Miscellaneous Experimental Design Topics Flashcards
What is a summated scale?
A composite measure created by averaging multiple responses that capture the same construct.
Why use summated scales?
They help reduce measurement error and improve reliability.
What is a key requirement for summated scales?
Only use items with the same scale (e.g., do not mix 7-point and 5-point scales).
What is validity in measurement?
The degree to which a scale accurately captures what it is intended to measure.
What are three types of validity?
Convergent, Discriminant, and Nomological validity.
What is reliability?
The consistency of a measurement over time.
How do you test reliability for a 2-item scale?
Use the Pearson correlation coefficient (r ≥ .70, p < .05).
How do you test reliability for a scale with 3 or more items?
Use Cronbach’s Alpha (α ≥ .70, but can be ≥ .60 in exploratory research).
What is the role of Component Factor Analysis?
It identifies which items provide the most reliable measure by examining inter-item correlations.
When is reverse coding needed?
When a survey item is worded opposite to other items measuring the same construct.
How do you reverse code an item in SPSS?
Compute a new variable using the formula: (Max Scale Value + 1) - Original Score.
How do you create a summated scale in SPSS?
Use Transform → Compute Variable, sum the items, and divide by the number of items.
Why must you use parentheses in the numerator when computing summated scales?
To ensure correct mathematical operations.
Why are manipulation checks important?
They confirm whether the manipulation worked as intended.
Where should manipulation checks be placed in a survey?
At the end of the survey, after the dependent variables but before demographics.
What statistical test is used for multiple-choice manipulation checks?
Crosstabs analysis (Chi-Square test).
What statistical test is used for continuous (scaled) manipulation checks?
ANOVA.
What percentage of correct recall in a manipulation check is ideal?
At least 85% correct recall.
What is a pretest in experimental research?
A small preliminary study used to refine stimuli, manipulations, or measures before the main study.
How does a pretest differ from a pilot study?
A pretest focuses on refining elements, while a pilot study stands as a small independent experiment.
What are some methods of pretesting?
Expert reviews, qualitative interviews, or running small-scale surveys.
What is a covariate in experimental research?
A continuous variable controlled in analysis to prevent confounding effects.
Why must covariates be measured before respondents see the manipulation?
To ensure they do not bias the manipulation.
What are two characteristics of a good covariate?
It is highly correlated with the dependent variable but not the independent variable.
What statistical tests involve covariates?
ANCOVA and MANCOVA.
Why do larger sample sizes increase statistical power?
They reduce the likelihood of Type II errors and improve the generalizability of results.
What is the minimum recommended number of respondents per condition?
At least 50 per condition.
Why should sample sizes be roughly equal across conditions?
To ensure that each factor’s effects are independently analyzed.
How can researchers remove extreme outliers?
Use General Linear Model → Univariate → Standardized Residuals (ZRE_1) and remove responses ±2.5 SD.
What is Winsorizing?
Replacing extreme outlier values with the next closest non-outlier value instead of deleting them.
What is a control condition in an experiment?
A baseline condition used to compare with treatment conditions.
Why are control conditions important?
They allow researchers to isolate the effects of an experimental manipulation.
What is an example of a control condition in an experiment?
Comparing a group exposed to a new product label versus a group with no label.
What is counterbalancing?
Systematically varying the order of stimuli or conditions to control for order effects.
Why is counterbalancing especially important in repeated measures designs?
To prevent carryover effects from one condition influencing responses in the next condition.
What is an example of counterbalancing?
Showing half of participants Ad A first and Ad B second, while the other half see Ad B first and Ad A second.
