PRINCIPLES OF SOCIAL RESEARCH
INTRODUCTION
Regardless of its diverse and pluralistic nature, structure and process, social research is generally expected to adhere (loyalty) to certain standards and principles. The nature of these standards and principles may vary, but their presence and necessity are taken for granted. It is worth noting that these principles are often referred to by other names; qualitative researchers, for instance, use different concepts when addressing the principles of their research model.
| Principles of quantitative research | |
| · Precision in measurement · Replication · Validity · Reliability | · Objectivity · Ethics · Representativeness · Generalisability |
MEASUREMENT
Introduction
The research process usually begins with the theoretical preparation or formulation of the research topic. This will establish the foundations for the remaining parts of the study. In the beginning we need to consider another general and fundamental element of the research process and one of the principles of social research, namely measurement.
Nature of measurement
Social research, irrespective of its type and nature, entails a degree of measurement this involves categorizing and/or assigning values to concepts, and is diverse in nature and level of operation. It is also a very useful procedure because it serves to ensure high quality in social research. Most of all, measurement is undertaken to facilitate adequacy, uniformity, comparisons, consistency, accuracy and precision in describing and assessing concepts.
| Measurement facilitates: · adequacy in description and assessment, offering a full account of the concept · uniformity in description and assessment, over time and among researchers · comparisons between complex concepts, enabling the identification of fine distinctions · consistency in the assessment of concepts, over time and among researchers · accuracy and precision in procedures, by taking into consideration all aspects of concepts , · replicability in social research, by the same or different researchers, in the same or different contexts. |
Generally, measurement may be quantitative or qualitative. Quantitative measurement concentrates on numerical values and attributes. Qualitative measurement refers to labels, names and qualities. Qualitative measurement describes attributes by using common concepts or symbols or introducing new ones; a common procedure involves description of categories and classifications. The classification of ‘residence' into 'urban" and "rural', for instance, is a qualitative measurement.
1.3 Variables
Definitions
Variables are empirical constructs that take more than one value or intensity, for example, sex (male, female), marital status (single, married, divorced, widowed, deserted), age and education are variables. The opposite of variables are constant. Constants take only one value or intensity. The researcher determines at the outset of the study which concepts will act as constants and which as variables.
The construction of variables follows a systematic procedure that adheres to the rules of measurement i.e.
· Variables must relate to one concept only
· Must be measurable
There are many types of variables. These types vary according to a number of criteria, such as their nature (geographic variables, demographic variables etc.) their position within the research context (dependent variables, independent variables etc.), or other factors. The following types are most common:
Dependent and independent variables: An independent variable (IV) is a variable that is set to cause changes in or explain another; a dependent variable (DV) is a variable that is set to be affected or explained by another variable.
Extraneous variables: These are variables which are 'outside’ the research question, argument or hypothesis; they are distinct from the dependent or independent variable.
Discrete and continuous variables: Discrete and continuous variables differ from each other in terms of scale continuity; the former are not continuous but use whole units only, whereas the latter are continuous and can be fractioned indefinitely. Discrete variables are counted, not measured; continuous variables are measured, not counted. Examples of discrete variables are ethnicity, race, sex, marital status, cause of death or blood type. Examples of continuous variables are height, distance, time, age, temperature or IQ scores.
Demographic variables: Demographic variables deal with demographic data such as age, residence, religion, marital status, family size, race, education and sexual preference. Remember, a demographic variable can be dependent or independent, discrete or continuous, depending on the nature of the research design. Discrete variables are measured at the nominal or ordinal level, and continuous variables at the interval or ratio level.
1.4 Levels of measurement
Measurement can be performed at four levels, they are: the nominal, the ordinal, the interval and the ratio level. Nominal-level measurement has the lowest and ratio-level measurement the highest match with the real-number system.
| Nominal-level measurement: · involves nominal categories and is essentially a qualitative and a non-mathematical · measurement: it actually names and classifies data into categories · does not have a zero point · cannot be ordered in a continuum of low-high · produces nominal (in name only) or categorical data · assumes no equal units of measurement | · assumes the principle of equivalence: all units of a particular group are taken to be the same |
Nominal-level measurement
This is the simplest, the lowest and the most primitive type of measurement. At this level, measurement involves classification of events into categories.
Numbers assigns to categories have no mathematical meaning, are used only for identification and cannot be added, subtracted, multiplied, divided or otherwise manipulated mathematically. Classifying the respondents in categories such as male-female black-white, young-old, single, married, cohabiting, separated, divorced, remarried or widowed, or Catholic, Protestant,
Ordinal-level measurement
Measurement at the ordinal level involves nor only categorizing elements into groups but also ordering data and ranking variables in a continuum ranging according to magnitude, that is, from the lowest to the highest point. They do not allow mathematical operations such as addition or subtraction.
| Ordinal measurement: · refers to ranks based on a clear order of magnitude of low and high signifying that some elements have more value than others · assigns numbers actual mathematical meaning as well as identification properties · is essentially a quantitative measurement · shows a relative order of magnitude |
With regard to the last point, order of magnitude allows categories to be ranked (who is first, second, last) but does not indicate the amount of difference between the groups
Example are: status (low, middle, high); size (smallest, small, big, biggest), quality (poor, good, very good, excellent); class (low, middle, high); achievement (poor, moderate, high); income (low, middle, high). Ranking occupations is another example.
Interval-level measurement
This level of measurement, as well as demonstrating the properties of ordinal-level measurement, provides information about the distance between the values, and cousins equal intervals, ordering subjects into them. This method allows the researcher to assess differences between respondents and to obtain more detailed information about the research topic.
Interval-level measurement allows the researcher, first, to establish whether two values are the same or different (as in nominal measurement), second, to determine whether the one is greater or smaller than the other (as in ordinal measurement and third, to ascertain the degree of difference between them. Nevertheless, it does not have a true zero point, and if a zero is used it is set arbitrarily, is done so for convenience and does not mean absence of the variable.
| Interval-level measurement: · includes equal units · is essentially a quantitative measurement · facilitates differentiation and classification · incorporates ordering of subjects · specifies the numerical distance between the categories |
For example, if the IQ of two students is 105 and 125 respectively, in nominal terms this means that they have a different IQ; in ordinal terms that the first student has a lower IQ than the second; and in interval terms, that: the IQ of the second student is 20 points higher than that of the first student, but not, say, one fifth greater than the other student.
In mathematical terms, at this level numbers assigned to categories are used to count and rank, but can also be added to and subtracted from each other. This indicates that interval-level measurement is superior to the other two. However, given that there is no true zero, they cannot be multiplied or divided. Statistical measures for nominal, rank and interval data can be used. Examples of this type of measurement are degrees of temperature, calendar time (day, week, month) attitude scales and IQ scores.
Ratio-level measurement
Measurement at this level includes all the attributes of the other three forms, plus the option of an absolute true zero (0) as its lowest value, which in essence indicates the absence of the variable in question. Simply, it is an interval-level measurement with an added true zero. Hence all attributes of interval-level measurement also apply here. Ratio-level measurement: allows the researcher to make statements about proportions and ratios, that is, to relate one value to another. For instance, a comparison of speed of response of two students to a stimulus - say, 10 seconds and 20 seconds - allows the researcher to conclude that the first is twice as fast as the second.
Levels of measurement: a summary
| Criteria | Nominal | Ordinal | Interval | Ratio |
| Properties of measurement | Naming | Naming and ranking | Naming, ranking and equal intervals | Naming, ranking, equal intervals and zero point |
| Nature of measurement | Categorical | Ranking | Scoring | Scoring |
| Mathematical functions | None | None | Addition and Subtraction | All four functions |
| Relevant statistical tests | Lambda test X2 test | Spearman’s P M-W U test Sign test | Pearson's r t-test; ANOVA | Pearson's r t-test; ANOVA |
| Nature of underlying construct | Discrete | Discrete or continuous | Continuous | Continuous |
| Examples | Marital status, gender, race, residence, ethnicity | Income, status, achievement, social class, size | Temperature, calendar time, IQ scores, attitude scales | Length, weight, distance, number of children, age |
| Typical answers to questions | Male, Female, Single, Married, Irish. | Always; often; sometimes; never | Scores, Likert scales, Degrees | Years, Kilograms, Kilometre |
In the social sciences this level of measurement is employed mainly when measuring demographic variables; however, it is considered inappropriate for measuring attitudes and opinions. This is because a, zero (0) option in an attitude scale means no attitude, or no opinion, which is misleading; even having ‘no opinion' is in itself an opinion.
Measuring variables
Variables are not measured at one specific level only. Whether a variable will be measured one way or another depends very much on how it is conceptualized and on what type of indicators have been used during measurement. The same variable can be measured in a variety of ways. Age, for instance, can be measured nominally, if it is defined in broad and discrete categories, such as infancy, adolescence, adulthood, middle age and old age; or as young and old. It can be measured also at the ordinal level, when respondents are ranked according to age from the oldest to the youngest.
Age can also be measured at the interval level, given that units are equal, and that we can determine how many units of difference there are between age levels. Interval-level measurement tells us not only whose age is higher (as in ordinal-level measurement) but also how much higher it is. Age can, finally, be measured in the ratio level, since it has an absolute (non-arbitrary) zero. One cannot be younger than 0; and a 20-year-old person is twice as old as a 10 year old person.
| Arbitrary and true zeros The use of true zero as the distinguishing characteristic of ratio scales has caused some confusion. This is due to the fact that it often is difficult to distinguish true zeros from arbitrary zeros. Zeros are not always 'true'. True zeros are meaningful; arbitrary zeros are not. For instance, when we measure temperature, a zero degree reading does not mean no temperature at all! And in measuring attitudes, a zero does not mean no attitude at all (having no opinion on an issue is an opinion!). These zeros are not true zeros, they are arbitrary zeros. However, when measuring income, number of cars, or number of children, a zero indicates no presence of these criteria: it means no children, no income, no cars. These are true zeros; and only measurement using these true zeros can be conducted at the ratio level. |
2- REPLICATION
· Replication applies to quantitative research.
· Can be repeated by other researchers for validity checking
· Guarantee the absence of subjective influence
· Full objectivity in the procedure
· The results here are expected to reflect the views of the respondents fully
· Same outcome are achieved each time the study is repeated.
· Researchers could use same research instruments to
· Full study can be replicated.
· Valid comparisons and more legitimate generalizations
3. SCALES AND INDEXES
3.1 Introduction
Scales are techniques employed by social scientists in a variety of contexts, particularly in the area of attitude measurement. They consist of a number of items (statements or questions) and a set of quantified response categories. Each item is chosen so that people with different points of view about it react to it in a different way. Scales are employed because they offer;
· High coverage: Scales allow a complete coverage of all significant aspects of the concept.
· High precision and reliability: Scales allow a high degree of accuracy and reliability.
· High comparability: The use of scales permits detailed and accurate comparisons between sets of data.
· Simplicity: Scales help to simplify collection and analysis of the data.
| Guidelines for scale construction · Language must be simple, clear and direct. · Items must be brief (up to 20 words) and contain one issue only. · Complex sentences must be avoided. · Items referring to past events and factual items must be avoided. · Ambiguous and irrelevant items must be avoided. · Items that may be accepted or rejected by all respondents must be avoided. · Words such as all, always, no one, never, only, exactly, almost should be avoided. · Use of professional jargon and double negations should be avoided. · Response categories must be mutually exclusive, complete and unidimensional (i.e. measuring one single construct). ' . |
3.2 Examples of scales
Likert scales present items in a continuum (range) that covers the whole range of possible responses, allowing respondents to choose the answer that fits their opinion. The following is an example of the type of questions employed in Likert scales.
Qu. 57. Gay marriage is as good as heterosexual marriage. (Please circle the number in front of the answer of your choice.)
Strongly agree
Agee
Undecided
Disagree
Strongly disagree
Another example is the Bogardus Social Distance Scale, which helps to test how close people allow others, for example strangers, to come to them. The content of this scale is as follows
| Bogardus Social Distance Scale Would you consider an asylum seeker as a [ ] close relative by marriage [ ] personal friend [ ] neighbour [ ] colleague at work [ ] speaking acquaintance only [ ] visitor to your country [ ] person to be kept out of the country |
3.3 Indexes
An index is a measure containing a combination of items, the values of which are summed up to provide a numerical score. An example is the Quality of Life index for the city of Vienna. Such an index may include the following items: employment opportunities, recreation opportunities, weather, pollution level, medical services, educational opportunities, child-care services, safety, crime rate and racial problems. These items will be transformed into questions/statements and the index presented for evaluation. Each question will be scored and the total will present a single measure.
The items of an index can be given the same weight (un-weighted index) or different values (weighted index). The latter option is taken when, for instance, some index items are thought to be more important than others.
4. VALIDITY
4.1 Validity in quantitative research
Validity is the property of a research instrument that measures its relevance, precision and accuracy. Validity tells the researcher whether an instrument measures what is supposed to measure, and whether this measurement is accurate and precise.
Relevance: An instrument is considered to have absolute validity when it measures what it is supposed to measure and nothing else - no more, no less. If a researcher wanted to know the distance between two cities, kilometers or miles would be a relevant instrument. Similarly, a scale of kilos is a relevant instrument when measuring a person’s weight, but is not relevant if used to estimate the person's intelligence.
Accuracy: validity also entails a degree of accuracy. Accuracy refers to the ability to identify the true value of the item in question. For instance, if you step on your bathroom scales (which measure whole kilos only), and you obtain a reading of 70 kilos (which is your real weight), the scale is accurate. However, if the reading were 68 kilos, the scale would have been inaccurate and hence invalid).
Precision: Validity requires also that a measure is precise. Precision implies accuracy, but in addition it requires that measurements employ the smallest possible measure. For instance, for a dietician who wants to measure the weekly weight gain or loss of a patient undergoing a special medical treatment, scale that read whole kilos only (68, 69, 70 kilos, and so on) are not precise enough.
| What is validity? Validity: · Is a measure of precision, accuracy and relevance · Reflects the quality of indicators and instruments · Refer to the ability to produce findings that are in agreement with theoretical or conceptual values · Answer the question: Do the instruments/indicators measure what they are supposed measure? |
4.2 Testing validity
In quantitative research, there are two ways of checking the validity of an instrument; these are empirical validation and theoretical validation. In this context, tests of internal and external validity are employed. In the former, the validity a measure is checked against empirical evidence. In the latter, the validity of an instrument is ascertained through theoretical or conceptual constructs. In both cases, validity is claimed if the test results are acceptable.
Empirical validation
Empirical validation tests pragmatic (realistic) or criterion (standard) validity. If an instrument has for instance, produced results indicating that students involved in student union activities do better in their exams, and if this is supported by available data, the instrument in question has pragmatic validity. Again, validity here is assumed if the findings are supported by already existing empirical evidence. In this case validity is concurrent validity.
If new findings support the predictions of the measure in question, this measure is said to be valid. For example, if a study found that an eventual introduction of advanced statistics into the social sciences degree would result in a significant drop-out of older students, and if meanwhile this prediction is supported by new findings, the measure has validity. This is known as predictive validity.
Theoretical validation
Theoretical or conceptual validation is employed when empirical confirmation of validity is difficult or impossible. A measure is taken to have theoretical validity if its findings comply with the theoretical principles of the discipline, that is, if they do not contradict already established rules of the discipline. There are several types of theoretical validity.
Face validity
An instrument has face validity if, ‘on the face of it', it measures what it is expected to measure. For example, a questionnaire aimed at studying sex discrimination has face validity if its questions refer to discrimination experienced by people because they are male or female. The standards of judgment here are based on general theoretical standards and principles, and on what other researchers consider to be the case. It should be noted that when there are no common standards and principles, and when there is disagreement as to what is generally right to expect this instrument has no face validity.
Content validity
A measure is considered to have content validity if it covers all possible dimensions of the research topic. If a researcher in a study of religiosity employs a questionnaire that contains questions only on 'church attendance', this research instrument has no content validity.
Construct validity
A measure can claim construct validity if its theoretical construct is valid, in other words, if it measures the constructs it is supposed to measure. Validation concentrates here on the validity of the theoretical construct. For example, if an instrument tests the attitudes of two groups of students known to have different views on the issue in question, and this instrument finds them to be different- that is, it verifies the known difference - this instrument is said to have construct validity.
Internal validity
Internal validity refers to the extent to which the research design impacts on the research outcomes. Internal validity checks ensure that the findings of the research have not been affected by instruments or procedures, and that they are the results of the independent variable. Examples of factors that can threaten internal validity, for example in experimental research, panel studies or trend studies, are given below (see Farber, 2001);
· Unexpected structural changes might occur during the course of the study, subjecting respondents to different conditions.
· Normal developmental changes are to be expected in longitudinal studies where data collection occurs in, say, five-year intervals.
· Diverse methods may be used over the course of the study, subjecting respondents to different research instruments.
· Different sampling procedures may be employed during the course of the study, leading to selection problems.
· There may be diverse personnel in the study, with different levels of competence, experience, knowledge and attitude.
· Changes or alterations in recording techniques may lead to inconsistent records.
In such cases, the respondents are exposed to factors that can affect the information collected in the study.
External validity
External validity refers to the extent to which research findings can be generalized, and is mostly relevant to explanatory studies. The following are a few examples of how conducting the research can threaten external validity.
· Testing: Being chosen to take part in the study can stimulate respondents to become more familiar with the study object and hence become more knowledgeable than the average population.
· Sampling: Inadequate or biased selection may lead to unrepresentative samples.
· Multiple exposures: Exposure to a variety of research instruments might cause an interaction effect and associated problems.
· Measures: Inappropriate measures may produce unrealistic responses.
4.3 Validity in qualitative research
Validity is a methodological practice not only of quantitative but also of qualitative research. Qualitative researchers aim to achieve validity, as it frees data from interference and contamination, control or variable manipulation; this is facilitated in a number of ways, particularly through their orientation towards, and study of, the empirical world through construction of appropriate methods of data collection and analysis or through specific measures such as communicative, cumulative, ecological or argumentative validity.
· Cumulative validation: A study can be validated if its findings are supported by other studies. The researcher can compare the various findings and make a judgment about the validity of the studies.
· Communicative validation: This form of validation entails the involvement of the participants - by checking accuracy of data, evaluation of project process, change of goals etc., by employing expert external audits, and by using triangulation - in order to achieve a multiple perspective, and to confirm authenticity.
· Argumentative validation: This form of validity is established through presentation of the findings in such a way that conclusions can be followed and tested.
· Ecological validation: A study is held to be valid if carried out in the natural environment of the subjects, using suitable methods and taking into consideration the life and conditions of the researched.
Other tactics
Validity is not a criterion of quantitative research but a common basis for most types of research. A view supported by many workers in this area is that investigators do not need to demonstrate validity but rather methodological excellence, that is, research performance in a professional, accurate and systematic manner. For some writers types of validation employed in quantitative research are even more effective.
· The data are closer to the research field that in quantitative research.
· The collection of information is not determined by research screens and directives.
· The data are closer to reality than in quantitative research.
· The opinions and views of the researched are considered.
· The methods are more open and more flexible than in quantitative research.
· There is a communicative basis that is not available in quantitative research.
· A successive expansion of data is possible.
| Validity in ethnographic research (1) The researcher should refrain from talking in the field but rather should listen as much as possible. He or she should (2) produce notes that are as exact as possible (3) begin to write early, and in a way (4) which allows readers of his or her notes and report to see for themselves. This means providing enough data for readers to make their own inferences and follow those of the researcher. The report should be (5) as complete and (6) as candid as possible. (7) The researcher should seek feedback for his or her findings and presentations in the field or from his or her colleagues. (8) Presentation should be characterized by balance between the subjects and (9) by accuracy in writing' |
5 RELIABILITY
Reliability refers to the capacity of measurement to produce consistent results. Reliability is equivalent to consistency. Thus, a method is reliable if it produces the same results whenever it is repealed, and is not sensitive to the researcher, the research conditions or the respondents. Reliability is also characterized by precision (accuracy) and objectivity. As in validity, so in reliability there are two major aspects of interest in this context; these are internal reliability and external reliability. Internal reliability means consistency of results within the site (location), and that data are plausible (reasonable, believable) within that site. External reliability refers to consistency and replicability of data across sites.
| What is reliability? Reliability: · is a measure of objectivity, stability, consistency, and precision · measures the quality of indicators and instruments · refers to the ability to produce the same findings every time the procedure is repeated · answers the questions: does the instrument/indicator produce consistent results? Is the instrument free of bias associated with the researcher, the subject or the research conditions? |
5.1 Reliability in quantitative research
There are at least three types of reliability, all of which are considered by social researchers. These are:
· Stability reliability: relating to reliability across time. Here the question is whether a measure produces reliable findings if it is employed at different points in time.
· Representative reliability: which relates to reliability across groups of subject. The question here is whether the measure will be reliable if employed in groups other than the original group of subjects.
· Equivalence reliability: which relates to reliability across indicators and to multiple indicators in operationalization procedures. The question here is: will the measure in question produce consistent results across indicators?
There are also several methods for resting reliability of an instrument. The most common methods are the following:
· Test-retest method: The same subjects are tested and retested with the same instrument. If the same results are obtained the instrument is reliable.
· Split-half method: Responses to the items of an instrument are divided into two groups (e.g. odd/even questions) and the scores correlated. The type and degree of correlation indicate the degree of reliability of the measurement.
· Inter-item test and item-scale test: Inter-item correlations or item-scale correlations indicate the degree of reliability of the instrument.
· Alternate-form reliability: Reliability is rested by administering two similar instruments in one session, and is assessed by the degree of correlation between the scores of the two groups.
5.2 Reliability in qualitative research
Qualitative researchers consider reliability an important parameter of research but, they employ different methods from quantitative research, such as increasing the variability of perspectives in research, or setting up a list of possible errors or distortions which they aim to avoid. In the majority of cases, they avoid the use of the concept 'reliability’, instead they use concepts such as credibility and applicability, or auditability.
Objectivity is replaced by confirmability; coherence (rationality), that is, the extent to which methods meet the research goals; openness, the degree to which otherwise suitable methods are allowed to be used; and discourse that is, the extent to which researchers are allowed to discuss the researched data and interpret them together and evaluate the consequences of such findings. References to trustworthiness, dependability credibility, transferability and confirmability also seem to be popular.
| Qualitative research and quantitative reliability In their quest for validity, quantitative researchers are thought to: · control the environment · employ high levels of measurement and standardization · restrict the researcher-researched relationship · create artificial situations which are different from those they intend to study · alienate the researcher from the research environment, which is counterproductive |
Internal and external reliability are addressed in qualitative research by using following paths:
· prolonged engagement and persistent observation
· peer review or debriefing
· analysis of negative cases
· checking ‘the appropriateness of the terms of reference of interpretation and their assessment’
· member checks (communicative validation)
· external auditing
Following steps should be followed if internal reliability is to be achieved:
· Use low inference (supposition) descriptors.
· Use multiple researchers whenever possible,
· Create a careful audit trail (a detailed record of data that can be used by other scholars to check internal validity).
· Use mechanical recording devices where possible (and with permission).
· Use participant researchers or informants to check the accuracy or congruence of perceptions.
With regard to external reliability following are the five steps:
· Specify the researchers' status or position clearly so that readers know exactly what point of view drove the data collection.
· State the identity of the informants (or what role they play in the natural context) and how and why they were selected (while maintaining confidentiality).
· Explain the context or set boundaries and characteristics carefully so that the reader can make judgments about similar circumstances or settings.
· Define the analytic constructs that guide the study (describe specific conceptual frameworks used in design and deductive analysis).
· Specify the data collection and analysis procedures carefully.
5.3 Validity and reliability
Example: If a person weighs himself 20 times and every time he receives a reading of 65 kg (which is also his true weight), the scale is both reliable and valid. If all recorded readings were 40 kg, the scale is reliable but not valid. And if he obtains 20 different readings (40 kg, 45 kg, 63 kg etc.), the scale is neither valid nor reliable.
| Criteria of validity and reliability | |
| Validity | Reliability |
| is a measure of the quality of measurement | is a measure of the quality of measurement |
| tests the quality of indicators and research instruments | tests the quality of indicators and research instruments |
| measures relevance, precision and accuracy | measures objectivity, stability, consistency and precision |
| tests the ability to produce findings that are in agreement with theoretical or conceptual values | tests consistency, i.e. the ability to produce the same findings every time the procedure is repeated |
| ASKS: Does the instrument measure what it is supposed to measure? | ASKS: Does the instrument produce the same results every time it is employed? |
6 OBJECTIVITY IN SOCIAL RESEARCH
6.1 The debate
Objectivity is the research principle that requires that all the personal values and views of the investigator must be kept out of the research process. The purpose of this is to minimize personal prejudice and bias, and to guarantee that social reality will be presented as it is, and not as the investigator interprets it, imagines it or wants it to be. There are two different views on objectivity. One is known as value neutrality, and the other as normativism. The former was the position of quantitative researchers, and the latter the stance of qualitative and other researchers.
| What is objectivity? Objectivity is the empiricist doctrine that the research process and design must be free of personal bias and prejudice. It rests on the belief that facts and values should be kept apart, and that research should focus on what really is and not on what ought to be. Objectivity reflects value neutrality |
Value neutrality
The notion of value neutrality reflects the requirement that investigators ought to minimize the effects of their own biases. Social researchers are seen as ‘technicians' or consultants and not as reformers; or better, as neutral observers and analysts and not as philosophers or moralists. The researcher’s personal views and value judgments are to be kept out of research. In a more general context, objectivity subscribes to a number of principles and convictions, three of which are the following:
· The social sciences are value free; their goal is to study what is and not what ought to be. Research should aim to achieve the highest possible degree or objectivity
· Social scientists should be value free; they should rule out value judgments, subjective views, personal bias and personal convictions.
· Value judgments should be reserved for policy makers, and not for social scientists.
Normativism
Normativism is critical of the value and usefulness of objectivity and proposes that value neutrality is not justified. More specifically it proposes that:
· Objectivity is unattainable, unnecessary and undesirable.
· Social science is normative; its goal is to study what ought to be and not only what is.
· People’s orientation is based on and constructed with values, which direct thinking and action, and cannot be neutralized, isolated or ignored.
· Being normative and disclosing the inevitable bias or personal beliefs is less dangerous than pretending to be value free.
Objectivity in quantitative research
For quantitative researchers, objectivity is regarded as a virtue that every social researcher should try to achieve. Although they are aware that it is difficult to reach a high degree of objectivity. Research has the task of capturing and presenting reality as it is and not as it is interpreted, imagined or wanted to be by the investigator. Objectivity serves to restrict the influence of personal biases and prejudices in the research process and to allow reality to come forward as it is without manipulation.
| The logic of objectivity · The purpose of research is to discover objective truths. · Objective truths can be verified only when contrasted with objective reality. · The task of verification is completed by researchers. · It is important that verification is conducted objectively and focuses on objective reality. · The subjective views and personal values of the researcher can only distort the process of verification, and cannot enhance the objectivity of truths. · Hence subjectivity distorts the process of discovery of objective truths and must be excluded from research. |
6.3 Objectivity in qualitative research
Qualitative researchers fundamentally reject the notion of objectivity. Given that qualitative research rests within the parameters of an interpretivist epistemology. Hence, involving personal views and interpretation in the research process is not only acceptable but advisable because, there is no objective reality to begin with. Researchers capture one aspect of reality - their reality - and this is what they can describe and present. Apart from this, value neutrality is considered to be unattainable, unnecessary and undesirable. They consider basic evaluation, feelings, beliefs and standards as significant and influential. Qualitative researchers give importance to value judgments to solve social problems.
| Where and how objectivity is practiced | |
| Research parameters | How objectivity is practiced |
| Reality | By perceiving reality in objective terms, as an objective reality that must be reproduced as it is, without distortions of any kind |
| The researcher | By respecting value neutrality, i.e. being free of personal values, bias, and prejudice, and studying 'what is' and not 'what ought to be" |
| Research topic | By conceptualizing the research topic in an objective manner; selection of indicators and definitions should be free of personal bias, experiences, and views |
| Methodology | By choosing the appropriate methodology in a process free of personal preferences, ideologies or bias |
| Design construction | By constructing the design using professional standards, avoiding personal bias, and assuring compliance to ethical standards throughout the study |
| Sample | By choosing the sample in compliance with research standards and practices, excluding personal bias and preference |
| Data collection | By choosing relevant methods and gathering data using professional standards, and by focusing on "what is' and not on 'what ought to be’ |
| Administration | By guiding arrangements towards facilitating the completion of the study, using fair and professionally acceptable standards and not personal or ideological preferences |
| Data analysis | By describing relevant methods clearly and by conducting analysis in a manner that reflects professionalism, and avoiding personal biases and preferences |
| Interpretation | By including clear and detailed justification of conclusions, revealing personal perceptions and interpretations of data and reality for verification |
| Reporting | By constructing the report in a manner that clearly outlines the data and the personal interpretations of the writer; personal views etc. should be made clear. |
7 REPRESENTATIVENESS
| What is representativeness? Representativeness is a research principle that reflects the capacity of social research to produce findings that are consistent with (representative of) what appears in the target population; this is a property of sampling. The aim of representativeness is to ensure that all relevant groups of the target population are adequately represented in the research sample. The degree of representativeness determines the extent to which the findings of a study can be generalized. |
Quantitative research
Representativeness has a central place in and is one of the aims of quantitative research. Several procedures have been developed and are currently employed to ensure the representativeness of the sample and the study in general. Most of these procedures deal with the nature of sample selection as well as with sample's size and composition. Representative studies speak for the whole population. Non-representative ones do not; they do not support generalization.
Qualitative research
In qualitative research representativeness is considered irrelevant and unimportant for several reasons. First, it is not consistent with the principle of the qualitative paradigm, and second, the size of the sample and the nature of qualitative sampling procedures do not allow any claims for representativeness. Nevertheless, this does not mean that qualitative researchers are not interested in representativeness. There are researchers who consider it to be an indispensable element of qualitative research take precautions to ensure it which, include avoidance of sampling non-representative informants, and of ‘generalizing from unrepresentative events or activities', or 'drawing inferences from non-representative processes'. Miles and Huberman (1994: 265) advise that weak non-representative cases should be expanded, and suggest the following ways to do so:
· Increase the number of cases.
· Search purposely for contrasting cases.
· Sort the cases systematically and fill out weakly sampled case types.
· Sample randomly within the total universe of people and phenomena under study
8 GENERALIZABILITY
| What is generalizability? In social research, generalizability refers to the capacity of a study to extrapolate the relevance of its findings beyond the boundaries of the sample. In other words it reflects the extent to which a study is able to generalize its findings from the sample to the whole population. Obviously, the higher the generalizability, the higher the value of the study. |
MAIN POINTS
· The principles of research are precision in measurement, accuracy, validity, reliability, objectivity, replication, representativeness and generalizability.
· Some form and degree of measurement is included in all types of research.
· There are four levels of measurement: nominal, ordinal, interval and ratio levels,
· Variables are measured at the highest level possible.
· Validity is the ability to produce accurate results and to measure what is supposed to be measured. It is an attribute of quantitative and qualitative research,
· Quantitative research employs many types of validation: for example, empirical, theoretical, face, content and construct validity.
· In qualitative research, validation takes the form of cumulative, communicative, argumentative or ecological validation.
· Reliability is the capacity to produce consistent results. It is an attribute of both quantitative and qualitative research
· Objectivity excludes personal values from research, and is valued in quantitative research.
· Representativeness is an important characteristic of social research that is closely adhered to in both quantitative and qualitative research.
No comments:
Post a Comment