survey and correlational research designs

SURVEY AND CORRELATIONAL RESEARCH DESIGNS

ROAD MAP FOR THIS TOPIC

  • Survey Design:
    • “Everyone’s an Expert”
    • Types of survey items
    • Rules for writing survey items
    • Administering surveys
    • Issues with sampling and nonresponse bias
    • Possible ethical issues
  • Correlational Designs:
    • Describing the relationship between variables
    • Limitations in interpreting correlational relationships
    • Correlation, regression, and prediction

SURVEY DESIGNS

  • Survey research designs are typically administered in written or oral form.
    • Consists of a series of questions or statements, referred to as items.
    • The purpose is to measure the self-reports or responses of respondents.
    • Surveys may include questions in which participants report about themselves, such as their attitudes, opinions, beliefs, activities, emotions, etc.

TYPES OF SURVEY ITEMS

  • Open-ended Items:

    • Respondents provide any response in their own words, without restriction.
    • Primarily used in qualitative research design.
    • Coding of responses is necessary, which involves anticipating possible examples of responses.
    • Multiple raters and additional statistical analyses may be employed for accurate coding.
    • Example: "What are your views on the role of mothers in children’s potty training?"

  • Partially Open-ended Items:

    • Includes a few restricted answer options with an open-ended option at the end for responses that do not fit provided choices.
    • Responses can be coded numerically.
    • Example: "In what capacity do you (students or faculty or other) view the role of parents as teachers? A. Teaching B. Assessment C. Curriculum development D. Other "

  • Restricted Items:

    • Contains a limited number of predefined answer options that participants must choose from.
    • Likert Scale:
    • A numeric response scale for indicating agreement or disagreement with a question.
    • Advantage: Easy entry or coding for statistical analysis.
    • Limitation: Analysis restricted to provided options.
    • Example: "I have nightmares: (Circle one) 1 2 3 Never Sometimes Often"

RULES FOR WRITING SURVEY ITEMS

  1. Keep it simple:
    • Use straightforward language.
    • Example: Replace "How satiated do you feel?" with "How full do you feel?"
  2. Avoid double-barreled items:
    • Questions requesting a single response to two different inquiries.
    • Example: "I enjoy the time we spend together and dislike the time we are apart."
  3. Use neutral or unbiased language:
    • Avoid language that is offensive; ensure that identification terms are respectful.
  4. Minimize negative wording:
    • Rephrase items to avoid negatives.
    • Example: Change "How much do you not like working?" to "How much do you dislike working?"
  5. Avoid response set pitfalls:
    • Be cautious of participants responding consistently across survey items without regard to content.
    • Reverse Coded Item: Phrased oppositely from most survey items.
  6. Use consistent rating scales:
    • Use a single rating scale throughout the survey.
  7. Limit the points on a rating scale:
    • Scale should be between 3 and 10 points, with exceptions for dichotomous (two response) or bipolar scales.
    • Example: "How do you feel about your ability to find a job that will make you happy? -5 -4 -3 -2 -1 0 1 2 3 4 5"
  8. Label or anchor the rating scale points:
    • Use anchors to clarify the meaning of rating points.
  9. Minimize survey length:
    • Write surveys to be concise yet effective in measuring intended concepts.

ADMINISTERING SURVEYS

  • Written Surveys:

    • In-person surveys:
    • Tend to have higher completion rates but are time-consuming.
    • Mail Surveys:
    • Higher nonresponse rates but require less time from the researcher.
    • Recommendations for increasing response:
      • Include a pre-paid return envelope.
      • Notify potential respondents in advance.
      • Provide a cover letter.
      • Include a small gift.
    • Internet Surveys:
    • Cost-effective and popular, but limited to those with computer access.

  • Interview Surveys:

    • Face-to-Face Interviews:
    • Allow time control but are prone to interviewer bias.
    • Telephone Interviews:
    • Can be interpersonal or automated, may save time but face low completion rates.
    • Random Digit Dialing:
    • Method for randomly selecting telephone interview participants.

  • Focus-Group Interviews:

    • Aiming to gather shared views on a topic among small groups (3-8 participants).
    • Primarily uses open-ended questions, moderated by the researcher.
    • Conversations are recorded for analysis, and carry similar issues to face-to-face interviews.

ISSUES WITH SAMPLING AND NONRESPONSE BIAS

  • A high response rate is critical when administering a survey.
    • Response Rate: The proportion of individuals completing a survey of those asked.
    • A low response rate limits the representativeness of survey results.
    • Aiming for a 75% response rate is ideal, but published research often reports less than 50% response rate.
  • Nonresponse Bias:
    • Occurs when selected participants do not respond, leading to biased results.

HANDLING AND ADMINISTERING SURVEYS

  • Researchers have ethical responsibilities when handling surveys:
    1. Surveys should not be overly offensive or stressful; debreifing may be required.
    2. Respondents must not be coerced into participation.
    3. Avoid harassment during recruitment.
    4. Ensure confidentiality or anonymity of responses.

CORRELATIONAL DESIGNS

  • Correlational Research Design:
    • Measures two or more factors to examine the relationship and patterns among them.
    • Correlation Coefficient:
    • A statistic measuring the strength and direction of the linear relationship between two factors.
    • Values for r can range from -1.0 to +1.0.

DESCRIBING THE RELATIONSHIP BETWEEN VARIABLES

  • Relationships can be positive or negative based on the pattern of change between variables:

    • Positive Correlation:
    • Positive value of r indicates variables increase or decrease together.
    • Negative Correlation:
    • Negative value of r indicates one variable increases while the other decreases.

  • Scatterplot:

    • A graphical representation displaying individual data points (x, y) to summarize the relationship between two factors.

  • Regression Line:

    • The best-fitting straight line minimizing the distance of all data points from it.

  • Strength of Relationship:

    • Determined by how closely the correlation coefficient approaches ±1 (strong) or 0 (weak).

  • Pearson Correlation Coefficient:

    • Measures direction and strength of a linear relationship when both factors are on interval or ratio scales.

LIMITATIONS IN INTERPRETATION

  • Causality:

    • Significant correlation does not imply causation.
    • Reverse Causality:
    • The direction of influence between two factors may be unclear.

  • Confound Variable:

    • An unaccounted variable influencing observed changes in measured variables.

  • Outliers:

    • Extremely high or low scores affecting correlation strength and direction.

  • Restriction of Range:

    • Limited data range alters generalizability of correlation results.

CORRELATION, REGRESSION, AND PREDICTION

  • Linear Regression:
    • A statistical method to derive the equation of a regression line from data points.
    • Predictor Variable (X): Known values predicting another variable.
    • Criterion Variable (Y): The variable being predicted based on known values of X.
    • The formula for prediction is:
      Y=bX+aY = bX + a
    • Where
      • Y = criterion variable
      • X = predictor variable
      • b = slope
      • a = y-intercept
    • Advantages and Cautions:
    • Predicting behavior using regression equations, but caution against predictions from small correlations (close to r = 0).