Survey and Correlational Research
MUHD SYAHIR ABDUL RANI2011603908
ZAIRRIENOOR ZAIRRIN ZAINUDIN2011342477
ROSLEEN KHAIRI2011999609
Definition SURVEY RESEARCH involves
collecting data to test hypotheses /to Answer Q about people’s opinions on some topics or issue.
A SURVEY = instrument to collect data that describe one or more characteristics of a specific population.
Purpose Gather information about groups beliefs,
attitudes, behaviour, and demographic composition.
SAMPLE SURVEY: researcher attempts to infer information about population based on a representative sample drawn from that population.
CENSUS SURVEY: researcher attempts to acquire information from every member of a population.
Cross-Sectional Surveys Data are collected from selected
individuals at a single point in time. Single, stand alone study. Effective for providing a snapshot of a
current behaviours, attitudes, and beliefs in a populations.
Provide data quickly.
Longitudinal Surveys
Data are collected at 2 or more times. Require an extended commitment by the
researcher and participants.
Longitudinal
Trend survey Cohort Survey Panel Survey Follow-up Survey
Trend Survey Examines changes over time in a
particular population defined by some particular trait/traits.
Researcher can analyze changes in attitudes, beliefs, behaviours within that particular population over time.
Cohort Survey Involves one population selected at a
particular time period but multiple samples taken and surveyed at different points of time.
Can be different samples, but in the same population.
Panel Survey The same individuals are studies over
time. Frequent problem: lost of individuals
from the study because of relocation, name change, lack of interest, or death.
Take long time.
Follow-Up Survey Addresses development or change in a
previously studied population, some time after the original survey was given.
To examine changes in the attitudes, behaviours or beliefs.
Conducting Survey Research Aim: collection of standardized, quantifiable
information from all members of a population or a sample.
The researcher must ask them each the same question.
A questionnaire is written collection of survey questions to be answered by a selected group of research participants.
An interview is an oral in-person question-and answer.
Conducting a Questionnaire Study-Stating the problem The problem or topic studied and the
contents must be sufficient significance to motivate potential respondents to respond, and to justify the research effort.
Researcher should set specific objectives indicating the kind of information needed.
Continued…Constructing The Questionnaire Questionnaire should be attractive, brief,
and easy to respond to Identify the sub-areas of research topics to
make the process of developing the questionnaire easier.
Types of items: scaled, ranked, checklist, free response. Pg187
Include only items that relate to RO Collect demographic information about the
sample if you want to compare to different subgroups.
Focus each question on a single concept
Continued…-Things to consider Define/explain ambiguous terms. Include a point of reference to guide
respondents in answering questions. Avoid leading questions. Avoid sensitive questions. Don’t ask a questions that assumes a
fact that not necessarily true.-Pilot Test the Questionnaire Provide information about deficiencies
and suggestion for improvement.
Continued… Choose 2 or 3 individuals who are
thoughtful, critical, and similar to the intended research participants.
-Preparing Cover Letter When necessary.
Conducting The Questionnaire
Select participants: simple/stratified random, cluster, systematic, nonrandom.
Distributing: mail, email, telephone.], personal admin, interview. pg191
Conduct follow-up for reminder. Rule of thumbs: must be more than
50%. Dealing with nonresponse: get new
participants OR make assumptions by generalization.
Analyzing Results: Select total sample size.
Definition Involves collecting data to determine
whether, and to what degree, a relationship exists between two or more quantifiable variables.
The degree of relation is expressed as a correlation coefficient. i.e. if two variables are related, scores within certain range on one variable are associated with the other variable.
Purpose To determine relations among variables
(i.e. relationship study) or to use these relations to make predictions (i.e. prediction study)
To determine various types of validity and reliability.
Problem Selection Variables to be correlated should be
selected on the basis of some rationale. It should be a logical one.
“Treasure hunts”- the researcher correlates all sorts of variables to see what turns up are strongly discourage (cause inefficiency and findings difficult to interpret).
Participant and Instrumental Selection Minimum sample is 30. The higher the validity and reliability of
the variables to be correlated, the smaller the sample can be, but not fewer than 30.
Design and Procedure Scores for two (or more) variables of
interest are obtained for each member of the sample, and the paired scores are then correlated.
The result is expressed as a correlation coefficient that indicates the degree of relation between the two variables.
Data Analysis and Interpretation
When two variables are correlated, the result is a correlation coefficient, which is a decimal number ranging from -.00 to +1.00
i.e. a person with a high score on one of the variables is likely to have a high score on the other variable, and a person with a low score on one variable is likely to have low score on the other.
Relationship Studies Is when: a researcher attempts to gain
insight into variables or factors that are related to a complex variables (e.g. academic achievement, motivation, and self concept)
Purpose 1. They help to identify related variables
suitable for subsequent examination in causal-comparative and experimental studies.
2. Relationship study provide information about the variables to control for in causal-comparative and experimental studies.
PREDICTION STUDIESo If two variables are highly related, scores on one can
be used to predict scores on the other.
o The variable used to predict is called predictor. E.g :
high school grades or certification exam.
o The variable is predicted is a complex variable called
the criterion. E.g : college grades or principals’
evaluations.
o Prediction study is an attempt to determine which of a
number of variables are most highly related to the
criterion variable.
o Are conducted to facilitate decision making about
individuals.
o To aid in various types of selection.
o To determine the predictive validity of measuring
instruments
o The results of prediction studies are used not only by
researchers but also by counselors, admissions directors
and employers.
o More than one variable can be used to make predictions.
o A combination of variables will be more accurate.
Data collectiono In all correlational studies, research participants
must be able to provide the desired data and
must be available to the researcher.
o Valid measuring instruments should be selected
to represent the variables.
o It is especially important that the measure used
for the criterion variable be valid.
PREDICTION STUDY RELATIONSHIP STUDYPredictor variables are generally obtained earlier than the criterion variable.
An interesting characteristic is shrinkage, the tendency for the prediction to be less accurate for a group other than the one on which it was originally developed.
All variables are collected within a relatively short period of time.
Data Analysis and Interpretationo Data analysis in prediction studies involves correlating
each predictor variable with the criterion variable.o For single variable predictions, the form of the prediction
equation is:Y = a + bX
whereY = the predicted criterion score for an individualX = an individual’s score on the predictor variable
a = a constant calculated from the scores of all participantsb = a coefficient that indicates the contribution of the
predictor variable to the criterion variable
Continued…o Because a combination of variables usually
results in a more accurate prediction than any one variable, a prediction study often results in a multiple regression equation.
o A multiple regression equation, also called a multiple prediction equation, is a prediction equation including two or more variables that individually predict a criterion, resulting in a more accurate prediction.
o An intervening variable, a variable that cannot be directly observed or controlled, can influence the link between predictor and criterion variables.
OTHER CORRELATION-BASED ANALYSES
More complex correlation-based analyses include:o Discriminant function analysis, which is quite similar to
multiple regression analysis with one major difference; continuous predictor variables are used to predict a categorical variable.
o Canonical analysis is an extension of multiple regression analysis. It produces a correlation based on a group of predictor variables and a group of criterion variables.
o Path analysis also allows us to see the relations and patterns among a number of variables. The outcome is a diagram that shows how variables are related to one another.
o An extension of path analysis that is more sophisticated and powerful is called structural equation modeling; or LISREL, clarifies the direct and indirect interrelations among variables relative to a given variable, but it provides more theoretical validity and statistical precision in the diagram it produces.
o Factor analysis.
PROBLEMS TO CONSIDER IN INTERPRETING CORRELATION COEFFICIENTSo The quality of the information provided in correlation
coefficients depends on the date they are calculated from. It is important to ask the following questions when interpreting correlation coefficients:
1. Was the proper correlation method used to calculate the correlation?
2. Do the variables have high reliabilities?3. Is the validity of the variables strong? Invalid variables
produce meaningless results.4. Is the range of scores to be correlated restricted or extended?
Narrow or restricted score ranges lower correlation coefficients, whereas broad or extended score ranges raise them.
5. How large is the sample? The larger the sample, the smaller the value needed to reach statistical significance. Large samples may show correlations that are statistically significant but practically unimportant