Descriptive and Correlational Studies

1
Descriptive and Correlational Studies

• Ed 250
• Dr. Barrick

2
Definition of a Descriptive Study

• Involves collecting data in order to test
  hypotheses or to answer questions about opinions
  of people about a topic or issue.
• Also called survey research.
• Useful in investigating educational problems--
  What are the major problems in the math classroom
  today?

3
Steps in Conducting Descriptive Research.

• Identify a topic.
• Select a representative sample.
• Collect valid and reliable data.
• Analyze.
• Report conclusions.

4
Classes of Descriptive Research

• Cross-sectional Collection of data for several
  groups at a single time. For example, the U.S.
  Census
• Longitudinal Collection of data over time. For
  example, data on how students develop
  mathematically from 10 to 15 years of age.

5
Four Types of Longitudinal Studies

• Trend survey--Select a new sample each year.
• Cohort survey--Select a representative group from
  which to select a sample each year.
• Panel survey--Select a sample and follow this
  sample over time.
• Follow-up study--a cross-sectional study is
  undertaken. Later, this sample is studied for
  changes.

6
Problems with Questionnaire Studies

• Validity. Does the question really ask what you
  want?
• Non-response bias- Halo effect.
• Instrumental bias.
• Each method has a disadvantage- questionnaire,
  interview, observation, and telephone.

7
Criteria for Writing Questionnaires

• Know what information you need.
• Know why you need each item.
• Write items that make sense.
• Define or explain ambiguous terms.
• Include only items that respondents can answer.
• Focus on a single idea or topic.
• Use short questions.

8
More Criteria

• Word question in a positive way.
• Avoid leading Questions.
• Organize items from general to specific.
• Use examples when necessary.
• Try to keep the questions to one page.
• Subject the questionnaire to a pilot test.

9
Non-Response

• If less than 40 of the sample responds, you
  should on generalize the finding to the
  population.
• Follow-up several times. Studies have shown that
  with 4 contacts 80 of the population will
  respond.

10
Cover Letter for a Mailing

• Explain--What is being asked and why.
• The significance of the study.
• Sponsoring institution.
• Status of anonymity and Confidentiality.
• Deadline for return.
• Promise of summary of results and conclusions of
  the study.

11
Observational Studies/ Interview

• Questions are presented orally in person and so a
  more in-depth study is possible.
• Written guide or script with additional
  standardized prompts.
• Other items the same as a well-constructed
  questionnaire.
• Establish rapport and be sensitive to the
  responses.
• Pre-test the interview procedure.

12
Hawthorne Effect

• Study in 1920 by Western Electric.
• To single a group of people out for a study makes
  them feel different regardless of the changes.
• The effect was to bring about a consistent change
  in productivity.

13
Pygmalion Effect

• Documented by Dr.Robert Rosenthal of Harvard.
• The bias the researcher is inclined to project
  into into his methodology and treatment that
  subtly shapes the data into the direction of his
  forgone conclusions.
• Self-fulfilling prophecy.

14
Correlational Studies

• DefinitionInvolves the collecting of data in
  order to determine whether a relationship exists
  between two or more variables.
• If the relationship exists the one variable can
  be used to predict the other.
• A high correlation does not imply causation.

15
Relational Studies

• r is calculated for two variables. From r the
  strength of the relationship is determined by
  
• Do not use the shotgun approach. There must be
  some logical implied relation before a
  coorelational study is conducted.
• After a relational study is completed the natural
  progression is the complete a predictive study.

16
Predictive Studies

• Predictive studies assume linear
  relationship--therefore an r usually greater than
  .80.
• One used linear regression or multiple regression
  to find a prediction equation.
• Forms

17
Example

• I want to find the prediction equation between
  grade in Algebra 2 and College GPA.
• Check that there is a high correlation between
  grade in Algebra 2 and College GPA.
• Obtain a sample.
• Then, perform the following steps to obtain the
  prediction equation.

18
Sample
19
Check the high correlation
20
Steps to calculate prediction equation

• Construct a scattergram to check linearity.

21
I scattergram is a line, continue these steps

• Go to Excel Spreadsheet with the data in columns.
• Click on tools--then data analysis.
• Click on regression.
• Choose the y-range as GPA, X-range as Grade in
  alg 2.
• Select labels and output range.

22
Summary Table
23
Predictions

• Predict the College GPA for a D.
• D1 so
• Predict the grade in Algebra 2, if GPA is 2.4
• x2.6 or c or b-

24
Multiple Regression

• Use several independent variables and one
  dependent----grade in algebra 2 and grade in
  fourth year English to predict college GPA.

25
Multiple independent Variables
26
Summary tables