Please enable JavaScript.

Coggle requires JavaScript to display documents.

R4 QN research (4.5 Instrumentations ( (role of measurement?, A powerful…

- - - - Identify the concept to be measured
      - Specify an indicator (or indicators) of the concept
      - Operationally define the way of ordering the data, such as determining how to assign ratings from low to high, rarely to frequent, disagree to agree, and so on
      - Determine the validity and reliability of the scores produced by the measurement procedure that was utilised.
  - - - according to its classification, mainly, measurements in the cognitive, affective and psychomotor
        domain.
        
        Affective
        
        personality scales
        
        attitude scales
        
        Projective devices
        
        Psychomotor
        
        Cognitive
        
        Aptitude tests
        
        Achievement tests
  - - - Step 3: Operationalise the construct definition
    - - Step 4: Choose an instrument
  - - - adapt a questionnaire
      - design new questionnaire
        
        problems faced when writing new questionnaire items.
        
        Clarity — Inexperienced questionnaire makers are likely to be naïve about the clarity of their questionnaire items as well as instructions to respondents, even though they have worked on it many times.
        
        Broad coverage — Novice researchers may attempt to cover their field broadly in the hope that some of the responses will provide the answers for which they are groping.
        
        Poor items or phrase — Often enough, they are not likely to ask the right questions or to phrase them properly. In addition novice researchers face difficulty in wording the questionnaire clearly.
        
        Limitation on the number of words — The limitation on the number of words is a particular hazard in the questionnaire.
        
        Different meanings — The same words mean different things to different people. After all, even questionnaire makers have their own interpretation, and the respondents may have many different interpretations.
        
        Written language — Questionnaire makers must depend on written language alone. Obviously they cannot be too careful in phrasing questions to ensure the clarity of purpose.
        
        improve questionnaire items
        
        Define or qualify terms that could easily be misinterpreted
        
        Be careful when using descriptive adjectives and adverbs that have no agreed-upon meaning
        
        Beware of double negatives.
        
        Be careful of insufficient alternatives
        
        Avoid the double-barreled question
        
        Underline a word to indicate special emphasis
        
        When asking for ratings or comparisons, provide a point of reference
        
        Avoid unwanted assumptions
        
        Phrase questions so that they are appropriate for all respondents
        
        Characteristic of a good questionnaire
        
        It deals with a significant topic, the respondent will recognise as important enough to warrant spending one’s time on. The significance should be clearly and carefully stated on the questionnaire or in the letter that accompanies it.
        
        It only seeks information that cannot be obtained from other sources such as school reports or census data.
        
        It is as short as possible and only long enough to get the essential data.
        
        Questions are presented in good psychological order, proceeding from general to more specific responses. This order helps respondents to organise their own thinking so that their answers are logical and objective.
        
        Other considerations in developing an instrument
        
        Practical limitations — develop items that are suitable for the respondents’ level. Perhaps mathematics attitudes can be measured in small children by providing a brief but clear statement on their level of understanding or the use of pictures may be more creative in its presentation.
        
        Participants must have enough knowledge to be able to objectively state their response.
        
        Allow time to revise, pilot test, re-revise, and re-revise the instrument.
        
        Keep the instrument as short as possible to avoid fatigue or boredom.
      - adopt a questionnaire
    - - Likert method
      - Thurstone technique
      - Semantic differential
  - - - Observation is carefully planned, systematic, and perceptive. Observers should know what they are looking for and what is irrelevant in a situation. They are not distracted by other dramatic or spectacular incidences.
        
        Observers are aware of the wholeness of what is observed. Although they are alert to significant details, they know that the whole is often greater than the sum of its parts.
        
        Observers are objective. They recognise their likely biases, and they strive to eliminate their influence on what they see and report.
        
        Observers separate the facts from the interpretation of the facts. They observe the facts and make their interpretation at a later time.
        
        Observations are checked and verified whenever possible by repetition or by comparison with those of other competent observers.
        
        Observations are carefully and expertly recorded. Observers use appropriate instruments to systematise, quantify, and preserve the results of their observations.
        
        Observations are collected in such a way as to make sure that they are valid
        and reliable.
  - - - intrusion may affect the behaviour of the person(s) being observed.
        
        Some methods to overcome this issue:
        
        Concealing the observer has been used to minimise this reactive effect by using cameras and one-way screens to make unobtrusive observations.
        
        One-way glass and concealed microphones and videotape recorders have been used when observing the behaviour of children in natural group activities so that the observers could see and hear without being seen and heard.
        
        Introducing observers as active participants in the activities of the group will minimise the reactive effect more efficiently.
        
        Make their purposes known to the members of the group observed. Some may feel that concealing the intentions of the participant observers raises ethical questions of invasion of privacy and establishes a false interpersonal relationship with the individuals in the group. However, in some cases informing those observed of the complete purpose of the study may affect behaviours so as to make the study meaningless.
- - - - Stability over time (test-retest).
        
        The scores in a test1 (first administration) will be highly correlated with the scores of the test2 (second administration) to the same subjects at a later date if the test has good test-retest reliability.
        
        Individual scores may change due to having taken the test before, but we are interested here in the relative position of the individual’s score.
        
        That is, does everyone’s score change in the same direction and by the same amount from test one to test two?
        
        The reliability coefficient obtained is a measure of coefficient of stability.
        
        CORRELATION IS IMPORTANT FOR THIS RELIABILITY
      - Stability over item samples (equivalent or parallel forms).
        
        Some tests have two or more forms(VERSIONS) that may be used interchangeably.
        
        In these cases, the scores in one version1 will be very similar with scores in the alternative version2 of the test (for example, scores in form A will be very close to scores in form B) if the test has this type of reliability.
        
        Correlation is not as important here as equal results for the two versions.
        
        Time and testers are not factors because the two versions will usually be combined and given as one longer test with the items separated later for separate scoring of the two versions. The reliability coefficient obtained is a measure of coefficient of equivalence.
      - Stability of items (internal consistency).
        
        Test items should be highly related to other test items. ---> This is important because the test, or in some cases the subtest, needs to measure a single construct. --> For instance, if some test items measure reading readiness and others measure musical ability, the test will not measure either construct consistently. --->Thus, each test or subtest is to be used individually, and must demonstrate item consistency. The reliability coefficient obtained is a measure of coefficient of internal consistency.
        
        methods for measuring internal consistency:
        
        a. Split halves.
        
        Method 1 - splitting by odd and even numbered items
        
        scores for the odd-numbered items can be correlated with the scores for the even-numbered items
        
        Method 2 - splitting by scores in first half and second half
        
        in some but not in most tests, the scores in the first half of the test can be correlated with scores in the second half of the test. (This only works if the test is long and the items are equally difficult throughout the test.)
        
        Owing to the fact that the correlations resulting from the above splits would be for only half a test and because generally the longer a test is, the more internal consistency it has, the correlation coefficient for the half-test is therefore corrected by using the Spearman-Brown formula.
        
        see notes for calculation
        
        b. Coefficient of consistency.
        
        Kuder-Richardson formula 20 (KR-20) and Cronbach’s alpha are estimates of internal consistency reliability of an instrument, by determining how all items of an instrument relate to all other items of the instrument and to the whole instrument.
        
        Internal consistency results when all the items or tasks in an instrument are related. In other words, the items measure similar things.
        Both techniques provide reliability estimates that are equivalent to the average of the split-half reliabilities computed for all possible halves.
      - Stability of scores (inter-scorer).
        
        Certain types of tests, in particular projective tests, leave a good deal to the judgement of the person scoring the test.
        
        Scorer reliability can be determined by having two persons independently marking the same set of test papers or videotapes of the test and then calculating the correlation between their scores.
      - Stability of testers.
        
        It also may be important to demonstrate that two different trained testers can administer the same test to the same people and not have their personality or other attributes affect the results. In this case, the same procedure is used as for test-retest reliability, but with two different testers administering the two tests, each one giving the test first half of the time.
      - Standard error of measurement.
        
        This statistic permits the interpretation of individual scores obtained in a test.
        
        Because no tests are perfectly reliable, we know that the score an individual receives in a given test is not necessarily a true measure of the trait being measured.
        
        The standard error of measurement tells us how much we can expect an obtained score to differ from the individual’s true score.
        
        Test makers usually provide this statistic in the test manual.
    - - three broad categories of reliability coefficients, namely:
        
        1) coefficients derived from correlating individuals’ scores in the same test administered on different occasions (test-retest coefficients); :red_cross:
        
        need 2 administrations
        
        2) coefficients derived from correlating individuals’ scores for different sets of equivalent items (parallel or equivalent-forms coefficient); :red_cross: and
        
        need 2 administrations
        
        3) coefficients based on the relationship among scores derived from individual items or subsets of items within a test (internal consistency coefficients). :red_cross:
        
        need 1 administration
        
        coefficient ranges from 0.00 (not reliable at all) to 1.00 (perfectly reliable). The higher the coefficient value, and approaching 1, the more reliable is the instrument. That means, the instrument is consistently measuring what it is supposed to measure.
    - - It is feasible through a variety of statistical treatments to quantify the reliability and validity of psychological tests and inventories.
      - It is more difficult, though usually possible, to determine these qualities for some other data-gathering instruments or procedures, such as observation, interview, or the use of questionnaire, in which responses are more qualitative and yield data that are not always readily quantifiable.
    - - In tests that have a high coefficient of reliability, errors of measurement have been reduced to a minimum.
      - Reliable tests are stable in whatever they measure and yield comparable scores on repeated administration.
      - One way to reduce the probability of measurement error and increase reliability is to increase the number of items in a test.
      - An unreliable test yields different measurements each time it is applied, thus, not reliable.
  - - - try to eliminate subjectivity from the judgements they make about the characteristics of samples, or their achievements, or performance.
        Subjectivity lends itself to many interpretations and one interpretation may differ from another.
        Unfortunately, it is not possible to attain complete objectivity.
    - - includes considerations of
        
        length of time it takes for the respondents to answer the instrument.
        
        Clarity of instructions given .
        
        To what extent are the content and the language used in the instrument suitable and readable for the age level of the group being investigated?
        
        For the researchers -- the ease of scoring and of interpreting the results of the data gathered using the instrument
    - - Objectivity and usability of instruments are also checked at the same time when content validity is being examined by a panel of experts.
        However, the researcher needs to prepare and include questions related to the objectivity of the instrument and usability of the instrument for the panel of experts to check on.
  - - - Ease of administration, scoring, and interpretation are important factors in selecting a test, particularly when expert personnel or an adequate budget is not available. This issue is very much related to usability of the instrument. However, many good tests are easily and effectively administered, scored, and interpreted by the classroom teacher, who may not be an expert.
    - - Tests that can be given in a short period of time are likely to be well received by the subjects and to save the time of all those involved in test administration.
        The expense of administering a test is often a significant factor if the testing programme is operating on a limited budget.
    - - Tests that are interesting and enjoyable will be well received by the subjects. Those that are dull or seem silly may discourage or antagonise the subjects. Under these unfavourable conditions the test is not likely to yield useful results.
    - - When psychological tests are used in educational research, one should remember that standardised test scores are only appropriate measures of the traits under consideration.
      - This limitation is inevitable and may be ascribed to a number of possible factors:
        
        Errors inherent in any psychological test — no test is completely valid or reliable.
        
        Errors that result from poor test conditions, inexpert or careless administration or scoring of the test or faulty tabulation of test scores.
        
        Inexpert interpretation of test results.
        
        Inappropriate test for the specific purpose in mind.
  - - - criterion-related validity
        
        broad term that actually refers to 2 different types of validity with different time frames:
        measured by validity coefficient
        
        Predictive validity refers to the usefulness of a test in predicting some future performance, such as the usefulness of the Sijil Pelajaran Malaysia (SPM) in predicting performances in college. ---- eg: If a test is designed to pick out good candidates for appointment as teachers and test scores show a high positive correlation with later actual success in the teaching profession, the test has a high degree of predictive validity, whatever factors it actually measures.
        
        i.e high predictive validity = good indication that the test is useful in predicting future performance
        
        Concurrent validity refers to whether a test is closely related to other measures, such as present academic grades, teacher ratings, or scores in another test of known validity.
        
        THEREFORE it is possible to demonstrate validity of a new measure by demonstrating that it is highly related to another measure of known validity or other important criteria that is known.
        
        example
        
        Looking at the well-known scale of personal adjustment, the self- regulated learning scale.
        
        A paper-and-pencil form was devised using the simple process of ticking responses to printed items on a form. ---This form was administered to a large group at one time and then scored by machine, all with little expense.
        
        The results were so similar to the more time-consuming expensive observational process.
        
        This is the process of establishing concurrent validity, in this case by comparing an observational individual device with an easy-to-administer group instrument.
        
        Similarly, performance tests have been validated against paper-and-pencil tests, and short tests against longer tests.
        
        Through this process, more convenient and more appropriate tests can be devised to accomplish the measurement of behaviour more effectively.
      - construct validity
        
        is the degree to which scores in a test can be accounted for by the explanatory constructs based on a sound theory.
        A construct is a trait that cannot be observed.
        
        hypothesis = that people who have this characteristic will perform differently from those who do not.
        
        Theories can be built describing how learning motivated people behave in a distinctive way.
        
        AND people of high and low motivation can be identified by observation of their behaviour, rating or classifying them in terms of the theory.
        
        A device designed MUST have construct validity to the degree that the instrument’s scores are systematically related to the judgements made by observation of behaviour identified by the theory as dominant.
        
        Intelligence tests are good examples of tests that require adequate construct validity.
        
        Different intelligence tests are based on different theories, each test should be shown to measure what the appropriate theory defines as intelligence.
        
        Construct validity is VERY important for personality and aptitude tests. : :no_entry: :no_entry:
        
        example --- if research is on construct such as learning motivation
      - content validity
        
        refers to the degree to which the test items actually measure, or are specifically related to, the traits for which the test was designed.
        
        It shows how adequately the test samples the knowledge and skills a student is expected to master.
        
        is based on careful examination of course textbooks, syllabi, objectives, and the judgments of subject matter specialists.
        
        The criterion of content validity is often assessed by a panel of experts in the relevant field who judge its adequacy, but there is no numerical way to express it.
        
        is particularly important for achievement tests but not very important for aptitude tests.
        
        Next, the test items should be shown to be representative of this construct or variable examined.
        
        These two steps in demonstrating content validity can be done in the test development stage or by using a panel of experts after the test has been developed.
        
        However, for users of tests there is still the issue of whether the test content is appropriate for the persons to be tested.
        
        For instance, a reading test that relies heavily on phonic skills may have good content validity for schools where that approach is emphasised but not for schools taking a different approach to teaching reading.
        
        In order to demonstrate content validity, the test maker needs to first define the construct or variable that could be included in the test.
        
        Content validity is the test measuring the content that it is supposed to measure.
    - - Validity coefficient (rxy) = The coefficient correlation between test scores and criterion
        
        -Like any correlation coefficient, the size of a validity coefficient is influenced by the strength of the relationship between test and criterion and the range of individual differences in the group.
        
        i.e the nearer the coefficient value is to 1.00, the stronger the evidence to show that the test is useful for the intended purpose.
        
        Validity index indicates the extent to which the instrument is useful as a predictor or as a substitute measure.
        
        If a test is shown to have a high correlation with a future criterion, then the test can later be used to predict that criterion.
        
        As mentioned above, criterion-related validity is especially important for scholastic aptitude tests because they are generally used to predict future academic performance.
        
        Thus, any test which is used to make selections needs to have criterion-related predictive evidence.
        
        However, in order to accumulate predictive evidence, the researcher needs time and patience, perhaps several years before he can confidently say that the performance of a measure is useful for predicting the success of a criterion.
        
        Concurrent criterion-related validity evidence is important in tests used for classifying, certification, or diagnosis.
        
        A researcher who is seeking concurrent validity evidence for a new intelligence test or a psychiatric test needs to compare and correlate the new test to existing well-established instruments.
        
        If the correlation is substantial, it is fair to conclude that the new instrument measures the same construct as the established instrument, thus, it can be used in place of the established instrument.
        
        However, as a general rule, one can only do the substitution if the validity coefficient is substantially high.
    - - For example, tests of intelligence are designed to predict academic achievement and are based on a psychological theory or construct. Thus, this type of test needs to demonstrate construct and predictive validity but not necessarily content validity.
        
        On the other hand, most achievement tests need to demonstrate only content validity.
  - - - i) get the data themselves,
      - ii) directly from the subjects of the study, and
      - iii) from others referred to as informants (they are trained and knowledgeable about the subjects under study).
  - - - Content validity is not critical for this type of test.
        
        Predictive validity and construct validity are usually important here.
        
        For standardised tests that are group-administered paper-and-pencil with multiple-choice items (e.g., a group intelligence test), the only forms of reliability that are critical are test-retest, stability of test items, and the standard error of measurement.
        
        If multiple forms are available, then one may add the need for stability of item samples.
        
        For tests that also have items that require some degree of subjectivity (e.g., individual intelligence tests in which the tester interacts individually with the person taking the test), inter-scorer and inter-tester reliability are needed.
    - - Content, predictive, and construct validity are usually important here.
        
        However, the actual purpose of the researcher will determine the types of validity needed.
        
        Test-retest, stability of items, and standard error of measurement are the relevant forms of reliability here.
        
        Stability of scores and inter-tester, depending on the nature of the specific test.
        
        Other forms depend on the type of test, multiple choice, interview, and so on.
    - - Content validity is critical for this type of test.
        
        Concurrent validity might be used to help establish a new achievement test validity.
        
        Other forms of validity are probably not necessary but would only be important if relevant for the test use.
        
        For standardised tests, and assuming the test is group-administered paper-and-pencil with multiple-choice items, the only forms of reliability are critical area test-retest, stability of test items, and the standard error of measurement.
        
        If multiple forms are available, one should add the need for stability of item samples. For tests that also have items requiring some degree of subjectivity (e.g., essay answer), inter-scorer or inter-rater reliability is needed.
    - - Construct validity is critical here.
        
        Inter-scorer and inter-tester forms of reliability are also important.
        
        Predictive validity is also probably important.
        
        Test-retest reliability may also be pertinent.
    - - As for establishing a satisfactory level of content validity, the truly significant incidents of behaviour must be identified and sampled.
        
        Supplementing the knowledge and skill of the researcher may increase validity.
        
        Recommendation of experts in the field may help in selecting a limited number of observable incidents whose relationship to the qualities of interest is based on sound, established theories.
        
        Criterion-related and construct validity may also be necessary, depending on the purpose of the study and inferences made regarding behaviours.
- - - - Internal validity IV
        
        -refers to the inferences about whether the changes observed in a dependent variable are, in fact, caused by the independent variable(s) rather than by some extraneous factors I.E the inferences to confirm that independent variable affected dependent variable ?
        
        -It refers to the degree to which observed differences on the DV are a direct result of manipulation of the IV - not some other variable.
        
        emphasises questions such as: did the experimental treatment cause the observed change in the dependent variable or did the independent variable really make a difference?
        
        -These variables are considered threats since, unless they are controlled, they may very well produce an effect that could be mistaken for the effect of the experimental treatment.
        
        IV focuses on threats, or rival explanations or extraneous variables, which are not part of the independent variable, but could influence the findings of an experimental study.
        
        -Campbell and Stanley (1963) - eight extraneous variables that frequently represent threats to the internal validity of a research design. --ways to overcome these threats are through random selection of subjects, assigning of the subjects to treatments, and control of other variables.
        
        statistical regression
        
        occurs when groups are selected based on extremely high or extremely low scores -- tend to regress towards the mean on retesting.
        -These situations clearly produce an effect that may be mistakenly interpreted as an experimental effect.
        
        instrumentation
        
        -The unreliability or lack of consistency in measuring or calibrating instruments that may result in an invalid assessment of measured dependent variable eg if the researcher uses two different tests, one for pretest and one for posttest, and the tests are not of equal difficulty.
        
        threat with respect to the administration of instrument.
        
        differential selection of subjects,
        
        usually occurs when already formed groups are compared.
        -There may be important differences between the groups before the application of the experimental treatment.
        
        testing
        
        = pretest sensitisation, --> which refers to the threat of improved scores on a posttest as a result of having taken a pretest
        
        maturation
        
        refers to biological or psychological changes such as physical, intellectual, and emotional changes that occur within the subjects simply as a function of the passage of time
        
        History
        
        refers to the occurrence of specific events or conditions, other than the experimental treatment that may occur between the beginning of the treatment and the posttest measurement of the dependent variables that are not part of the experimental treatment.
        
        experimental mortality,
        
        occurs when there is attrition or differential loss of respondents from the treatment and control groups.
        
        selection-maturation interaction
        
        -This threat occurs if participants selected into treatment groups mature at different rates during the study.
        i.e the effect of differential selection may interact with the effects of maturation in such a way that the combination results in an effect on the dependent variable, is mistakenly
        attributed to the effect of the experimental treatment.
        
        A. experimenter effect
        
        is related to unintentional effects that the researcher has on the study.
        
        B. subject effect
        
        Subjects’ attitude --change their behaviour because they realise that they are participating in an experiment
        
        C. Diffusion
        
        may occur when subjects in one group, usually the experimental group communicate information about the treatment to subjects in the control group, in a way that influences the behaviour of the subjects in the control group.
      - External validity (EV) = ecological validity
        
        Campbell and Stanley (1963) identified 4 external validity threats that may interfere with the generalisability of the research.
        
        2.interaction of selection and X
        
        Also known as selection-treatment interaction, this occurs when the research findings apply only to the unrepresentative groups involved and are not representative of the treatment effect in the extended population.
        
        reactive arrangements
        
        precludes generalisation about the effect of the experimental variable on persons who were exposed to it in non-experimental settings.
        It is associated with the way in which a study is conducted and the feelings and attitudes of the subjects involved.
        
        testing and X (treatment)
        
        = pretest-treatment interaction, --occurs when subjects respond or react differently to a treatment because they have been pretested.
        
        multiple-X interferences
        
        likely to occur if the same subjects receive more than one treatment in succession.
        there is a carryover effect from an earlier treatment, making it difficult to assess the effectiveness of a later
        treatment because the effects of prior treatment are not usually erasable.
        
        is the degree to which study results are generalisable or applicable, to groups and environment outside the experimental setting
        
        Thus, external validity focuses on threats that disallow the results of a study to be generalised to other settings or groups.
        
        -If research results can never be generalised to similar situations outside the experimental setting, then no one can benefit from the research. ----
        
        -An experimental study can contribute to theory and practice only if the results are replicable and generalisable to other places and groups.
        
        If the result of a research is not replicable in other settings by other researchers, the study has low external or ecological validity.
  - - - if control over the extraneous variables is done, but random assignment is NOT done,
    - - control over other influencing factors or variables is done and the subjects involved in the study are randomly assigned to either the control group or the experimental group.
        
        7 diff models
        
        1) Randomised posttest only control group design;
        
        2) Randomised pretest-posttest control group design;
        
        3) Randomised Solomon three-group design;
        
        4) Randomised solomon four-group design;
        
        5) Randomised matched subjects posttest only control group design;
        
        6) Randomised matched pretest-posttest control group design; and
        
        7) Factorial design.
    - - if the experimental design has little or no control of the extraneous variables, the research becomes a weaker pre-experimental design.
        
        3 diff models
        
        1) One-shot case study design;
        
        only one group is involved.
        
        This single group is exposed to a treatment and the dependent variable is subsequently observed or measured.
        
        The effect of the treatment is assessed.
        
        2) One-group pretest-posttest design; and
        
        3) Static group comparison design.
- - - - use statistical data - 3 steps
        
        determining the sample characteristics,
        
        inferring the characteristics of the population based on the sample.
        
        randomly selecting a sample from a defined population,
  - - - three main threats to internal validity in non-experimental research:
      - Mortality — this arises in longitudinal studies as there would be drop outs in such studies. Hence to overcome the mortality threat, a researcher will need to delete all the data related to drop-out subjects.
      - Location — a location threat occurs when a researcher collects data in places that may affect responses. For example, a survey on the use of library facilities being conducted at the main entrance of the library on a Saturday. Another example is when a researcher conducts an interview on attitudes towards religion in a particular religious place such as mosque, church or temple.
      - Instrument and instrument decay — this is pertaining to lack of valid and reliable instruments being used by the researcher. As such, the information obtained may be biased or not valid. In other situations, a researcher may employ interviewers who are tired and in a rush when conducting the survey. This may lead to futile use of the research instrument resulting in a systematically biased set of data.
  - - - census
        
        survey
        
        characteristics
        
        1) Information is collected from a group of people in order to describe some aspects, e.g., abilities, opinions, attitudes, beliefs and/or knowledge, of the population of which that group is a part of,
        
        2) The main mode of collection of information is through asking questions, and answered by members of the group,
        
        3) Information is collected from a sample rather than from every member of the population.
        
        time dimension - according to the time of data collection
        
        cross-sectional surveys.
        
        1 more item...
        
        longitudinal surveys
        
        7 more items...
    - - also known as associational research.
      - the relationships among two or more variables are studied without
        any attempt to influence them.
      - In its simplest form, correlational design is used to identify the relationship between two variables, even though it is common to study more than two variables.
        
        -No manipulation of variables occurs in a correlational study.
        
        -This design provides answers to how much variations are there in the independent and dependent variables that occur in a natural setting.
        
        -The purpose of the study is to determine the relationship between the two variables, and to what degree the relationship exists.
- - - - 2 important applications
        
        Randomly selecting individuals to form a sample group of individuals for observation who are representative of the population about which the researcher wishes to generalise, or
        
        Equating experimental and control groups in an experiment by randomly assigning subjects to the two groups, is the best method of providing for their equivalence.
        
        error in a statistical analysis arising from the unrepresentativeness of the sample taken. ==the deviation of the selected sample from the true characteristics, traits, behaviors, qualities or figures of the entire population.
        
        sampling error,
        
        Characteristics of successive random samples drawn from the same population may differ to some degree, but it is possible to estimate their variation from the population characteristic and from each other.
        The variation, known as sampling error, does not suggest that a mistake has been made in the sampling process.
        
        Types of sampling procedure
        
        probability sampling -- 4 types
        
        simple random sampling = fishbowl technique
        
        each member of the population under study has an equal chance of being selected, and the probability of each choice is unaffected by the selection of other choices from the population (independent of any other choice).
        
        can use table of random numbers
        
        systematic sampling
        
        is a sample obtained by selecting every n th subject in a population.
        
        10 more items...
        
        stratified random sampling
        
        is a sample selected so that certain characteristics are represented in the sample in the same proportion as they occur in the population.
        ---a type of sampling that maintains a smaller homogeneous group in order to ensure a more accurate representation.
        
        5 more items...
        
        cluster sampling = area sample
        
        sample is obtained by using groups as the sampling unit rather than individuals. - is a variation of the simple random sample
        
        is appropriate technique when: the population of interest is infinite, and when a list of every individual member of the population does not exist, or when the geographic distribution of the individuals is widely scattered. - i.e when it is impossible or impractical to obtain and prepare a list of the unit of analysis.
        
        2 more items...
        
        non-probability sampling -- 2 types use whatever subjects are available rather than following a specific and systematic subject selection process.
        may produce samples that do not accurately
        reflect the characteristics of a population of interest.
        may lead to unwarranted generalisations and should not be used if random selection is feasible.
        
        convenience sampling
        
        5 more items...
        
        purposive sampling
        
        3 more items...
  - - - defined as all members of any well-defined group of individuals, events or objects, which have one or more characteristics in common.
        
        -The population is of interest to the researcher, since it is the group to which the results of the research will be ideally generalised.
        
        -The population may be all the individuals of a particular type, or a more restricted part of that group.
        
        2 types of population
        
        Target population is a group of individuals (or a group of organizations) with some common defining characteristics, i.e, the actual population to whom the researcher would ideally like to generalise his/her findings, but is usually rarely available.
        
        Accessible =available population is the population that the researcher can realistically select from, based on readily available list and to whom the researcher is entitled to generalise.
        
        If a population is broadly defined (like ‘all teachers’ as in the above example), external validity or generalisability will be maximised, although such a broad definition may make obtaining a representative sample difficult, and may require a large sample size.
      - -A measured value based on data collected from a population is called a parameter.
    - - -A measured value based on data collected from a sample is called a statistic.
      - comprises individuals, items, or events selected from a population. It is a small proportion of a population selected for observation and analysis and it is representative of the population.
  - - - 100 for a descriptive study
      - 50 for a correlational study
      - 30 in each group for experimental and casual-comparative study
      - The use of 15 subjects per group should probably be replicated
    - - There are many variables
      - Only small differences or small relationships are expected or predicted
      - The sample will be broken down into subgroups
      - The sample is heterogeneous in terms of the variables under study
      - Reliable measures of the dependent variable are unavailable