Competence II: Ch 10 & 20 Flashcards
The heart of descriptive designs:
the quality of observation is the heart of descriptive designs; Ex: Investigator can exercise considerable care in making observations so as to minimize measurement errors which can reduce considerable error variance.
Exercising experimental control in descriptive designs
to reduce bias and extraneous variables by using random sampling.
Author-made surveys:
instruments lacking reliability or inappropriate for variables being studied. Created but not lack knowledge about conducting sound survey research.
Most frequent way of collecting survey research data:
self-report questionnaires either mailed or online
Advantage: ease of data collection
Disadvantage: difficulty of getting participants to respond and return completed questionnaires.
The purpose of follow-up letters in survey research:
This acts to remind participants to complete the survey thus increasing response rate.
-Ex: 30%-40% response rate from initial mailing, 20% 1st follow-up, 10% second follow up (50% response rate is recommended minimum)
Adequate return rates in survey research:
Recommend 50% minimum is adequate, others at least 80-90%.
Challenges peculiar to survey research with diverse populations
- How to tailor the survey to the particular group under study; 2. Catering to participants who do not speak standard English, translation process; 3. Distribution and collection of survey- some groups do not trust researchers connected with white middle class institutions.
Construct of interest in scale construction:
The first step in scale construction; Before develop new scale researcher should consider 1. Assessing the necessity for a new scale for the construction and population of interest, 2. Conceptualizing the content of the construct and writing its operational definition, 3. Consulting with colleagues regarding the utility of the scale and appropriateness of the definition.
-In order for an item to be operationalized must be written in a statement that is measurable.
Item generation:
Third step in scale construction; Poor items jeopardize the construct validity of the scale, misleads future researchers who wish to use the scale. Enhance quality of items by 1. Basing the items on solid lit review and conceptual models or theories 2. Using qualitative methods such as focus groups & interviews with relevant groups of people to id prototypical dimensions or indicators of the construct 3. Writing conceptually and linguistically clear items
-Kline has 9 rules for development of writing items p.502
Content validity of test items:
Ensure respondents do not perceive an item differently from what the researchers intended. Therefore conducting a content analysis and consult with domain experts increases content validity of scale.
Purpose of pilot testing test items:
identify potential problems with their wording; Involves asking participants not only to respond to the items as if they were a participant but also to identify unclear or ambiguous elements about the items. Provides powerful tool for researchers to identify items that may be misunderstood or unclear to respondents.
Translation and back translation:
lack of care in translation/ back translation can result in measurement error in the scale and mislead readers about the generalizability of the construct across cultures.
Classification or data reduction research designs
reduce data to only a few variables by developing categories, subgroups, or factors, can also have important theoretical implications; Two common strategies: Factor analysis and cluster analysis
Exploratory factor analysis:
attempt to determine underlying dimensions without a priori specification of the number or content of the constructs (Ex: Enloe assessment ex: Perihan data set)
Confirmatory factor analysis:
Based on theory or previous research there is an existing model of a factor structure to be examined to determine how well the existing model fits the actual relationship observed in a new set of data to be either confirmed or disconfirmed. Begin with identifying the number of dimensions one expects to find along with items in the dat set that will correlate with (or load on) each dimension.