Research Methodology: Types, Designs, and Variables
The methodology involves consideration of the logical design and research process. It answers the question: How will the research be conducted?
Regarding the final report of the thesis (or work), the following points must be developed:
1. Types of Research
It is important to determine the type of research, considering the research question, the objectives, and hypotheses presented throughout the work, taking into account the scope of the collection and analysis of results. Although related, it should not be confused with the type of research design.
The type of research refers to the end, the main focus of the work. Instead, the design relates to the overall structure to achieve that purpose.
- Exploratory:
- Investigates problems (issues) that are little studied.
- Exploring from a new perspective.
- Helps identify promising concepts.
- Prepares the way for further studies.
- Descriptive:
- Defines and measures variables.
- Collects information on variables.
- Specifies properties, characteristics, and traits of an emerging phenomenon.
- Makes predictions.
1.3. Correlational Types:
- Associates variables through a predictable pattern for a group or population.
- Its purpose is to understand the relationship between two or more variables, concepts, categories, or variables in a particular context.
- Offers predictions.
- Explains the relationship between variables.
- Quantifies the relationship between variables.
1.4. Explanatory Types:
- Determines causes.
- Generates a sense of understanding.
- They are more structured, going beyond mere exploration, description, or listing of variables or phenomena.
2. Research Designs
It refers to the condition in which the investigator will interfere with the manipulation of reality, that is, the manipulation of variables under study. The research design refers to the plan or strategy designed to respond to research questions. It allows the researcher to point out what to do to achieve their study objectives, answer the questions that have arisen, and analyze the accuracy of the assumptions made.
In the social sciences, different classifications of designs can be found.
- 2.1 Non-Experimental: a) Transactional b) Longitudinal: b1: Trend, b2: Evolution, b3: Panel
- 2.2 Experimental: a) Pre-experimental, b) Quasi-experimental c) 2.3 Factorial Experimental
How is Design in particular? This will be set depending on:
The level of manipulation of independent variables, the number of independent and dependent variables, the degree of control of extraneous or intervening variables, and the type of sampling method.
It should be emphasized that the distinction between the independent variable (that which the researcher manipulates) and the dependent variable (which receives the independent effect) makes sense to apply only to explanatory studies where interest rate relationships are cause-effect.
What is it to manipulate one or more variables?
Manipulation of variables is characteristic of quasi-experimental or experimental studies.
It indicates that the project aims to intervene in one or more variables, that is, change its current state: improve, reduce, increase, enhance, strengthen, and so on, one or more given variables.
- Non-Experimental Design
- These are those that are made without deliberately manipulating variables. That is, where research is not done to intentionally vary the independent variables. The phenomenon under investigation is seen as it occurs in reality. Also called ex post facto research.
In non-experimental designs, there are two major classifications:
2.1.a) Non-Experimental Transactional Designs
These collect data at a single moment in time, that is, a single opportunity. The purpose of these studies is to describe variables and analyze their impact and interaction at a given time or a single measurement. They may include several groups or subgroups of people, objects, or indicators. They are divided into descriptive designs and correlational designs.
2.1.b) Longitudinal Non-Experimental Designs
These designs are used to view and analyze changes over time in certain variables or relationships between them. They collect information on specific points of the study period. For example, developmental psychology has been studied with this model.
The overall non-experimental longitudinal design is:
Longitudinal Trend Design: These designs focus on a specific population. Measurements are taken at different periods of time, say every 3 months for 5 years in a given community.
Longitudinal Group Evolution Design: Determines changes over time in sub-populations or groups. These small groups can be defined by age, sex, color, or other distinctive and relevant characteristics to the study.
Longitudinal Panel Design: They are similar to those defined above, but here the same group is used for measurements throughout the period of investigation.
- Experimental Design
This is a research study in which one or more independent variables (called causes) are deliberately manipulated to analyze the consequences of the manipulation on one or more dependent variables (effects), in a control situation for the researcher.
Classification of experimental design: Pure Experimental (total manipulation of variables), Pre-experimental (lower degree of manipulation of variables), Quasi-experimental (much smaller degree of manipulation of variables).
Pure Experimental Design: Pure experiments manipulate independent variables
for their effects on dependent variables in a situation of control.
The researcher selects the group.
To consider a situation as an experiment, certain requirements should be present:
- Manipulation of one or more independent variables to analyze the effect on a dependent variable.
- To measure the effect that the independent variable has on the dependent variable.
- Control and internal validity of the experiment.
How is Condition No. 1 (manipulation of one or more independent variables…)?
- Presence or absence of that variable
This involves having two groups, one exposed to the independent variable and the other not.
The group exposed to the independent variable is called the experimental group, and the other is the control group.
- Varying the exposure to the independent variable
This involves observing whether the magnitude of the effect on the dependent variable depends on the stimulus intensity of the independent variable.
Another means of manipulating the independent variable relates to the different categories that it can take that are strengths, but different categories.
How is Requirement N° 2 (measuring the effect the independent variable has on the dependent variable)? Validity and reliability.
How is Requirement N° 3 (control or validity of the experimental situation)?
INTERNAL SOURCES
History, Maturation, Instability, Test Management, Instrumentation, Regression Analysis, Selection, Attrition, Interaction between selection and maturation, Other interactions.
How is the control and internal validity of the experiment?
1. Several comparison groups: To compare the results of an experiment (intervention), it is necessary to have more than one group. If only compared to a group, the effect cannot necessarily be attributed to the intervention.
2. Equivalence of groups: First, during the experiment or intervention, randomization.
Experimental Design: Pre-experimental
The pre-experiment is so-called because its degree of “control” is minimal.
The researcher selects the group.
We found: A case study design with a single measurement of pre-trial and post-test with a single group.
Experimental Design: Quasi-experimental
In quasi-experimental designs, subjects are not randomized or matched groups, but these groups were already formed before the experiment; they are intact groups, therefore the degree of control is minimal.
There are several types of quasi-experimental designs:
Posttest design with only intact groups, Design with pre-trial and post-test and intact groups (one control), Design-time series, of a single group and multiple groups.
3. Conceptual and Operational Definition of Variables
What are Variables? Tentative explanations of the phenomenon under investigation are formulated as propositions.
They are the aspects and features of interest to study a phenomenon or problem. They are defined and deducted from the objectives and hypotheses. The objectives are operationalized through variables.
It must be disclosed when appropriate to the design of the investigation: the independent and dependent variables of the study, together with their level (nominal, ordinal, interval, or ratio).
Conceptual Definition of Variables
The conceptual definition of a variable means explaining the variable in terms of other terms. It gives us the definition of literature or the theoretical framework of the attribute. Example: Variable: Perceived work environment (level range).
Concept definition:
“Quality of internal environment relatively long for an organization that is experienced by its members, influences their behavior, and can be described in terms of values of a particular set of characteristics or attributes of the organization” (Stringer, 2002, pg. 83).
Operational Definition of Variables
Operationally defining a variable involves establishing how to measure it.
For example, the average score obtained by a research participant in any of the items of the Organizational Climate Questionnaire by Litwin and Stringer.
4. Research Hypothesis
What are the assumptions? They are tentative proposals (responses) to the research questions that the researcher develops based on the theoretical framework. Hypotheses are statements that could be verified, proven, or tested. They are an anticipated response to a research problem and are derived from the objectives of the study. There is a close relationship between the research problem or situation to be studied and the objectives and hypotheses. The researcher, after formulating the problem, develops a possible answer that must be verified during the investigation.
Considerations: Not all research hypotheses are formulated. Indeed, exploratory studies that explore unfamiliar issues do not raise them. Descriptive studies do not usually raise them either; they are usually present in accordance with the discretion of the investigator, setting out the problem and findings previously established.
The researcher, after formulating the problem, develops a possible answer that must be verified during the investigation. Research hypotheses, depending on the depth of study, can be descriptive, correlational, or explanatory.
- Sources Assumptions: Review of the literature, The problem
- REQUIREMENT OF THE HYPOTHESIS: Correspondence with reality, The relationship between variables, Measurable variables, Techniques for testing.
- Types of Research Research Hypothesis Hi, Null Ho, Alternatives has statistics, types of statistical assumptions, (typical of the quantitative estimation of correlation of mean difference).
Descriptive: a figure forecast Grounds: What happens in a variable may occur in other (bivariate) or other (multivariate) Correlational: establish relationships between variables, group differences.
- FORMULATION: Exploratory studies SCOPE OF THE STUDY: NO hypothesis
- Descriptive studies: Study data to predict a correlation: Specify any relationships between variables.
- Explanatory studies: Determine the reasons why a phenomenon occurs.
General hypothesis: Those that emerge from the goal and are tested through their respective specifics.
Example: There is a relationship between the perceived work environment and the level of motivation shown by the employees of Alfa Compensation Fund.
Specific hypotheses: The general hypothesis needs a detailed definition of the behavior of relevant research variables. They relate directly to specific objectives and include a set of assumptions derived from general assumptions, which aims to help clarify the content and make more explicit the different aspects covered.
Examples:
Workers Compensation Fund of Alfa have a positive perception of your organization’s work climate.
Workers Compensation Fund of Alfa demonstrate a high level of motivation in carrying out their work activities.
There is a direct and positive relationship between perceived work climate and motivation levels expressed by the workers of the Alfa Compensation Fund.
In qualitative research, hypotheses are not usually used, or at least not in the traditional manner typical of quantitative studies. What is often used instead of questions is the marking guidelines (guiding questions or axes), which are previous approaches that would guide the actions of production (computers that will shape the data collection instruments). Objectives relate to the extent that they permit to be built.
5. Population, Sample, and Sampling Process
Types of sampling:
In an investigation, one can work with the entire population or a subset of this (sample) for reasons of practicality and economy of resources. It can be: probabilistic (simple random or random, systematic, stratified, or cluster) or non-probabilistic (accidental or casual, intentional, or quota).
Size: In the literature on social statistics, there are different formulas to calculate the sample size depending on the parameter estimate and the sampling error to consider; however, there are general criteria and practices that facilitate the process. In terms of a statistically representative sample, it should be at least 5% of the population; in other cases, it is suggested, where possible, to work with large samples, that is, at least 30 cases (law of large numbers).
Rationale and characteristics, considering its representative:
Some considerations for achieving Representativeness: Age, socioeconomic status, educational level, gender, and any other consideration relating to the further analysis.
Example:
The population of this study consisted of all workers in the administrative area of the company Alfa Compensation Fund, and the sample to which the data collection instruments were applied consisted of 315 employees, 220 females and 95 males, whose ages ranged between 18 and 71 years of age.
Most employees who participated in the study were employed in sales positions (57%) and Customer Service (35%); the remaining 8% corresponded to the Telefónica Platform staff.
The sample type can be classified as intentional non-probabilistic, while employees were selected according to certain criteria administrative headquarters. In this regard, it was a fundamental requirement for the selection of the participants that they had at least one year of seniority. In addition, the participation of the subjects was voluntary, so that under this approach, about 15 workers answered the instruments.
It is important to note that quantitative methodology is expected that the results (when obtained from a sample) are generalized to the population from which it was extracted. For this, the importance of statistical significance. This condition is not true for methods such as qualitative, losing relevance. In other words, it may be used to refer to the sample or selected from the study that account for the final implementation of the results; hence, they are often called in qualitative studies “study participants”.
The ratings on the selection of participants (or sample) depend on the authors used for their argument. However, they are usually classified as type-intentioned in its most general.
Within these opinion-based sampling can be found (they are selected according to specific criteria of research and know the final size of the participants) and theoretical sampling (they are intended for the construction and emergence of theories, the final size is unknown – it is a successive sampling).
Similarly, relevant characteristics related to the study, the reasons for the choice, and the type of sampling method should be specified.
DEFINE THE SAMPLE DESIGN Probability sampling Stratified sampling
SAMPLES randomly selected The basis is that all components of a universe have an equal chance of being included in a sample, for which once identified, the universe is assigned to each component item a unique identification.
NON-PROBABILITY SAMPLE SELECTION In non-probability samples or directed samples, the choice is not dependent on probability. In these cases, the selection of units or cases is at the discretion of the investigator to consider properties according to specific purposes of the study.
TYPES of probability sampling
Simple, stratified, cluster or clusters, multi-stage.
TYPES OF NON-PROBABILISTIC SAMPLES: Subject type, volunteer subjects, expert subjects, quota, snowball.
6. Data Collection Techniques
Data collection techniques:
At this point, you should describe the instruments used to collect research data.
In the case of working with questionnaires or scales, you should describe the number of items, dimensions, theoretical background, correction method, etc., and technical characteristics such as indicators of reliability, validity, and norms (also note the authors of it).
If the instrument is self-built, the final report should provide information on the procedures used in its construction (pilot implementation, evaluation of judges, etc.).
Example: To find the motivation of workers, the Work Motivation Scale developed by Perez, Arroyo, Garcia, and Torres (2005) was applied.
This instrument consists of 47 items presented in a Likert scale of 4 points, with its endpoints “always motivates me” and “never motivates me”.
Research has shown the adequacy of psychometric properties, which facilitates the correct application and interpretation. In this regard, Perez and Amador (2005) illustrate a factor structure that matches the one originally submitted by the authors, highlighting the appropriate factorial construct validity of the instrument. At the same time, these researchers note that the reliability of the instrument was good (Cronbach’s alpha = 0.93).
Requirements for data collection instruments
Validity: The instrument measures what it purports to measure and nothing else.
Reliability: That measurement instrument applied at other times throws similar results.
As data collection for qualitative studies, the term “information production techniques” is often used. These strategies are often based on open and flexible approaches, so that they account for unmeasured aspects of reality constructed through the social world.
These include: in-depth interviews, semi-structured biographical interviews (life stories), participant and non-participant observation, and field observations, among others.
For each of them, their characteristics and the particular form it would take to conduct their own research shall be described.
7. Data Analysis Procedure
The procedure for data analysis should be mentioned and justified, which will depend mainly on the nature of the data, the measurement level of variables (nominal, ordinal, interval, or ratio), and the mode of study for depth (exploratory, descriptive, correlational, or descriptive). In particular, identifying the elements analyzed with descriptive statistics and inferential statistics, along with the type of statistical test used (parametric or non-parametric).
Specify also the software that has been used. Thus, in quantitative research, the statistical techniques used and the software and version used (SPSS 16.0, STAT, EXCEL, etc.) will be described and justified.
Example: To represent the behavior of individual perception of work climate variables and motivation of workers from Alfa, descriptive statistics were applied. In this regard, the use of frequency tables, contingency tables, graphs (histograms), and measures of central tendency (mean) and variability (standard deviation) are stressed.
To conduct inferential analysis (to assess the correlation between two variables), the parametric Pearson correlation coefficient test was used, considering that both variables are interval-level measures and their relationship (represented by a dispersogram) can be classified as linear (Triola, 2004).
In qualitative research studies, the data analysis procedure should specify whether it will refer to content analysis, discourse analysis, the type of encoding to use (open coding, axial, selective), etc. The foregoing is consistent with the epistemology and related methodological schools.
1. Simple Probability Sample
When all units in the universe are known and have an equal chance of being selected in the sample.
The simplest, but not very fast, is to sign each of the sample items and draw out the cards to complete the sample size.
Simple Probabilistic Sample
Another method is the Random Number Table. For this, any point in the table (row and column) is chosen at random, and from it, in consecutive order, select the numbers that are below the value that represents the size of the sample to complete the sample.
There is another very useful procedure and easy selection from a range (K) which is called systematic selection of sample elements. This interval is determined by the size of the population and sample as follows: K = N / n.
2. The Stratified Probability Sample
When there are layers of importance for research into the universe, the sample is selected within the stratum, which can be used for different procedures.
Stratified Random Sample Types
Proportional stratified sampling.
When it is not enough for each sample element to have the same probability of being
chosen, but also to keep the same proportion in different strata or categories that appear in the population and that are relevant to the objectives of the research. What we do is divide the population into subpopulations and select a sample of each. In this case, simple random sampling is applied within each layer.
Non-proportional stratified sampling.
It is used when justified according to the objective of the investigation and involves manipulating the number of cases to be selected in each stratum.
3. Probability Sample of Clusters
It is one in which the sampling unit (the elements of the universe that are selected) is not the unit or element of the population, but the conglomerate.
It is used to reduce costs, time, and energy, and is based on the consideration that often analysis units are encapsulated or enclosed in some physical or geographic locations that are called clusters.
EXAMPLE
In the above example, the sample will be selected from the class groups:
We know that the 500 students are grouped into 20 groups (different courses).
Define the sample group: 8 groups (courses).
Define the amount to be sampled in each group: 120 ÷ 8 = 15.
Select by lot the 8 groups.
Choose by lot 15 students in each group selected.
4. The Multiple Phases Probability Sample
It is based on a process of subdividing the sampling units.
Initially, some groups or clusters called primary sampling units can be built, then split into smaller groups or clusters identified as secondary sampling units, and so on, until the criterion of research is satisfied.
EXAMPLE: Following the example above, if the U. Autonomous Region has 12 runs, we first define a sample of careers (e.g., 5).
In each race, a sample of class groups is defined, and in each class group, the sample of students surveyed is defined.
Then, schools, groups within each race, and students within each group are selected by lottery.
II. Non-Probability Sample Selection: In non-probability samples or directed samples, the choice is not dependent on probability. In these cases, the selection of units or cases is at the discretion of the investigator to consider properties according to specific purposes of the study.
Types of Non-Probabilistic Samples
1. Subject-type 2. Volunteer subjects 3. Experts 4. Quota 5. Snowball
1. Non-Probabilistic Sample of Subject-Type
Groups are based on typical subjects in relation to a particular feature, where the goal is the richness, depth, and quality of information. It is used in exploratory studies and qualitative research. Example of values, rules, and meaning of gang membership in the institute. Subject types: subjects who belong to gangs.
2. Non-Probabilistic Sample of Volunteer Subjects
Used in studies where the intention is that subjects are homogeneous in certain variables so that the results or effects do not obey individual differences but the conditions they were subjected to, using as sample items individuals who voluntarily agree to participate in the research. EXAMPLE: Study on motivation in the college student using a specific test. Among the volunteers who showed up, those who possess certain characteristics that give the group homogeneity (age, sex, IQ, etc.) are selected so that individual differences do not affect the results.
3. Non-Probabilistic Sample of Experts
Used when the opinion of subject matter experts is necessary. These samples are common in qualitative and exploratory research. Example: Research on students, what are the most efficient methods? Experts: students of higher academic index.
4. Non-Probabilistic Samples of Quota
Interviewers are instructed to conduct questionnaires to persons selected by them, shaping or filling quotas according to the proportion of certain variables of the population. This type of sample is often used in opinion research and marketing.
These samples are to some extent the opinion of the interviewer. This makes the design vulnerable, but if there is good control on the aspect of savings made on location, you can justify this kind of design shows.
EXAMPLE: In the example above, survey IPLA students on teaching methods used by teachers.
Pollsters are told they have a quota of 8 groups of 4 men that have to survey respondents to 1 female to reach the figure of 120 students.
5. Non-Probability Sampling Snowball
Contact consists of an initial number of subjects in a random fashion, and from information provided by these, additional subjects are selected. This sampling has enough use in social sciences, in particular to establish communication networks.
EXAMPLE: The effect drug use can have on academic performance.
Start with 1 or 2 students who use drugs and, through them, locate others, and so on.
C. Define the Sample Size
When a probability sample is formed, it is less important to pinpoint the number of sample units necessary but sufficient to ensure that the results can be extended to the population with a high probability of success.
Types of Instruments in Social Research
1. Scales for Measuring Attitudes 2. Questionnaires 3. Content Analysis 4. Observation 5. Standardized Tests
6. In-Depth Sessions 1. Likert Scale
Rensis Likert developed the procedure in the early 1930s, but it is still applicable and recurrent in several investigations to measure the attitudes of subjects against a reagent. It consists of a set of items presented as statements against which the reaction of the subjects is called for.
Attitudes
Social Sciences frequently study attitudes as they relate to three aspects of a subject’s personality:
Cognitive Aspect, Emotional Aspect, Behavioral Aspect
2. Questionnaires
What kind of questions? / Closed and open questions.
One or more questions to measure a variable? It depends.
“Are the questions pre-coded or not?” / Yes (statistical package).
What are the characteristics of a question? Clear and understandable.
The respondent should not be bothered, refer mainly to one aspect.
The questions should not lead to the answers. / They cannot rely on institutions.
The order of responses should not affect the election.
The language should be adapted to the characteristics of the respondent.
How should the first questions of a questionnaire be? / Neutral or easy.
What does a questionnaire consist of? / Instructions, letter, letter of intent, a guarantee of confidentiality of information.
What size should a guest be? / Depends on the number of variables and dimensions to be measured, respondents’ interest, and the manner of how it is administered.
How are open-ended questions coded? / Find and name the general patterns of response. List signs and numerical patterns (categories).
In what contexts can a questionnaire be administered or applied? / Self-administered
Personal interview / Telephone interview / Self-administered mailed
Tips
Illiterate = Questionnaire interview
Basic reading level = Simple questionnaires, interviews, or questionnaires
High level reading = Self-administered questionnaires or telephone
Personal Interviews = Appropriate atmosphere
3. Content Analysis
Technique to study and analyze communication in an objective, systematic, and quantitative manner. Berelson (1952). Krippendorff extends the definition of content analysis to a research technique for making inferences valid and reliable data about their background, Hernández et al.
4. Observation
Systematic registration, valid and reliable behavioral or overt behavior. (Hernández et al.)
Steps: Precisely define the universe of issues, events, or behavior to observe.
Draw a representative sample of the issues, events, or behavior to observe.
Establish and define the observation units. (Every time, every minute, etc.).
Establish and define the categories and subcategories of observation.
Distance Physics > Approach (affiliation) Walk away (avoidance)
Body movements: Relaxation (affiliation) Voltage (avoidance)
Directed visual behavior: The party (insured) to anywhere (avoidance)
Conduct verbal phrases or sentences (affiliation) dichotomous phrases and silences (avoidance)
Observers: Choose, select the means of observation, develop coding sheets.
Provide training for coders, calculate the reliability of the observer, perform coding, empty observation data, perform analysis, perform observation types:
Participant. Non-participating.
Advantages of observation Unstructured observation techniques stimulate behavior while encouraging response instruments. Accept unstructured material.
Can handle large volumes of data.
5. Standardized Tests
These measure a large number of variables.
Type: Skills and abilities, personality, interests, values, performance, motivation, etc.
Problem of standardized testing: The application context.
6. In-Depth Sessions
It brings together a group of people and works with them in relation to the variables investigated.
One or more meetings can be used.