For more information
Call 1-800-351-0222

DATA COLLECTION.
  Term Paper ID:19904
Essay Subject:
Concepts & methods, advantages & disadvantages. Surveys, interviews, questionnaires, reliabilty & measurement, tabulation/analysis, goals. Tables.... More...
31 Pages / 6975 Words
10 sources, 113 Citations, APA Format
$100.00

More Papers on This Topic


Paper Abstract:
Concepts & methods, advantages & disadvantages. Surveys, interviews, questionnaires, reliabilty & measurement, tabulation/analysis, goals. Tables.

Paper Introduction:
DATA COLLECTION: SURVEY, INTERVIEW, AND TABULATION/ANALYSIS This research provides an indepth review of selected aspects of the data collection process used in the conduct of research studies. The aspects of the data collection process reviewed are (1) the survey method, (2) the interview method, (3) questionnaire development, and (4) the tabulation/analysis of data collected through the application of the survey and interview methods. The Survey Method Objective and descriptive data from subjects are required in the conduct of many types of research studies. The use of surveys to collect such data is probably the most efficient of the various means available (Dyer, 1985, p. 24). The survey is also the most effective means of collecting objective data from

Text of the Paper:
The entire text of the paper is shown below. However, the text is somewhat scrambled. We want to give you as much information as we possibly can about our papers and essays, but we cannot give them away for free. In the text below you will find that while disordered, many of the phrases are essentially intact. From this text you will be able to get a solid sense of the writing style, the concepts addressed, and the sources used in the research paper.


Semistructured interviews demand greatertraining of the interviewer than is required with structured interviews. The central tendency theorem, more commonly referred to asthe central limit theorem, holds that the totals (and therefore the means)of random samples will be normally distributed no matter what thedistribution in the population is like, provided only that the samples arelarge enough. The reliability of a data collection procedure is a contributor tothe validity of that procedure, rather than the other way around (Emory,1988, pp. 2) There is no assurance that respondents understand thequestions asked. The initial combination of variables constitutes the first principalcomponent (Kim, 1975, pp. These weightsare referred to as factor coefficients, or factor loadings. Advantages: 1) Less costly than face-to-face interviews. Tabulation/Analysis of Data Tabulation refers to the arrangement of data in a table or othersummary format (Zikmund, 1992, p. 7) Summarization of findings is often difficult. 298). A simple tabulation includesmaking a response count and put them into a frequency distribution table.Tabulation also includes the measurement of central tendency in data(Zikmund, 1992, p. b. It often represents an impracticalapproach to a solution to a problem (Kim, 1975, pp. 468-514). ornearly so. Responses should never be summarized or paraphrased. Group interviews: a. 2. SPSS: Statistical packagefor the social sciences. While multiple choice questions provide therespondent with a greater choice of responses, they may also demand morequalified responses than a respondent is willing to make (Pfaffenberger andPaterson, 1987, p. Surveys are dependent on the direct communication with personspossessing the characteristics or factors to be measured (Isaac andMichael, 1991, p. Six specific steps are involved in the planning ofa survey (Pfaffenberger and Paterson, 1987, pp. Factor analysis seeks to interpret each factor identifiedaccording the variables included in the factor. A negative correlation,regardless of how high it was, would not be an indication of the validityof an instrument; for an instrument to be valid, its validity coefficientmust be positive and it must be high (Thorndyke and Hagen, 1969, pp. In formulating questions for the instrument, it is also necessary todecide whether to disguise the study objectives (Emory, 1988, p. 4) Similar to brainstorming, in that this method stimulatesthe activity of others in the group. The basic theoretical principle upon whichinferential statistics are based is probability. 468-514).The two basic types are (1) R type analysis, which involves the correlationof pairs of scale items, and (2) Q type analysis, which involves thecorrelation of pairs of individuals (Kim, 1975, pp. 4. In bootstrap analysis, different sets ofweights are assigned to variables, as a means of identifying the principalcomponents. One of thedeficiencies of the brainstorming technique is that, by design, it isintended to produce somewhat superficial ideas, which must later be refinedand developed. Further, there are twodifferent approaches--group and individual--to the conduct of interviews. The Eigenvalues represent the sums of the factorvalues, and, when the Eigenvalues are divided by the number of variables,the result is an estimate of the amount of the total variance which isexplained by the factor (Kim, 1975, pp. b. New York: John Wiley & Sons, Inc. The data collected through theconduct of a survey must be amenable to expression in numerical terms(Isaac and Michael, 1991, p. Schaum's outline series of theory andproblems of probability and statistics. The values in the tableunder "h2" are referred to as commonalities, and are representative ofestimates of the variance in each variable which is explained by thecombination of the factors (Kim, 1975, pp. Factor analysis. (1992). 128). 5) An extended geographic coverage is feasible. The planning and execution of thesurvey must assure appropriate content coverage and sound, efficient data collection (Isaac and Michael, 1991, p. Further, there are twodifferent approaches--group and individual--to the conduct of interviews(Emory, 1988, pp. The survey is also the most effective means ofcollecting objective data from subjects, because the responses can bewithout attribution (Dyer, 1985, p. The second principal component isdefined as the best linear combination of variables which accounts for thevariance remaining subsequent to that explained by the first principalcomponent (Kim, 1975, pp. In the test of significance, if thechi square is significant at the selected level of probability, theinference is that the residual matrix retains significant levels ofcovariance. 2) Databases are often incomplete and inaccurate. Respondents must be motivated to accept their assigned roles, andto fulfill the requirements of those roles (Emory, 1988, pp. Maximum likelihood factor analysis is used to identify the relevantfactors distinguishing subjects and groups in a sample (Kim, 1975, pp. 2) Inexpensive. An introduction to statistical methods and dataanalysis. While multiple choice questions provide the respondent with agreater choice of responses, they may also demand more qualified responsesthan a respondent is willing to make. The initial step involved in factor analysis is the construction ofa matrix of correlation coefficients between all of the variables involved(Kim, 1975, pp. 18). 542). 468-514). A decision must be made with respect to the number of subjects tobe surveyed. 468-514). As the data presented in Table 2 indicate, the rotatedfactor loadings provide far stronger explanations for each of the variablesthan did the unrotated factor loadings presented in Table 1. S. 468-514). 17). In maximum likelihood factor analysis, the assumption is made thatvariables stem from a designated number of factors (Kim, 1975, pp. (4th ed.). The sampling procedure mustassure that the characteristics of the research sample closely approximatesthe characteristics of the population from which it was drawn (Isaac andMichael, 1991, p. 2. 4) Self-administering. Individuals selectedfor a brainstorming group should be (1) from diverse backgrounds, (2) mixedby gender, where possible, and (3) from differing hierarchical levels of anorganization. 468-514). William. 2-4, 43-51, 1 7-117; Spiegel, 199 , pp. 4) Telephone interviewing is often confused by respondentswith sales pitches. Thus, each of the new components, in turn, account fora maximum amount of variance in each of the variables measured. b. 138). 5) Susceptible to personality conflicts between interviewerand respondent. A survey is a process of collecting data from existing populationunits, with no particular control over factors that may affect thepopulation characteristics of interest in the study. 4) Permit baseline comparisons. 3. The Survey Method Objective and descriptive data from subjects are required in theconduct of many types of research studies. Questionnaire Development Questionnaire development requires a brainstorming of analysis needsto assure clarity and comprehensiveness (Isaac and Michael, 1991, p. Bootstrap analysis is most often used in the analysis of variability. D., & Cohen, H. (1988). In the common factor model, the factors are divided into two groups:(1) the common factors themselves, which consist of those factors whichcontribute to two or more of the variables; and (2) non-common factors,which are summarized into what is called a unique factor, which containsall of the remaining scores necessary to complete the prediction of avariable (Kim, 1975, pp. Brainstorming groups are generally six-to-nine in number, and groupslarger than 12 seldom produce satisfactory results. 468-514). H. (3rd ed.). Disadvantages: 1) May intimidate and suppress individual differences. Reliability, as opposed to validity, is concerned with the accuracyand precision of a measurement procedure. 3) Changing record-keeping practices may make year-to-yearcomparisons impossible. 468-514). Surveys contact only those respondents who are both accessibleand cooperative (Isaac and Michael, 1991, p. for perfect correlation, or reliability,to . Pretests arealso referred to as pilot tests. New York: McGraw-HillBook Company. Advantages: 1) Nonreactive. Theunstructured approach provides the respondent with great freedom inexpression; however, this approach is the most vulnerable to subjectivebias. The interpretation of factor loadings is a largely subjective process(Kim, 1975, pp. Exploring marketing research. Onceagain consider the desire to measure the height of a number of individualsin feet and inches. 468-514). The testis given twice to the same group, and the results are then correlated.High correlations are viewed as indices that the test is reliable. Statisticalmethods for business and economics. The respondent's own words should be recorded (Zikmund, 1992, p.466). Bootstrap analysis is a method for the estimating the statisticalaccuracy of r from the data in a single, relatively small sample. Because ofthese characteristics, surveys pose risks for the researcher because: 1. Mailed questionnaires: a. Conclusions must be drawn from the results of the conduct of thesurvey. 468-514).In such instances, a third step in factor analysis is performed. Survey of records: a. The second step requires the construction of anew set of variables, on the basis of the relationships identified in thematrix of correlations. As acorrelation coefficient, a perfect validity coefficient would be +1. For mostsurveys, where an external standard of measurement may not be available,the pretest serves to establish the validity of the questionnaire. As is true for validity coefficients, the obvious question concernshow high reliability coefficients should be for a procedure to beconsidered reliable. Therefore, thecentral tendency theorem assumes a high-level of importance in businessdecision-making hypothesis testing. 468-514). 468-514). 17). 3 C .68 .1 D . Respondents must understand their roles in the interview process(Emory, 1988, pp. Thus,hypothesis testing is applied to sample data in business decision-makingapplications. Surveys tend to engender a proneness to agree with positivestatements or questions (Isaac and Michael, 1991, p. 468-514). The principal advantage of bootstrap analysis in principalcomponent analysis is that the procedure, in theory, enhances the validityof the results, when only a relatively small sample may be used. 17). 3. 138). 5. 155-193). In maximum likelihood factor analysis, the emphasis is on theestimation of population parameters from sample statistics (Kim, 1975, pp.468-514). 138). Ott, Lyman. References Dyer, W. The framework of the survey must be designed. If abattery of tests is being used, a multiple correlation of .5 may beworthwhile as a predictor, particularly if large numbers of cases areinvolved. To illustrate these points, consider, again, the case ofthe instrument calibrated in feet and inches which may be used to measurethe height of a group of individuals. A brief assertion of understanding and interest at the beginningof the interview (Emory, 1988, p. Another useful tabulationformat is cross-tabulation. Further, an instrument cannot be valid if it is notalso reliable; however, just because it is reliable, it is not, ofnecessity, valid. Thereare two ways by which the variables may be rotated--orthogonal and oblique(Kim, 1975, pp. 468-514). 3) Reveals group interaction patterns. From this core, theinterviewer may branch out to explore one or more points in greater depth.Semistructured interviews demand greater training of the interviewer thanis required with structured interviews. Confidenceregions for INDSCAL using the jackknife and bootstrap techniques.Psychometrika, 49, 475-491. Where problems are encountered in the application of factor analysisto reduce data in the tabulation/analysis process, the bootstrap proceduremay be applicable. These principles are as follows: 1. These steps are asfollows: 1. The purpose of the brainstorming technique is to develop a largenumber of new potential solutions to a specific problem. To describe events. anda complete absence of validity would be . 294-195). Further, the conditions under which bootstrap mayfail are predictable. 4. Pfaffenberger, Roger C., & Paterson, James H. It is with respect to samples and their characteristics, and theiruse in hypothesis testing, that the central tendency theorem issignificant. From the quantity of ideas,combinations and improvements are sought. Clarificationand elaboration of responses are permitted only within narrow limits.Structured interviews are typically factually oriented, aimed at specificinformation, and brief in duration. 2. The quality of any statistical analysis can be only as good as is thequality of the data upon which it is based; thus, the reliability andvalidity of data collected for use in statistical analysis is of paramountimportance (Ott, 1988, pp. 3) There is no assurance that the addressee is theindividual who actually responds to the questionnaire. Bootstrap is a methodology advanced as a means ofalleviating the problems associated with factor analysis, while retainingthe efficacy of factor analysis. To infer causes or future events. On the other hand, if an instrument does measure the height of thesame individual as being the same in measurement-after-measurement, theinstrument is a reliable instrument. 2) Can be conducted in either daytime or evenings. 4) Respondents are more at ease, as they are in their ownhomes or offices, and, thus, tend to be more candid in their responses. The bootstrap procedure is capable of estimating variability for anyprincipal component, without assuming that the data have a normaldistribution. b. Thus, if itwere desired to measure the height of a number of individuals in terms offeet and inches, it would be necessary to have a measuring instrument whichwas calibrated in feet and inches. Insight into impact: Strategies for interpersonaland organizational change. The interviewer must be able to stimulate the respondent to answer.The preferred process is called probing, and includes the following actionson the part of the interviewer: 1. Principal component procedures are developed within theframework of the multivariate linear model (Kim, 1975, pp. The initial data generated by principal component analysis will looksomething like that presented in Table 1, which may be found on thefollowing page. 3. 14-15). 6) Requires skillful, trained interviewers. Repeating the respondent's reply (Emory, 1988, p. 6) Only factual data are available in databases, as opposedto values or attitudes. These data are referred to as unrotated factors (Kim,1975, pp. Disadvantages: 1) Individual interviews are both expensive and time-consuming to conduct. Handbook in researchand evaluation. The primary purpose offactor analysis, thus, is exploration; an exploratory process to identifypotentially fruitful areas of research. Measurement andevaluation in psychology and education. 6 .7 E .13 .77 F . The unstructured interview is most frequently used in situationswhere the information sought is either highly personal or potentiallythreatening to the respondent. In mostinstances, if the data are to be used effectively, it is necessary toreduce the number of explanatory variables to more manageable proportions. 468-514). 468-514). As istrue of validity coefficients, reliability coefficients, as correlationcoefficients, may range from +1. 3 1). 4) May be misleading unless information is availablepertaining to the way in which the data base was compiled. b. Structured interviews are typicallyfactually oriented, aimed at specific information, and brief in duration(Isaac and Michael, 1991, p. If time is critical, the interviewer should use some sort ofshorthand during the initial recording of the responses, and write completeresponses at a later time. 138). The most frequently used of theseprocedures is principal component analysis (Kim, 1975, pp. 128-134). Homewood,Illinois: Dow Jones/Irwin. 3) Flexible and adaptable. 2) May annoy respondent when interviewer background--racial, ethnic, or socio-economic--is different from that of therespondent. 3) Permit unlimited callbacks to respondents. Factor analysis attempts to summarize many variables in to a fewfactors. 4. 294-195). Disadvantages: 1) Databases required may be inaccessible because ofconfidentiality. Thus, when the Table 2 Rotated Factor Analysis Matrix (N = 1 )------------------------------------------------------------- Rotated Factors -----------------Variable I II-------- ----- ------ A .79 .15 B .75 . Two axes, separated by 9Odegrees at a common intersection, are used to cut this space (Kim, 1975,pp. As indicated earlier in this discussion, one common measure ofreliability is the test-retest correlation-coefficient technique. Individual interviews: a. A neutral question or comment (Emory, 1988, p. There are two basic types of factor analysis, although there are anumber of different factor analysis procedures (Kim, 1975, pp. In principal factoranalysis, by contrast, the presence of population parameters is assumed(Kim, 1975, pp. The orthogonal procedure is the most widely used(Kim, 1975, pp. 468-514). Factor analysis studies the correlations of a large number ofvariables by clustering those variables into factors in a way that thevariables included in each factor are highly correlated. Clarification and elaboration of responses arepermitted only within narrow limits. 468-514). 13 -132): 1. The choice in this instance is between a census of thepopulation of interest, or a sample of that population (Pfaffenberger andPaterson, 1987, pp. 3. Reliabilityrefers to the accuracy and precision of a of data collection procedure.Validity, thus, is the extent to which differences found through aparticular data collection procedure reflect true differences among thosevariables being measured, while reliability refers to the capacity of aninstrument to yield similar measurements under similar conditions. (5th ptg.). 294-195). Samples which are too small willnot always yield valid results. The three general objectives of factoranalysis, as follows (Kim, 1975, pp. Factor analysis is a general descriptor for a group of specificcomputational procedures (Kim, 1975, pp. 5. Theassumptions required for the full components models are that: (1) thevariables may be calculated from the factors by multiplying each factor bythe appropriate weight, and then summing across all factors; and (2) allfactors have means of zero and standard deviations of one (Kim, 1975, pp.468-514). These three ways are asfollows: 1. In each instance, thestatistical significance of the results must be established (Kim, 1975, pp.468-514). 468-514). The principal benefit of the brainstorming technique isthat it fosters within an organization an atmosphere of free-thinking. . 298). Theaspects of the data collection process reviewed are (1) the survey method,(2) the interview method, (3) questionnaire development, and (4) thetabulation/analysis of data collected through the application of the surveyand interview methods. The data sought through the conductof the survey must be observable and they must be capable of explicitmeasurement (Isaac and Michael, 1991, p. 468-514). The major difficulty with free response questions is theclassification of responses, which, more often than not, is difficult,arbitrary, and time-consuming. Spiegel, Murray R. Zikmund, William G. 468-514). 5. Free response questions permit respondents to answer in any way theychoose. Bootstraprelies heavily on correlation analysis, as does traditional factoranalysis. A survey must be systematic. The major difficulty with free response questions is theclassification of responses, which, more often than not, is difficult,arbitrary, and time-consuming (Pfaffenberger and Paterson, 1987, p. Brainstorming sessions are typically short--4 minutes or under. Factoranalysis is a widely used statistical procedure with data-reductioncapabilities that is used to determine the underlying pattern ofrelationships among a set of variables or conditions which may be taken assource variables, accounting for the observed interrelations in the data. 468-514). Advantages: 1) Personalized. A pretest of the questionnaire instrument is required to assure thatthe data of interest are actually generated by the questionnaire in a validand reliable way (Pfaffenberger and Paterson, 1987, p. 128). Second, there must be a relatively small variance acrosssamples (Kim, 1975, pp. for a completely unreliable instrument. Multicollinearity is eliminated in principal component analysis (Kim,1975, pp. (1988). Decision theory providesmanagers and administrators with knowledge about events and relationshipswhich reduce the level of uncertainty in the data upon which decisions foran organization are based. In the common factor model: (1)variables are estimated from the common factors, by multiplying each factorby the appropriate weight, and summing across factors; (2) common factorscorrelate zero with unique factors; (3) unique factors correlate zero withone another; and (4) all factors have zero means and standard deviations ofone. 468-514). 468-514). A measurement procedure is,thus, reliable to the extent that it supplies consistent results. 468-514). 2. 468-514).Through principal component analysis, the initial set of variables istransformed into a new set of composite variables, or principal components,which are referred to as factors (Kim, 1975, pp. The unstructured approach provides therespondent with great freedom in expression; however, this approach is themost vulnerable to subjective bias. In orthogonal rotation, the variables are, ineffect, plotted in an "n" dimensional space. Theaccuracy of the maximum likelihood method increases, as the sample sizeincreases. Tests of statistical significance have been developed for the maximumlikelihood factor analytic procedure. (1975). Theresult is an orthogonal transformation, in which each of the originalvariables are describable or explainable in terms of new principalcomponents (Kim, 1975, pp. 468-514). If recording errors are made, the result is aresponse error-- discrepancies between recorded and actual responses(Emory, 1988, p. For use ininferential statistics, this probability must be statistically independent. When the principal components analytical procedure is applied tofactor analysis, it becomes principal factor analysis (Kim, 1975, pp. Dichotomous questions ask the respondent to selectbetween to possible responses--usually "yes" or "no." Multiple choicequestions ask the respondent to select from among a number of potentialresponses. San Diego, California: EdITS Publishers. 468-514). Kim, J-O. 2. 4. Irwin, Inc. 468-514). G., Steinbrenner, K., & Bent, D. The full component model is based on the perfect calculation of thevariables from the components, while the common factor model also considerssources of variance which may not be attributable to common factors. Under this assumption, there are an infinite number of positions towhich the factors could be rotated to satisfy the likelihood function.Each of these positions, it is assumed, would provide the same estimatedreproduced correlation matrix. The procedure seeks to generalize from a sample of individualsto a population of individuals (Kim, 1975, pp. Respondents must have access to the needed information (Emory,1988, pp. Structured interviews follow a well-defined format. The design may include threetypes of questions-- dichotomous, multiple choice, or free answer(Pfaffenberger and Paterson, 1987, p. The full component model produces exact relationships (Kim, 1975, pp.468-514). Throughthe bootstrap procedure, an enormous number of different samples may begenerated from a single sample. Isaac, Stephen, & Michael, William B. These conclusions must relate directly to the problem statementdeveloped for the survey (Pfaffenberger and Paterson, 1987, pp. 468-514). 3. Theproblem to be considered is not revealed prior to the beginning of thesession, and the problem is clearly stated in a relatively narrow context. Descriptive statistics describe theperformance or activity of one group or class, without attempting to makegeneralizations about other groups or classes. 468-514). The procedure attempts to generalize from asample of individuals to a population of individuals. Interviews may be unstructured, semistructured, or structured (Isaacand Michael, 1991, p. 14-15). The most widely used method of decomposition to identify factormatrices is principal factor analysis (Kim, 1975, pp. 468-514). 2. For practical purposes, however, the process istypically terminated after a few principal factors have been identified. Sample selection must be generally random, or thebootstrap procedure will not provide valid results for the factor analysismodel. 3) Can be well designed, simple, and clear. Theprocedure is concerned with estimating the population correlation matrix.The method itself does not attempt to uniquely locate the factors (Kim,1975, pp. 128). The test/retest aspect of pilot testing provides for thereliability of a survey questionnaire. Disadvantages: 1) Unlisted telephones cause validity problems. If, however, this identical time-after-time measurement is not in agreement with that yielded by apreviously validated external standard, the reliable instrument is not avalid instrument. Bootstrap, however, contrives thousands of samples and, inturn, correlations, from a single sample. A survey must be representative. In statistical applications in business decision-making, probability is afunction of sampling, as well as of relationships, because it is usuallynot feasible, or even possible, to measure an entire population. Three steps are involved in the construction of a questionnaire--designing the instrument, conducting a pretest, and editing the results.The design may include three types of questions--dichotomous, multiplechoice, or free answer. 14-15). (4th ed.). 542). A cross-tabulation is a contingency table displaying a joint-frequency distribution of the observations applicable to two or morevariables (Zikmund, 1992, p. Thismethod extracts the maximum amount of variance possible by a given numberof factors. It may also be used, however, in more sophisticated applications, such asprincipal components analysis. Principal components analysis is concerned with the examination ofthe total variance among all of the variables, while principal factoranalysis is concerned with the examination of the commonality of all of thevariables (Kim, 1975, pp. 4. First, the sample must be large (Kim, 1975, pp. The maximum likelihood method of factor analysis is a procedurewhich obtains the best estimates of factor loadings from informationprovided by the correlation matrix, and is based on an assumption that thevariable is explained by the designated number of factors (Kim, 1975, pp.468-514). Surveys, thus, are reactive in nature. 2) Permits in-depth, free responses. 468-514). The unstructured interview is mostfrequently used in situations where the information sought is either highlypersonal or potentially threatening to the respondent (Isaac and Michael,1991, p. H.,Jenkins, J. (1969). In most instances of the measurement of social or psychologicalcharacteristics, however, it is unrealistic to hope for a validitycoefficient in this range. DATA COLLECTION: SURVEY, INTERVIEW, AND TABULATION/ANALYSIS This research provides an in-depth review of selected aspects of thedata collection process used in the conduct of research studies. To meet these usage demands, the tabulation/analysis process mustprovide, in most instance, for the reduction of the data collected. Summary This research provided an in-depth review of selected aspects of thedata collection process used in the conduct of research studies. Component analysis involves therotation of variable coordinates to form new frames of reference. Dichotomous questions ask therespondent to select between to possible responses--usually "yes" or "no."Multiple choice questions ask the respondent to select from among a numberof potential responses. Each of the proceduresincluded in the group, however, are intended to reduce a large number ofmeasures to a smaller number which provides a more efficient and powerfulmeasure of the same thing. Bootstrap has been demonstrated as a useful procedure in conjunctionwith relatively simple (mathematically) statistical procedures, such ascorrelation analysis, and with relatively complex (mathematically)statistical procedures, such as factor analysis. 468-514). There are five basic types of surveys--survey of records, mailedquestionnaire, telephone surveys, group interviews, and individualinterviews (Isaac and Michael, 1991, p. Within the variants of the multivariate model of factor analysis, anyscores given weights and, subsequently added together, are defined asfactors of the resulting variables (Kim, 1975, pp. The instrument calibrated in feet and inches would beconsidered to be reliable, if it yielded consistent results in themeasurement time-and-time-again of the same group of individuals. A clear and detailed statement of the problem tobe investigated must be developed. The values in the table under "I" and "II" arecorrelation coefficients, which are referred to as factor loadings (Kim,1975, pp. 5) Can be anonymous. Advantages: 1) Inexpensive. In most instances where inferential statistics is applied inhypothesis testing, population distributions are unknown. 298). New York: McGraw-Hill Book Company, pp.468-514. (199 ). The validity of this measuringinstrument calibrated in feet and inches would be determined on the basisof its ability to accurately measure feet and inches in accordance with anexternal standard, such as a master measurement instrument maintained bythe National Bureau of Standards. 468-514). The bootstrap procedures relies on the use ofrandom numbers and computer support. Repeating the question (Emory, 1988, p. Research methods. If those who stand high on a test,as an example, stand high on the outside criterion, and those who stand lowon the test are low on the outside criterion, the test is valid. Unrotated factor loadings often do not provide as efficient andstrong explanations of variances as are required (Kim, 1975, pp. The use of surveys to collectsuch data is probably the most efficient of the various means available(Dyer, 1985, p. Many research studies generate vast quantities of data. In the maximum likelihood method, R coefficients are used to obtainthe best estimates of factor loadings (Kim, 1975, pp. (1984). Surveys are dependenton the direct communication with persons possessing the characteristics orfactors to be measured. 14-15). Telephone surveys: a. 128). 138). If the instrument does not measurethe height of the same individual as being the same in measurement-after-measurement, it is not a reliable measurement instrument; and, if theinstrument does not measure the height of the same individual as being thesame in measurement-after-measurement, the instrument, obviously, does notalways yield results which are in agreement with the external standard,such as a master instrument maintained by the National Bureau of Standards. Most investigators hold that a coefficient of .7 or higher is necessary for a single test to show validity. An instrument designed to measure somepsychological characteristic or factor would have to be evaluated in thecontext of validity on the basis of its ability to yield results consistentwith those of another instrument whose validity has previously beenestablished. 4) Allows interviewer to record respondent gestures, voicetone, home environment, and so forth. (3rd ed.). The obvious question which arises concerns how high a validitycoefficient should be in order for a measurement procedure to be consideredas valid. A simple tabulation includes making a response count andput them into a frequency distribution table. 3. A relaxed and pleasant atmosphere is desirable for abrainstorming session, and either a recording secretary or recordingmachine should be employed to capture the ideas generated. Ineach instance--full component and common factor, the models may be furthersubdivided on the basis of the correlation or absence thereof of thefactors. 3) Telephone interviewing is often viewed by respondents asbeing intrusive. Further, allprobing actions on the part of the interviewer should be recorded. 468-514). Theaspects of the data collection process reviewed were (1) the survey method,(2) the interview method, (3) questionnaire development, and (4) thetabulation/analysis of data collected through the application of the surveyand interview methods. Semistructured interviews are built around a core of structuredquestions. Surveys are vulnerable to the tendency of some respondents togive consistently high or low ratings (Isaac and Michael, 1991, p. (2d ed.). In some instances, however, psychological tests areconsidered reliable, if their reliability coefficients are .7 or higher. Larger samplesizes and small variances improve the results of maximum likelihood factoranalysis (Kim, 1975, pp. Boston: PWS-Kent Publishing Company. 468-514). Interviews may be unstructured, semistructured, or structured. 468-514). This report shouldinclude a description of the problem investigated, the sampling design, thetesting method, and inferences derived from the results and conclusions ofthe survey (Pfaffenberger and Paterson, 1987, pp. 214-215). The maincharacteristic of the principal factor method is that each factor accountsfor the maximum possible amount of the variance of the variables beingfactored. Thus, if an instrument were designed to provide a measurementof a tendency toward abnormal behavior, the measurements provided by theinstrument would have to be consistent with those of other instruments ofproven reliability in the measurement of tendencies toward abnormal behavior. In thecase of the measurement of tendencies toward abnormal behavior inindividuals, an instrument, to be considered as reliable, would have toprovide consistent scores for the same individuals within a short-span oftime when those individuals were both tested and retested with theinstrument. 5 5). 18). Astronger solution is one in which different factors provide more powerfulexplanations of the Table 1 Unrotated Factor Analysis Matrix (N = 1 )------------------------------------------------------------- Unrotated Factors ---------------------------Variable I II h2-------- ----- ------ ------ A .7 - .4 .65 B .6 - .5 .61 C .6 - .35 .48 D .5 .5 .5 E .6 .5 .61 F .6 .6 .72 ----- ------ ------Eigenvalue 2.18 1.39Percent of variance 36.3 23.2Cumulative percent 36.3 59.5-------------------------------------------------------------variances in each of the variables than is provided by the combined factors(Kim, 1975, pp. 2) Results reflect group behavior and consensus. Free response questions permit respondents to answer in any way theychoose. 468-514). (1987). 3. 128). Therefore, in this example, because the measuring instrument is notreliable, it cannot be valid (Emory, 1988, pp. The interviewing method of collecting data is a survey type, asopposed to being a completely separate procedure. Weinberg, S. For the maximum likelihood method to be used, two characteristicsmust exist. Such an assumption is required in traditional factoranalysis. Semistructured interviews are built around a core of structuredquestions (Isaac and Michael, 1991, p. Inferential statistics permitthe findings with respect to one set of relationships to be extended toother relationships and to generalize findings and conclusions on the basisof statistical inference. 14-15). An expectant pause (Emory, 1988, p. From this core, the interviewer may branch out to explore oneor more points in greater depth. 24). There are a number of procedures by which thesecond step may be accomplished. Thisthird step is a rotation of the variables (Kim, 1975, pp. The survey must bedesigned to isolate the characteristic or characteristics of interest fromother characteristics or factors (Pfaffenberger and Paterson, 1987, pp. This method may be used with either the full component modelor the common factor model (Kim, 1975, pp. Surveys, thus, are reactive in nature. Provo, Utah: Brigham YoungUniversity Press. Therefore, it can easily be seen that reliability in aninstrument is required before that instrument can be considered to bevalid, but that reliability in an instrument cannot, alone, guarantee thatsuch instrument will be valid (Emory, 1988, pp. Thistransformation is performed by determining the best linear combination ofvariables which accounts for the total variance in the data set (Kim, 1975,pp. 3) Permit historical comparisons and trend analysis. The problem statement should clearlyindicate what is to be determined by the conduct of the survey(Pfaffenberger and Paterson, 1987, pp. A cross-tabulation is a contingency tabledisplaying a joint-frequency distribution of the observations applicable totwo or more variables. 299). 32). 468-514). Requirements for the conduct of successfulinterviews include the following: 1. 128). The process of identifying principalcomponents may, in theory, continue until all of the variance in the dataset is explained. Structured interviews follow a well-defined format (Isaac andMichael, 1991, p. 468-514). (3rd ed.). 5 5). Advantages: 1) More efficient and economical than face-to-faceinterviews. To enhance the decision making process. 3) Open to manipulation by the interviewer. Disadvantages: 1) Low response rates are common. This dual requirementinvolves the administration of the survey instrument, the tabulation of thedata on the completed instruments, and the analysis of that data(Pfaffenberger and Paterson, 1987, pp. 468-514). Thorndyke, Robert L., & Hagen, Elizabeth. The general rule is that reliability coefficientsshould be .9 or higher in order for a psychological measurement to beconsidered reliable. In the maximum likelihood approach to factor analysis, the centralfocus is on the estimation of population parameters from simple statistics(Kim, 1975, pp. The conditions underwhich the use of bootstrap may fail for the factor analysis model arestraightforward. 13). Another useful tabulation format is cross-tabulation (Zikmund, 1992,p. The large number of samples, in theory,leads to improved estimates. 2) Many persons, particularly in low-income categories, donot have telephones; another factor that causes validity problems. Tabulation refers to the arrangement of data in a table or othersummary format. 468-514). A validity coefficient is the correlation coefficient between ameasuring procedure and an outside or independent measure of the functionthat the test was designed to measure. 13 -132). (3rd ed.).Chicago: The Dryden Press.----------------------- 3 A survey is "a process of collecting data from existing populationunits, with no particular control over factors that may affect thepopulation characteristics of interest in the study" (Pfaffenberger andPaterson, 1987, p. Factor analytic techniques may be grouped into (1) principalcomponent procedures, and (2) maximum likelihood procedures (Kim, 1975, pp.468-514). 14-15). 468-514). 2) May be wide-ranging. Analysis of variance procedures are used to assess the effectivenessof the intervention (Kim, 1975, pp. 4) Vulnerable to manipulation by an influential or skillfulmember of the group. 468-514). Principal component analysis is the fundamental approach to datareduction (Kim, 1975, pp. These dataare, more often than not, multidimensional, and are characterized bymulticollinearity (Weinberg, Carroll, and Cohen, 1984, 475-491). Validity refers to the extent to which data or a data collectioninstrument measures what it is actually desired to measure. 2. 3. Statistics are used in three general ways. 298).Recording of responses should occur as the responses are made (Emory, 1988,p. 5) The purpose of constructing the database is seldom thesame as the purpose of conducting the survey. 128-134). 14-15). 128). Probability is a numberexpressing the likelihood of occurrence of a specific event. The Interview Method As indicated in the preceding discussion, the interviewing method ofcollecting data is a survey type, as opposed to being a completely separateprocedure (Isaac and Michael, 1991, pp. G. Factor meanings may not becalculated mathematically (Kim, 1975, pp. 128). 6. 468-514). 298). In Nie, N. (1991). In the simple example used above--that of an instrument designedto measure the height of individuals in feet and inches, it would bedesirable for the instrument to have a validity coefficient of 1. The axes are rotated until a stronger solution is found. Because surveys cause respondents to feel special or unnatural,they may elicit artificial or slanted responses (Isaac and Michael, 1991,p. 128-134). These advantages anddisadvantages are as follows(Isaac and Michael, 1991, p. These situations are (1) whererelationships are nonlinear, (2) where the relationship between thevariable and the factor is not stable through all levels of the factor, and(3) where the relationships of several factors to a single variable arevirtually interchangeable. 5-2 ). 128). The problem with the full component model is the mass of dataand ties required for its execution. 4) Vulnerable to interviewer biases. 217).Disguising study objectives is a common practice used to minimize theintroduction of distortions in responses. Factoranalysis is a widely used statistical procedure with data-reductioncapabilities that is used to determine the underlying pattern ofrelationships among a set of variables or conditions which may be taken assource variables, accounting for the observed interrelations in the data.Factor loadings are the coefficients of factors identified in factoranalysis, and are used as measures of the degree of the relationshipsbetween factors and variables. As long as data collection is a relatively randomprocess, and the sample size is moderate, bootstrap may be expected toperform satisfactorily. Emory, C. Sample sizes must be moderate. A survey must be objective. 13 -132). A survey must be quantifiable. (1985). The accuracy of the maximum likelihoodmethod increases, as the variance across samples decreases (Kim, 1975, pp.468-514). 128). 468-514). There are advantages anddisadvantages associated with each type of survey. To meet these usage demands, the tabulation/analysis process mustprovide, in most instance, for the reduction of the data collected. The findings of the survey must be reported. The bootstrap procedure is a powerful data-analytictool, which permits the construction of distributions of sample statisticsand their variability. H., Hull, C. 2) Fosters conformity. The test of significance yields achi square statistic, wherein the null hypothesis is that all of thepopulation multicollinearity has been extracted by the hypothesized numberof factors (Kim, 1975, pp. 129).In the brainstorming process, freewheeling is welcomed, and, initially, atleast, quantity is sought in preference to quality, although judicialthinking and evaluation is not prohibited. Three steps are involved in the construction of a questionnaire--designing the instrument, conducting a pretest, and editing the results(Pfaffenberger and Paterson, 1987, p. 2. There are three instances wherein multivariate models of factoranalysis may not provide the best representation of the factor and variablerelationships (Kim, 1975, pp. Again, as was true forvalidity coefficients, reliability coefficients must be positive, asnegative reliability coefficient is not an indication of reliability,regardless of how high the negative coefficient may be. Homewood, Illinois: RichardD. The data must be collected and analyzed. L., Carroll, J. 3) Intensifies group loyalties, and can rigidly polarizepositions. There are several variants of the multivariate model of factoranalysis; however, these variants may be grouped into two principal classes--the full component model, and the common factor model (Kim, 1975, pp. Four guiding principles underlie effective surveys (Isaac andMichael, 1991, p. To accomplish the objective of validity in measurement, it isnecessary to have some standard which is external to the measurementprocedure, in order to evaluate the validity of the procedure. 468-514): 1. 7 .85 ----- -------------------------------------------------------------------interpretation of factor loadings commences, factor analysis ceases to be asophisticated mathematical procedure, and becomes, instead, aninterpretative art (Kim, 1975, pp. Within a given specified number of factors, thefactors could be rotated into an infinite number of positions, all of whichwould satisfy the maximum likelihood function. Using the data presented in Table 1, the rotatedfactors would be those presented in Table 2, which may be found on thefollowing page.

If this paper is not what you are looking for, you can search again:

Search for:

or

We can write a Custom Essay just for you.


Browse Essays by Subject