|| Introduction|| Descriptive Research || Correlational Research || Survey Research || Ex Post Facto || Course Overview Map ||

Introduction

Nonexperimental research designs describe something that has occurred, or examine the relationships between things. They are directed toward determining the nature of a situation as it exists at the time of the study. Nonexperimental research involves no administration or control of a treatment as is found in experimental research. With the exception of causal-comparative research, nonexperimental research is not generally directed toward hypothesis testing. The aim is to describe "what exists" with respect to the variables or conditions in a situation. There are four types of nonexperimental designs: Descriptive, correlational, survey, and ex post facto. The statistical analysis procedures involved in nonexperimental research are descriptive data analysis techniques. This session will focus on the above four types of nonexperimental research designs and the associated descriptive data analysis techniques.

Descriptive Research


1. Variable Types and Permissible Statistical Analysis

Variable Type Characteristics Nature of Data Sources Applicable Statistical Analysis
Nominal (Disrete) Numbers represent categories. Numbers do not reflect differences in magnitude. Numbers serve to distinguish groups Qualitative and quantitativeFrequencies
Ordinal (Discrete) Numbers indicate rank order of observations Qualitative and quantitativeFrequencies
Interval (Continuous) Numbers represent equal units (intervals). Intervals between observations can be compared Quantitative Frequencies, central tendency and variability
Ratio (Continuous) Numbers represent equal units from absolute zero. QuantitativeFrequencies, central tendency, variability, and ratio

2. Descriptive Statistical Analysis

  1. Frequencies
  2. Central tendency: means, median, and mode
  3. Variability: variance, standard deviation, and range

3. Standards of Adequacy

  1. Is the research problem clearly descriptive in nature?
  2. Is there a clear description of the sample, population, and procedures for sampling?
  3. Will the sample provide biased or distorted results?
  4. Is the instrumentation reliable and valid?
  5. Do graphic presentations of the results distort the findings?
  6. Are inappropriate relationship or causal conclusions made on the basis of descriptive results?
  7. If cross-sectional, do subject differences affect the results?
  8. If longitudinal, is loss of subjects a limitation?
  9. Are differences between groups used to identify possible relationships?

Correlational Research

1. Simple correlational studies

Type of Correlation Type of VariablesSymbol
Pearson product-moment Both continuous
Spearman Both rank-ordered (Ordinal)
Biserial One continuous, one artificially dichotomous (Nominal)
Point-biserial One continuous, one truly dichotomous (Nominal)
Tetrachoric Both artificially dichotomous
Phi Both truly dichotomous
Contingency Both 2 or more categories (Nominal or/and ordinal)
Eta (correlation ratio for curvilinear relationships) Both continuous

2. Prediction Studies

3. Statistical Analysis

  1. Scatterplots and correlation coefficients
  2. Simple and multiple regression

4. Standards of Adequacy

  1. Does the research problem clearly indicate that relationships will be investigated?
  2. Is there a clear description of the sampling? Will the sample provide sufficient variability of responses to obtain a correlation?
  3. Is the instrumentation valid and reliable?
  4. Is there a restricted range on the scores?
  5. Are there any factors that might contribute to spurious correlation?
  6. Is a shotgun approach used in the study?
  7. Are inappropriate causal inferences made from the results?
  8. How large is the sample? Could sample size affect the "significance" of the results?
  9. Is the correlation coefficient confused with the coefficient of determination?
  10. If predictions are made, are they based on a different sample?
  11. Is the size of the correlation large enough for the conclusions?

Survey Research

1. Types of Survey

2. Constructing the Survey

3. Selecting the Survey Method

4. Standards of Adequacy

  1. Are the objectives and purposes of the survey clear?
  2. Is it likely that the target population and sampling procedure will provide a credible answer to the research question(s)?
  3. Is the instrument clearly designed and worded? Has it been pilot tested? Is it appropriate for the characteristics of the sample?
  4. Is there assurance of confidentiality of responses? If not, is this likely to affect the results?
  5. Does the letter of transmittal establish the credibility of the research? Is there any chance that what is said in the letter will bias the responses?
  6. What is the return rate? If low or borderline, has there been any follow-up with nonrespondents?
  7. Do the conclusions reflect return rate and other possible limitations?

Ex Post Facto (Causal Comparative) Research

1. Procedures of Ex Post Facto research

  1. Formulate the research problem that includes possible causes of the dependent variable
  2. Identify plausible rival hypotheses that might explain the causal relationship
  3. Select the groups that will be compared
  4. Collect and analyze data on the subjects
  5. Interpret the causal relationship

2. Standard of Adequacy

  1. Was the primary purpose of the study to investigate cause-and-effect relationships?
  2. Have the presumed cause-and-effect conditions already occurred?
  3. Was there manipulation of the independent variable?
  4. Were the groups being compared already different with respect to the independent variable?
  5. Were potential extraneous variables recognized and considered as plausible rival hypotheses?
  6. Were causal statements regarding the results made tenuously?
  7. Were threats to external validity addressed in the conclusions?