Module 8, Quantitative Research Design Flashcards
Cross Sectional vs. Longitudinal Designs
cross sectional
- collect data at only 1 time point
- snapshot of the data
- pros: easier & cheaper, less participant burden, more people in your study (because cheaper and easier)
- cons: cannot look at different changes and trends overtime (interested in this)
longitudinal
- repeated measurement of variables over time (usually the same variables)
- multiple snapshots overtime - looking at someone’s trajectory over a long period of time
- participants often drop out and no longer want to participants and thus very challenging to do
- it is important so they can capture how people develop overtime
Between-Groups Designs
- separate group for each condition
◦ participants only provide
data once - when we have a between-groups designs, that between groups or whatever your grouping is, it is oftentimes one of your independent variables
Repeated-Measures Designs
- observations are taken from the same participants more than once
- in a repeated measures design ‘time of observation’ is an independent/predictor variable
my notes:
- same variables are being measured over several observations
- the number of observation is going to be the level of that IV (5 observation would mean 5 levels)
- it is when you actually takes measures from participants: before intervention, after intervention, a month after etc. (explain why that time needs observations (why and when you take measurements)
- we can look at change over time with repeated-measures designs
Mixed Factorial Designs
- both between-group and repeated measures elements
- this design is to see how groups differ in their change overtime
- 2 levels of the IV (the levels of the IV will vary to each between groups and repeated measures) - treat them separately when quantifying levels
Non-Experimental Designs
- research designs where there is a theoretically presumed cause and effect
- no control group and random assignment
- “observational design”; “correlational design”
- predictor/criterion
Experimental Designs
- the researcher manipulates the independent variable(s) to determine if it has an effect on the dependent variable(s)
- manipulation of the independent variable is what separates the experimental from non-experimental
Notation (TORN)
T = treatment (intervention)
O = observations
- represent when variables are measured, or in other words, when data are collected
R = randomly assigned groups
- equal probability of being assigned to any of the groups
N = non-equivalent groups
- groups of participants are typically already intact
- templates that can be adapted or adjust them (change) to help address your research question
Pre-Experimental Designs
so you can identify what is not a good experimental design - you cannot determine cause and effect
- participants receive the intervention/treatment of interest
- limited control over threats to internal validity (this is bad as we want control over internal validity in experimental designs - to see if the IV is the cause in DV)
- participants not randomly assigned to groups (no random assignment)
- changes in the dependent variables cannot be attributed to the manipulation of the independent variable
One-Shot Study (pre-experimental)
group A T 01
- can conclude that at the time of the observation participants performed in a certain manner
- 1 group that receives treatment or intervention and then there is 1 observation
example: we have a PE class, do PA intervention (PE class) and at the end of the year you do a cooper’s test
- cannot really say if the intervention did anything because we did not do a pre-test
- did not have a control group so we do not know if improvements are being seen overtime or the causation of extraneous variables
- all we can say is whatever they scored in the cooper test (snapshot) - this many km on a cooper test
One-Group Pre Test - Post Test Study (pre-experimental)
group A 01 T O2
- try to determine the magnitude of the treatment effect
- repeated-measures design
- researcher cannot attribute any changes in performance to the treatment
my notes:
- there is an observation before and after intervention
- you can see if there was a change from beginning to end of year of the cooper’s test (if there is change overtime) - did they change in their aerobic capacity
- cannot say what those changes were due to (maturation etc.)
- time of observation is an independent variable (2 levels)
Post Test Only with Non-Equivalent Groups (pre-experimental)
static group comparison
N group A T 01
N group B O2
- between-groups design (IV)
- groups are not randomly formed (intact groups)
- cannot be determined if group A and B are equivalent at the beginning of the study
◦ differences between O1 and
O2 cannot be attributed to
the treatment
◦ all we can see is that group a
performed this on the
coopers test and group b
performed this
- one class gets aerobic capacity increasing intervention and group b plays benchball
- no repeated measures element
- non-equivalent means that the people who are taking class a to class b might be different (have this when you do not do random assignment)
True Experimental Designs (random assignment)
random assignment
- equal probability of a participant being placed into a level (group) of the independent variable
- assumes personal factors (selection bias) that could influence participants scores on the dependent variable are distributed evenly across groups (groups are equivalent), thus changes in the dependent variable are more likely due to the manipulation of the independent variable
helps control for:
- past history
- maturation
- testing
does not control for:
- measurement errors
- something happening other than the treatment to one group, but not the other group
- experimental mortality
my notes:
- there is variability in groups but personal characteristics get divided equally between groups
Post Test Only Control Group Design (2 groups) - true experimental designs
R group A T O1
R group B O2
- theoretic argument that differences between 01 and O2 could be attributed to the treatment
- can start to make inferences about independent variable impacting dependent variable
- between groups design / no repeated measures
- making assumption that pre-test would be equivalent
- can attribute differences to the IV
Post Test Only Control Group Design (3 groups) - true experimental designs
R group A T1 01
R group B T2 O2
R group B O3
- 3 levels of the independent variable
- determine if one treatment is more effective than another treatment and if it is better than nothing
- no repeated measures
- between groups design
- group a is getting one treatment and group b is getting another treatment
- is this intervention better than what is already available
if you add observations before and after for instance:
if you add a pretest before the intervention can see if groups are actually equivalent at beginning of intervention and see how much they have changed over intervention (now you have repeated measures so you have a mixed factorial design now)
- 2 levels of the independent variable for time of observation
- independent is time of observation and what type of elbow support you have
- do follow up to see if treatment is actually sustained
Pre Test-Post Test Control Group Design
R group A 01 T O2
R group B O3 O4
- mixed factorial design (both repeated measures and between group measures)
- two independent variables (time of observation and something that represents group A and group B)
- determine how much more change is observed in Group A vs. Group B
- can be extended to include more observations and independent variables
