Chapter 6 Flashcards Preview

A descriptive question is designed to produce information about what is or has been happening in relation to the target of the research. 

Examples:
How much debt do undergraduate students accumulate upon graduation?

How do counselors in high schools support students with symptoms of depression?

A correlational question is designed to determine whether, and to what degree, a relationship exists between two or more variables (numerical). It implies prediction, but NOT causation. 

Examples:
What is the relationship between student debt load and student status as a first or second generation college goer?

What is the relationship between enrollment in a company mentoring program and length of time employed with the company?

An experimental question is designed to investigate a cause-effect relationship. An experimental treatment is administered and includes random-assignment of participants.

Examples:

What is the effect of self-paced instruction on student self-concept?

What effect does positive reinforcement have on student attitudes towards school?
 

A quasi-experimental question is designed to investigate a cause-and-effect relationship when an experimental treatment is administered, but with intact groups that cannot be manipulated.

Examples:
What is the effect of gender on science achievement?

What effect does participation in an intensive reading program in grade 1 have on student  reading performance at the end of grade 5?

creating conditions that allow the researcher to get a clear look at the variable of interest by limiting or eliminating the influence of some variables and explaining the influence of others

It is variability between or among individuals, groups, and/or conditions on outcomes of interest

-can be expressed quantitatively as a real, positive number

- variance of zero indicates that all scores in a distribution or identical

“It is creating conditions
that allow the researcher  to get a clear look at the variable of interest b
limiting or eliminating the influence
of some variables  and
explaining the influence of others.” 

1. Randomization (to your treatment & control group)

2. Building conditions or factors into the design as independent variables

3. Holding conditions or factors constant

4. Statistical adjustment/control
 

The random assignment of participants spreads the effect of whatever is causing the variance evenly across the groups in the study.   This could be ability, motivation, math knowledge, prior chemistry knowledge.

Two key points:
-randomization essentially equalizes the groups with respect to ability level UT it also equalizes them with respect to together variables such as motivation, attitude, and prior achievement
-Although the process distributes the effect of ability level evenly , it does not allow the researcher to quantify the influence of ability level on the scores of chemistry test.
 

If  measures are available for a second variable, we can include it as a second independent variable

Scenario: 15 Chemistry students from each of two ability groups – high and low – are assigned to  each method
How? Rank; split sample 50/50: median split

Result:  Research can determine if there is a difference among 3 methods and a difference also among higher or lower ability groups.

When a factor is held constant– that is, It becomes a characteristic that has the same value for all individuals in the study – a potential variable is then reduced to a constant.

This eliminates, or substantially reduces, any effect the factor may have on the dependent  variable.

Ex: In short, eliminating the effect of ability in the study 

Holding factors constant can have some disadvantages
-- logistical problems – you are eliminating  individuals from the study

-- you are reducing amount of data  available on the dependent variable

-- you are reducing how broadly you can generalize your results. Now results generalize to a restricted group: high ability students. This restricts external validity.

1. Freedom from Bias

2. Freedom from Confounding

3. Has Controls on Extraneaous Variables

4. Statistical Precision to Test Hypothesis with confidence

Provides data that allows a fair, unbiased comparison of groups

Care taken to ensure that an difference between groups can be attributed to the independent variable understudy.

Data does not vary in any systematic way beyond that due to the independent variable

Managed by randomization
 

Good design ensures that the  effects of independent variables can be separated and results interpreted without confusion.

Confounded variables – when the effect of two or more independent variables on a dependent variable cannot be separated

Attempt to apply treatment so that outcomes are only attributable to known/desired variables

Good design balances, minimizes or eliminates the effect of variables that may influence the dependent variable.

Scenario: Chemistry Class Example

Student ability is an extraneous variable and
is controlled by (1) including ability as a constant (only high ability students participate) or  (2) as an independent variable (sort students as high/low ability).

Precision increases with large samples and
when random/error variance is decreased
by building additional variables into the design.