms such as the Survey Monkey or other software such as Qualtrics. Currently there exists rising interest in virtual platforms such as Twitter and other virtual means, however, there is a risk that these virtual environments may not provide the researcher with the expected return rates.

Data Analysis Procedures

The researcher should have some basic knowledge about statistics before reading empirical research or have a desire to conduct research in a field of interest and suggested that the researcher be equipped with the essential skills in statistical analysis. One advantage of such statistical knowledge facilitates political scientists to aggregate information from large numbers of cases, for instance, to understand individual attitudes, beliefs, voting or other behavior and extract even basic summary statistics from such a mass of data revealing unsophisticated analysis such as the sample mean, percentages or standard deviations as valuable, and such knowledge is obviously a great opportunity and should not be overlooked.

One easy way of calculating descriptive statistics is using tools such as the Microsoft Excel, or SPSS software data analysis packages. Through using such tools, the researcher or political scientist can easily convert the data collected from the online surveys to the program or manually insert the data collected through interviews, phone calls, or filled out questionnaires. Through selecting the proper analysis boxes, the researcher can summarize big datasets in a press on a button period to obtain summaries that facilitate understanding demographic information about the participants. For analysists using figures with visual reports such as tables, charts, or scatter boxes may appear more comprehensible as the reader can observe the magnitude or the differences of the figures on those tables or charts. On the other hand, use of causal data will enable the researcher to find out the relationships between the variables. In doing data analysis based on correlational research, the researcher must be careful and avoid understanding or reporting cause-and-effect relationships.

Basics of Doing Research

Regardless if the researcher is aiming at doing descriptive research or correlational research, there are some basics of doing scientific research, which relates to asking for research questions or stating hypothesis if the design selected requires one. Next, it is also important to make sure that the data collected and analyzed is scientifically valid and reliable.

Writing Research or Hypothesis

To find scientific evidence for the phenomenon or problem areas revealed by the politician, the policy analysist, or the economist, one needs to test the hypothesis in a systematical way and offer reliable data to accept or refute the hypothesis we stated. The scientific findings we reach at in turn will provide implications for the improvement of political science and international relations.

Good Research Questions

• Research questions should be feasible so that there is available time, energy, and even sufficient budget to investigate them.
• They should be clear so that they do not give way to ambiguity.
• Research questions should be significant, in other words, they should be worth to investigate.
• Research questions should be ethical, indicating that any research should harm no human being or living being, or the natural environment.
• Research questions are likely to suggest, but not always, a relationship to be investigated.
• Good quantitative research questions should be specific.
• Finally, good research questions should be measurable; and capable of conducting rigorous statistical analysis.

Variables

A variable is a concept that may refer to any variation in a context. For example, artifacts used in elections can be a variable. Gender, education, nationality are all nouns that can be counted as variables. For instance, let us use the example of a voting box. If the votes in the box vary from rightist parties to liberal and leftist parties, we call them variables. Nevertheless, if all the votes are representing the same party in the box we call them ‘constants’. On the other hand, variables may also refer to measures such as motivation, happiness, and well-being. In order to measure or manipulate a variable, we need to define it specifically.

Variables can be defined as ‘quantitative variables’ and ‘categorical variables., Quantitative variables are mostly able to be divided into smaller chunks such as weight in kilograms, or height in centimeters, or age in numbers, or as magnitude of interest represented in figures like 1 for little interest or 5 for high interest, or 1 for disagreement and 5 for agreement. All those ratings hold a continuum in themselves and therefore are called quantitative. Categorical variables, on the other hand, are not varying in degrees, or amounts or quantity, they rather are part of a category such as gender, one is either male or female; or marital status, married, single, divorced, or other, or such as nationality, or school type, either public school or private school.

In order to prepare good research questions or identify hypotheses it is essential to understand the difference between independent and dependent variables. Through an independent variable, a researcher can study its effect on one or more variables. Thus how the independent variable affects the dependent variable is studied.

Census, Population, and Sampling

The individual or the sample is the unit of analysis in a study. While samples are only about a part of a population, a census tries to include the entire population. Since it is not feasible and fit to reach the entire population (or target population) such as in a census, the sampling process is a crucial process to select individuals who will become representative of the larger group, in other words, the actual or target population. In this case, it is essential that you determine the sample size.

What is the sample size that you need to consider to generalize your findings? Although there is no strict rule behind what percentage is needed, the researcher may draw his or her own limitation and decide that he or she is likely to reach 10% of the population. A simple formula is suggested for decision-making by dividing the population by the desired sample size. For instance, assume you have a population of 1000 individuals, and decided to represent a 10% of your population, the calculation would be selecting randomly 100 individuals to be included in your sample.

Random Sampling

‘ Random sampling’ is a strategy that enables every member in the population to become selected and be a part of the study. Random sampling is essential as it avoids the likelihood to select a biased sample. ‘Random sampling’ is the most reliable strategy to identify your sample because it enables every member in the population to become selected at an equal chance and be a part of the study. Consequently, the findings of the study that are based on random sampling can be generalized to its population.

Stratified Random Sampling

In stratified random sampling the researcher has the opportunity to exploit any available information on the population to improve the efficiency of the sample in case they know the population characteristics and can divide them into groups or strata such as male and female, or age groups, political orientations and other. Stratified sampling strategy is one of the most preferred methods in opinion studies in political science research as it ensures representativeness of the population studied.

Systematic Sampling

Systematic sampling , on the other hand, is the selection of a sample systematically from the population- it belongs to the non-random sampling strategy.

Cluster Sampling

There are occasions in which it is not feasible to conduct a random assignment of a number of individuals from a population, the reasons may be multiple, for instance, the list of individuals may not be obtainable, or it is not feasible to reach out the random assigned individuals as it may cause ethical considerations, loss of energy and time. In such cases cluster sampling may be the best solution to adopt as the sampling strategy. As the word itself indicates, cluster sampling deals with the assignment of the subjects (or individuals) not one to one, but based on each cluster identified. In case the researcher wants to generalize the findings to the population, he or she can select the clusters randomly and the sampling method will be called cluster random sampling .

Two-stage Random Sampling

In the case you implement a two-stage sampling method, then you may consider the cluster random sampling method explained above and randomly select the subjects or individuals randomly from those randomly selected clusters.

Purposive Sampling

‘Purposive sampling’ method is employed when the researcher intentionally is interested in a population with certain knowledge, experience, or history.

Convenience Sampling

In the cases that it is not feasible or fit to conduct a random sampling as suggested earlier, ‘convenience sampling’ method can be implemented, nevertheless, it should be highlighted that convenient sampling does not enable the researcher to generalize the findings from the sample to the targeted population and it is the least suggested sampling strategy

Validity and Reliability

Any quantitative research conducted needs to be valid and reliable. Validity refers to the correctness, appropriateness, meaningfulness, and usefulness of the data collected or the relevancy of the data collection instrument.

‘Reliability’ is about the consistency of the scores obtained. For instance, if we have both open-ended items and close-ended items in an instrument that measures political tendencies of subjects, the responses need to be consisted in both methods used to check for the consistency in responses. In addition, data may not be reliable based on how it is collected. Consider that in a survey the respondents are asked for their political tendencies, in such occasion the individual may not feel comfortable, especially, if the item includes controversial issues, the respondent may feel reluctant in answering his or her actual feelings or attitudes. Ultimately, the data may not be reliable.