ZIMBABWE OPEN UNIVERSITY
IN COLLABORTATION WITH
THE UNIVERSITY OF ZAMBIA
MASTER OF EDUCATION IN ADMINISTRATION, PLANNING
AND POLICY STUDIES
NAME: PHIRI, CHICCO
COURSE: MDEA 560
COMPUTER: 717819407
Question:
discuss the survey method and show its relevance to educational
research.
Page 1 of 16
The quality of the educational system and infrastructure is central to every nation’s economic,
social, political and technological well-being. The quality of education depends and builds on the
quality, rigor and relevance of available educational research. It is therefore of critical
importance to secure and raise the standards for conducting research in order to improve
education. “Educational research is a systematic approach of gaining a better understanding of
educational processes, generally with a view of improving educational efficiency whilst applying
scientific methods to the study of educational problems,” (Munroe, 2000: 14 )
In Zambia, the Research Council holds a critical position when it comes to organizing and
funding educational research. The Research Council has been funding educational research
programmes since the mid-1990s, starting with the research programme Competence, Learning
Processes and Value Creation in Work life, and the evaluation of the 1976 educational reforms.
However, it has been reported that all research initiatives within the educational sciences have
lacked a long-term perspective, sufficient volume for funding and infrastructures that paid
attention to processes of synthesizing and accumulating research within the education sector. It
was therefore a huge step forward when the Ministry of Education together with the Research
Council of Zambia launched the new research programme Educational Research towards 2020;
dubbed UTDANNING2020. The UTDANNING2020 research programme is designed to address
and challenge scientific merits, multidisciplinarily, rigor and relevance in educational research.
In order to promote scientific quality and merits of educational research, UTDANNING 2020
uses different tools and strategic actions. For this reason, funding of high quality research
relevant to the educational sciences holds a key position in this tool kit. Through a rich portfolio
of varied and intriguing research projects the programme aims to contribute to new insight,
accumulate knowledge, support methodological awareness and growth and contribute to
fostering research capacity within the educational sciences.
Annual seminars and conferences as mechanisms for knowledge exchange and knowledge
building are examples of other activities meant to foster quality in educational research. Within
the programme, these seminars and conferences are targeting different groups and audiences like
policymakers and stakeholders, the teaching profession, researchers and other knowledge
brokers.
Page 2 of 16
Therefore, against the above background the aim of this paper is to discuss the survey methods
and show their relevance to educational research.
Educational research has long recognized that every method of scientific inquiry is subject to
limitations and that choosing among research methods inherently involves trade-offs. With the
control of a laboratory experiment, for example, comes an artificiality that raises questions about
the generalizability of results. And yet the 'naturalness" of a field study or an observational study
can jeopardize the validity of causal inferences. The inevitability of such limitations has led
many methodologists to advocate the use of multiple methods and to insist that substantive
conclusions can be most confidently derived by triangulating across measures and methods that
have no overlapping strengths and weaknesses (Ibid).
Survey research involves the collection of data about a research issue from the target population
to answer questions that have been raised. Survey researchers have a variety of techniques at
their disposal for evaluating survey questions (Presser et al. 2004). These range from cognitive
interviews (Willis: 2005), to the conventional pretests recommended in many questionnaire
design texts (Converse and Presser: 1986), to behavior coding of various types (Maynard et al:
2002), to question wording experiments (Fowler: 2004), to the application of statistical
procedures, such as latent class analysis (Biemer: 2004) or structural equation modeling that
provide quantitative estimates of the level of error in specific items. These different evaluation
techniques do not bear a close family resemblance. Although they all share the general goal of
helping question writers to evaluate survey questions, they differ in their underlying
assumptions,the data collection methods they use, the types of problems they identify, the
practical requirements for carrying them out, the type of results they generate, and so on.
To illustrate the differences across techniques, consider cognitive interviewing and the use of
latent class modeling methods for evaluating survey items. Cognitive interviewing is a practice
derived from the protocol analyses used in the work of Simon and his collaborators. Loftus
(1984) first pointed out the potential relevance of Simon’s work to the testing of survey items
more than 25 years ago. The key assumptions of protocol analysis and its latter-day survey
descendant, cognitive interviewing, are that the cognitive processes involved in answering
survey questions leave traces in working memory (often intermediate products of the process of
formulating an answer) and that respondents can verbalize these traces with minimal distortion
Page 3 of 16
(Ericsson and Simon 1980). Cognitive interviews added several techniques to the think-aloud
methods introduced by Simon and his colleagues, especially the use of specially designed
probes, or follow-up questions; responses to these probes are also thought to provide important
clues about how respondents come up with their answers and about potential problems with
those processes. Some researchers see later developments of cognitive interviews as a departure
from the original paradigm proposed by Ericsson and Simon, and argue that the use of probes
has largely supplanted the use of think-aloud methods in cognitive interviewing (for example,
Schaeffer and Presser 2003, p. 82; see also Beatty and Willis 2007; Gerber 1999). Cognitive
interviews are rarely subjected to formal analyses; instead, the questionnaire testing personnel,
often staff with advanced degrees, draw conclusions about the questions from their impressions
of the verbal reports produced by respondents during the cognitive interviews (Willis 2005 for a
thorough discussion of cognitive interviewing).
On the other end of the continuum stands the application of quantitative methods, such as latent
class modeling, to assess problems in survey questions. In a series of papers, Biemer and his
colleagues (Biemer 2004) have used latent class models to estimate error rates in survey items
designed to assess such categorical constructs as whether a person is employed or not. Latent
class analysis is sometimes described as the categorical analogue to factor analysis
(McCutcheon: 1987). It is used to model the relationships among a set of observed categorical
variables that are indicators of two or more latent categories (for example: whether one is truly
employed or unemployed). In contrast to cognitive interviews, latent class analysis is a statistical
technique that yields quantitative estimates. It uses maximum likelihood methods to estimate
parameters that represent the prevalence of the latent classes and the probabilities of the different
observed responses to the items conditional on membership in one of the latent classes.
Given the large number of different evaluation methods and the large differences between them,
it is an important theoretical question whether the different methods should yield converging
conclusions and, if not, whether they should be used alone or in combination with each other. In
practice, the choice between techniques is often dictated by considerations of cost and schedule,
and it is an important practical question whether clear conclusions will result even if different
methods are adopted
Page 4 of 16
The answers to the questions of whether converging conclusions should be expected and how to
cope with diverging conclusions about specific items depend in part on how researchers conceive
of the purpose of the different evaluation methods. Much work on question evaluation and
pretesting tends to treat question problems as a binary characteristic – the question either has a
problem or it does not. Question evaluation methods are used to identify the problems with an
item and group questions into two categories – those with problems that require the item to be
revised, and those without such problems. Under this conceptualization, all of the question
evaluation methods flag some items as problematic and others as non-problematic, even though
the methods may differ in which items they place in each category. Of course, questions may
have problems that differ in seriousness, but ultimately questions are grouped into those that
require revision and those that do not. Different question evaluation methods are, then, compared
on their success in identifying question problems and correctly placing items into one of the two
categories. Presser and Blair’s (1994) study is a classic example of such a conceptualization.
Implicit in such work is the assumption that if any method reveals a problem with an item, that
problem should be addressed. That assumption has been challenged recently by Conrad and Blair
(2004; see also Conrad and Blair 2009), who argue that the “problems” found in cognitive
interviews may well be false alarms.
Recently, the field of question evaluation and pretesting has seen a shift towards a more general
conceptualization of question problems and goals of question evaluation methods (Miller 2009).
Survey questions measure the construct they are supposed to measure more or less well (Saris
and Gallhofer 2007). Thus, it is possible to conceive of question problems as a matter of degree,
and the purpose of question evaluation methods is to determine the degree of fit between
questions and construct (Miller 2009).
There is limited empirical work comparing different question evaluation methods, especially
work comparing qualitative methods (like cognitive interviews and expert reviews) with
quantitative methods (like measurements of reliability and validity). The few prior studies that
have been done seem to suggest that the consistency between the different methods is not very
high, even at the level of classifying items as having problems or not (Presser and Blair 1994;
Rothgeb et al. 2001; and Willis et al. 1999, for examples).
Page 5 of 16
Most researchers use the term 'sample" to refer both to (a) the set of elements that are sampled
from the sampling frame from which data ideally will be gathered and (b) the final set of
elements on which data actually are gathered. Because almost no survey has a perfect response
rate, a discrepancy almost always exists between the number of elements that are sampled and
the number of elements from which data are gathered. Lavrakas (1993) suggests that the term
'sampling pool' be used to refer to the elements that are drawn from the sampling frame for use in
sampling and that the term 'sample" be served for that subset of the sampling pool from which
data are gathered.
Moreover, survey researchers can make use of non-probability sampling methods. Such methods
have been used frequently in studies inspired by the recent surge of interest among social
psychologists in the impact of culture on social and psychological processes (Kitayama, 1996).
In a spate of articles published in top journals, a sample of people from one country has been
compared with a sample of people from another country, and differences between the samples
have been attributed to the impact of the countries' cultures. In order to convincingly make such
comparisons and properly attribute differences to culture, of course, the sample drawn from each
culture must be representative of it. And for this to be so, one of the probability sampling
procedures described above must be used.
Alternatively, one might assume that cultural impact is so universal within a country that any
arbitrary sample of people will reflect it. However, hundreds of studies of Americans have
documented numerous variations between subgroups within the culture in social psychological
processes, and even recent work on the impact of culture has documented variation within
nations (Cohen, 1996). Therefore, it is difficult to have much confidence in the presumption that
any given social psychological process is universal within any given culture, so probability
sampling seems essential to permit a conclusion about differences between cultures based on
differences between samples of them.
On the other hand, nearly all recent social psychological studies of culture have employed
nonprobability sampling procedures. These are procedures where some elements in the
population have a zero probability of being selected or have an unknown probability of being
selected. For example, Heine and Lehman (1995) compared college students enrolled in
psychology courses in a public and private university in Japan with college students enrolled in a
Page 6 of 16
psychology course at a public university in Canada. Rhee et al. (1995) compared students
enrolled in introductory psychology courses at New York University with psychology majors at
Yonsei University in Seoul: Korea. Han and Shavitt (1994) compared undergraduates at the
University of Illinois. With students enrolled in introductory communication or advertising
classes at a major university in Seoul. And Benet and Waller (1995) compared students enrolled
at two universities in Spain with Americans listed in the California 'Ibvin Registry.
In all of these studies, the researchers generalized the findings from the samples of each culture
to the entire cultures they were presumed to represent. For example, after assessing the extent to
which their two samples manifested self-enhancing biases, Heine and Lehman (1995) concluded
that 'people from cultures representative of an interdependent construal of the self," exhibited by
the Japanese students, "do not self-enhance to the same extent as people from cultures
characteristic of an independent self," exhibited by the Canadian students. Yet, the method of
recruiting potential respondents for these studies rendered zero selection probabilities for large
segments of the relevant populations. Consequently, it is impossible to know whether the
obtained samples were representative of those populations, and it is also impossible to estimate
sampling error or to construct confidence intervals for parameter estimates. As a result, the
statistical calculations used in these articles to compare the different samples were invalid,
because they presumed simple random sampling.
More important, their results are open to alternative interpretations, as is illustrated by Benet and
Waller's (1995) study. One of the authors' conclusions is that in contrast to Americans,
"Spaniards endorse a 'radical' form of individualism". Justifying this conclusion, ratings of the
terms "unconventional," "peculiar," and "odd" loaded in a factor analysis on the same factor as
ratings of "admirable" and "high-ranking" in the Spanish sample, but not in the American
sample. However, Benet and Waller's American college student sample was significantly
younger and more homogeneous in terms of age than their sample of California twins (the
average ages were 24 years and 37 years, respectively; the standard deviations of ages were 4
years and 16 years, respectively). Among Americans, young adults most likely value
unconventionality more than older adults, so what may appear in this study to be a difference
between countries attributable to culture may instead simply be an effect of age that would be
apparent within both cultures.
Page 7 of 16
Even researchers who recognize the value of survey methodology in educational research inquiry
are sometimes reluctant to initiate survey research because of misconceptions regarding the
feasibility of conducting a survey on a limited budget. And indeed, the cost of prominent largescale national surveys conducted by major survey organizations are well outside of the research
budgets of most schools. But survey methodologists have recently begun to rekindle and expand
the early work of Hansen and his colleagues (Hansen & Madow, 1953) in thinking about survey
design issues within an explicit cost-benefit framework geared towards helping researchers make
design decisions that maximize data quality within the constraints of a limited budget. This
approach to survey methodology, known as the 'total survey error" perspective can provide
educational research with a broad framework and specific guidelines for making decisions to
conduct good surveys on limited budgets while maximizing data quality, (Dillman: 1978,
Fowler: 198, Groves: 1989 & Lavrakas: 1993).
The total survey error perspective recognizes that the ultimate goal of survey research is to
accurately measure particular constructs within a sample of people who represent the population
of interest. In any given survey, the overall deviation from this ideal is the cumulative result of
several sources of survey error. Superficially, the total survey error perspective disaggregates
overall error into four components: coverage error, sampling error, non-response error, and
measurement error. Coverage error refers to the bias that can result when the pool of potential
survey participants from which a sample is selected does not include some portions of the
population of interest. Sampling error refers to the random differences that invariably exist
between any sample and the population from which it was selected. Non-response error is the
bias that can result when data are not collected from all of the members of a sample. while
measurement error refers to all distortions in the assessment of the construct of interest,
including systematic biases and random variance that can be brought about by respondents' own
behavior (such as misreporting true attitudes, or failing to pay close attention to a question),
interviewer behavior (manifested in misreporting responses, or providing clues that lead
participants to respond in one way or another), and the questionnaire (when the instrument has
ambiguous or confusing questions, or biased question wording or response options). The total
survey error perspective explicitly advocates taking into consideration each of these sources of
error and making decisions about the allocation of finite resources with the goal of reducing the
sum of the four. In the sections that follow, we consider each of these potential sources of survey
Page 8 of 16
error and their implications for psychologists seeking to balance pragmatic budget considerations
against concerns about data quality.
Surveys offer the opportunity to execute studies with various designs, each of which is suitable
for addressing particular research questions of long-standing interest to social psychologists. In
this section, we will review several standard designs, including cross-sectional, repeated crosssectional, panel, and mixed designs, and discuss when each is appropriate for social
psychological investigation. We will also review the incorporation of experiments within
surveys.
Cross-sectional surveys involve the collection of data at a single point in time from a sample
drawn from a specified population. This design is most often used to document the prevalence of
particular characteristics in a population. For example, cross-sectional surveys are routinely
conducted to assess the frequency with which people perform certain behaviors or the number of
people who hold particular attitudes or beliefs. However, documenting prevalence is typically of
little interest to social psychologists, who are usually more interested in documenting isolations
between variables and the causal processes that give rise to those associations. Cross-sectional
surveys do offer the opportunity to assess relations between variables and differences between
subgroups in a population. Although many scholars believe their value ends there, this is not the
case. Cross-sectional data can be used to test causal hypotheses in a number of ways. For
example, using statistical techniques such as two-stage least squares regression, it is possible to
estimate the causal impact of variable A on variable B, as well as the effect of variable B on
variable A (Blalock: 1972). Such an analysis rests on important assumptions about causal
relations among variables, but these assumptions can be tested and revised as necessary (James
& Singh: 1978).
Furthermore, path analytical techniques can be applied to test hypotheses about the mediators of
causal relations, thereby validating or challenging notions of the psychological mechanisms
involved. Moreover, cross-sectional data can be used to identify the moderators of relationships
between variables, thereby also shedding some light on the causal processes at work, (Kenny:
1979).
In a panel survey, data is collected from the same people at two or more points in time. Perhaps
the most obvious use of panel data is to assess the stability of psychological constructs and to
Page 9 of 16
identify the determinants of stability. At the same time, with such data, one can test causal
hypotheses in at least two ways. Firstly, one can examine whether individual-level changes over
time in an independent variable correspond to individual-level changes in a dependent variable
over the same period of time. So, for example, one can ask whether people who experienced
increasing interracial contact manifested decreasing racial prejudice, while at the same time,
people who experienced decreasing interracial contact manifested increasing racial prejudice.
Secondly, one can assess whether changes over time in a dependent variable can be predicted by
prior levels of an independent variable. Hence for example, do people who have had the highest
amounts of interracial contact at Time 1 manifest the largest decreases in racial prejudice
between Time 1 and Time 2? Such a demonstration provides relatively strong evidence
consistent with a causal hypothesis, because the changes in the dependent variable could not
have caused the prior levels of change in the independent variable (Blalock, 1995).
One application of this approach occurred in a study of a long-standing social psychological idea
called the projedm hypothesis. Rooted in cognitive consistency theories, it proposes that people
may overestimate the extent to which they agree with others whom they like, and they may
overestimate the extent to which they disagree with others whom they dislike. By the late 19805,
a number of cross-sectional studies by political psychologists yielded correlations consistent with
the notion that people's perceptions of the policy stands of presidential candidates were distorted
to be consistent with attitudes toward the candidates (Granberg: 1985, Kinder: 1978). However,
there were alternative theoretical interpretations of these correlations, so an analysis using panel
survey data seemed in order. Krosnick (1999) did such an analysis exploring whether attitudes
toward candidates measured at one time point could predict subsequent shifts in perceptions of
presidential candidates' issue stands. And he found no projection at all to have occurred, thereby
suggesting that the previously documented correlations were more likely due to other processes
(such as deciding how much to like a candidate based on agreement with him or her on policy
issues).
Furthermore, with each additional wave of panel data collection, it becomes increasingly difficult
to locate respondents to re-interview them, because some people move to different locations,
some die, and so on. This may threaten the representativeness of panel survey samples if the
members of the first-wave sample who agreed to participate in several waves of data collection
Page 10 of 16
differ in meaningful ways from the people who are interviewed initially but do not agree to
participate in subsequent waves of interviewing.
Also, participation in the initial survey may sensitize respondents to the issues under
investigation, thus changing the phenomena being studied. As a result, respondents may give
special attention or thought to these issues, which they have an impact on subsequent survey
responses. For example, Bridge et al. (1977) demonstrated that individuals who participated in a
survey interview about health subsequently considered the topic to be more important. And this
increased importance of the topic can be translated into changed behavior. For example, people
interviewed about politics are subsequently more likely to vote in elections. Even answering a
single survey question about one's intention to vote, increases the likelihood that an individual
will turn out to vote on Election Day (Greenwald & Young: 1987).
Additional evidence of causal processes can be documented in surveys by building in
experiments. If respondents are randomly assigned to 'treatment' and 'control" groups,
differences between the two groups can then be attributed to the treatment. Every one of the
survey designs described above can be modified to incorporate experimental manipulations.
Some survey respondents (selected at random) can be exposed to one version of a questionnaire,
whereas other respondents are exposed to another version. Differences in responses can then be
attributed to the specific elements that were varied.
Many social researchers are aware of examples of survey research that have incorporated
experiments to explore effects of question order and question wording. Less salient are the
abundant examples of experiments within surveys that have been conducted to explore other
social psychological phenomena, (Presser: 1981).
What is the benefit of doing these experiments in representative sample surveys? Couldn't they
instead have been done just as well in laboratory settings with college undergraduates? Certainly,
the answer to this latter question is yes; they could have been done as traditional social
psychological experiments. But the value of doing the studies within representative sample
surveys is at least three-fold. Firstly, survey evidence documents that the phenomena are
widespread enough to be observable in the general population. This boosts the apparent value of
the findings in the eyes of the many no psychologists who instinctively question the general
liability of laboratory findings regarding undergraduates.
Page 11 of 16
Secondly, estimates of effect seen from surveys provide a more accurate basis for assessing the
significance that any social psychological process is likely to have in the course of daily life.
Effects that seem large in the lab (perhaps because undergraduates are easily influenced) may
actually be quite small and thereby much less socially consequential in the general population.
Lastly, general population samples allow researchers to explore whether attributes of people that
are homogeneous in the lab but vary dramatically in the general population (for instance: age and
educational attainment) moderate the magnitudes of effects or the processes producing them
(Sanders: 1990).
Carefully monitoring ongoing data collection permits early detection of problems and seems
likely to improve data quality. In self-administered surveys, researchers should monitor
incoming data for signs that respondents are having trouble with the questionnaire. In face-toface or telephone surveys, researchers should maintain running estimates of each interviewer's
average response rate, level of productivity, and cost per completed interview. The quality of
each interviewer's completed questionnaires should be monitored, and if possible, some of the
interviewees themselves should be supervised. When surveys are conducted by telephone,
monitoring the interviews is relatively easy and inexpensive and should be done routinely. When
interviews are conducted face-to-face, interviewers can tape record some of their interviews to
permit evaluation of each aspect of the interview.
When data collection occurs from a single location (like telephone interviews that are conducted
from a central phone bank), researchers can be relatively certain that the data are authentic.
When data collection does not occur from a central location (in a case of face-to-face interviews
or telephone interviews conducted from interviewers' homes), researchers might be less certain.
It may be tempting for interviewers to falsify some of the questionnaires that they turn in, and
some occasionally do. To guard against this, researchers often establish a procedure for
confirming that a randomly selected subset of all interviews did indeed occur (recontacting some
respondents and asking them about whether the interview took place and how long it lasted).
Perhaps a first step for many scholars in this direction might be to make use of the many
archived survey data sets that are suitable for secondary analysis. So even when the cost of
conducting an original survey is prohibitive, survey data sets have a lot to offer educational
researchers. It is hoped that more educational researchers will take advantage of the opportunity
Page 12 of 16
to collect and analyze survey data in order to strengthen our collective enterprise. Doing so may
require somewhat higher levels of funding than has been evidenced from the pas. It is anticipated
that theoretical richness, scientific edibility, and impact across disciplines are likely to grow as a
result, in ways that are well worth the price.
In conclusion, over the last several decades, educational researchers have come to more fully
appreciate the complexity of social thought and behavior. In domain after domain, simple 'main
effect" theories have been replaced by more sophisticated theories involving moderated effects.
Statements such as 'People
Learners behave like this" are being replaced by 'Certain types of learners behave like this under
these circumstances." More and more, educational researchers are recognizing that learning
processes apparent in one class may operate differently among learners in other classes and that
personality factors, social identity, and cultural norms can have a profound impact on the nature
of many educational problems being researched on.
Despite this, the bulk of research in the education continues to be conducted with a very narrow,
homogeneous base of participants. As a result, some scholars have come to question the
generalization of educational research findings, and some disciplines look at the findings with
skepticism.
In this paper, survey method has been presented as an overview research methodology that
allows educational research to address pertinent problems in education. Surveys enable scholars
to explore educational research phenomena with samples that accurately represent the population
about whom generalizations are to be made. And the incorporation of experimental
manipulations into survey designs offers special scientific power. It is the writer’s anticipation
that the discussed advantages of survey research make a valuable addition to the methodological
arsenal available to educational research. The incorporation of this methodology into a full
program of research enables researchers to triangulate across measures and methods, providing
more compelling evidence of educational research phenomena than any single methodological
approach can offer.
Page 13 of 16
REFERENCES
Adomo, T. W., Frenkel-Brunmrik, E., Levinson, D. J., 6. Sanford, R. N. (1950). The
authoritarian personally. New York: Harper 6. Row.
Ahlgren, A. (1983). Sex differences in the correlates of cooperative and competitive school
attitudes. Devr&pmrntalPsychology, 19,881-888.
Alwin, D. F., Cohen, R. L., 6. Newcomb, T. M. (1991). The women of Bmningtmr: A of political
orientations over the life span. Madison: University of Wiscansin Press.
Alwin, D. F., 6. Jackson, D. J. (1982). Adult values for children: An application of factor
analysis to ranked preference data. In K. F. Schuessler (Ed.), Sociological methodology 1980.
San PransW Jossey-Bass.
Babbie, E. R. (1990). Survcy research methods. Belmont, CA: Wadsworth.
Bardn, R. M., 6. Kenny, D. A. (1986). The moderator mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations.
Journal of Personalily and Soda1 Psychology, 51.1 17 3- 1182.
Bass, R. T., 6 Tortora, R. D. (1988). A comparison of centralized CAT1 facilities for an
agricultural labor survey. In R. M. Groves, P. P. Beimer, L. E. Lyberg, J. T. Massey, W. L.
Nicholls, 6. J. Waksberg (Eds.), Trlephonesumymethodow(pp; 497-508). New York: Wiley.
Bearden, W. Q., Netemeyer, R. G., 6. Mobley, M. F. (1993). Handbook of marketing scales.
Newbury Park, CA: sage.
Ban, D. J., 6 McConneU, H. K. (1970.). Testing the self-perception explanation of dissonance
phenomena: On the salience of pre-manipulation attitudes. Journal of Pmmral- 3
ilyandSodalPrydtology, 1123-31.
Page 14 of 16
Benet, V., 6. Waller, N. G. (1995). The big seven factor model of personality description:
Evidence for its cross-cultural generality in a Spanish sample. Journal of Personality and soda1
Prydtology, 69,701-718.
Bischoping, K. (1989). An evaluation of interviewerdebriefingin survey pretests. In C. F.
Cannell, (pp. 15-29). Ann Arbor, MI:Survey Research Center.
Bishop, G. F., 6 Fisher, B. S. (1995). 'Seaet ballots" and self -reports in an exit-poll experiment.
Public Opinion Quartn&, 59, 568-588.
Bradburn, N. M., Sudman, S., 6. Associates. (1 981). Improving interringmethodand
qwstt'onnairc&sign. San Francisco: Jossey-Bass.
Brehm, J. (1993). The phantomrrspondntts. Ann Arbor: University of Michigan Press.
Campbell, D. T., 6 Stanley, J. C. (1963). Experimental and quasi-experimental designs for
research. In N. L. Gage (Ed.), Handbook of March on teaching (pp. 171-246). Chicago: Rand
McNally
Cohen, D., Nisbett, R. E., Bowdle, B. P., 6 Schwm, N. (1996). Insult, aggression, and the
southern culture of honor: An 'experimental ethnography.." Journal of P.-ality and Soda1
Psychology, 70,945-960.
Granberg, D., 6 Holmberg, S. (1992). The Hawthorne effect in election studies: The impact of
survey participationon voting. British Journal of PoliticulScim, 22, 240- 247.
Greenwald, A. G., Carnot, C. G., Beach, R., 6 Young, B. ( 1987). Increasing voting behavior by
asking people if they expect to vote. Journal of Applied Psychology, 72, 31 5- 318.
Hansen, M. H., 6 Madow, W. G. (1953). Survey methodsand theme. New York: Wiley.
Miethe, T. D. (1985). The validity and reliability of value measurements. Journal of Pmonaliv,
119,441453.
Presser, S., 6 Blair, J. (1994). Survey pretesting: Do different methods produce different results?
In P. V. Marsden (Ed.), Socblogicul methodology, 1994 (pp. 73-104). Cambridge, MA:
Blackwell.
Page 15 of 16
Schumau, H., 6 Presser, S. (1981). Qucsrions and ansum in attit& sum. San Diego, CA:
Academic Press.
Weisberg, H. F., Krosnick, J. A., & Bowen, B. D. (1996). An introduction to survey march,
polling, and hta analysis (3rd ed.). Newbury Park, CA: Sage.
Page 16 of 16