RM1 Step 6: Select a Research Design (pt 2)

Within-subject design

Within-subject design (WSD)

Research by comparing scores within the same group of individuals

Procedure of a repeated-measures design

The same group of individuals go through the control condition, then the experimental condition

Advantages of WSD

1. Within-group variances are not damaging as participants are being compared to themselves

2. No noise = can easily detect differences between treatment and control

Disadvantages of WSD:

1. Time-related variables could confound findings, and threaten IV, as it often requires a series of measurements made over time

5 categories of time-related confounds

1. History

2. Maturation

3. Instrumentation

4. Regression towards the mean

5. Order effects (practice, fatigue, and contrast effects)

These can only be considered confounds when:

It influences enough of the participants to have an effect on the dependent variable 

How do these threaten IV?

Differences in treatment and condition may be caused by these factors instead of the treatment 

History

Environmental events (external events that occur in participants’ lives) that affect their performance or behaviour, which may affect scores in one treatment

Maturation

Inevitable systematic change in participants’ physiology or psychology across a research study, which affects scores

Instrumentation

Changes in measuring instrument over time

Maturation and instrumentation are of concern when:

The study extends over a long period of time

Regression to the mean

The tendency for extreme scores on any measurement to move towards the mean (regress) when the procedure is repeated; after an outlier score, one will likely go back to their normal true ability for the next round(s)

How else could regression to the mean threaten the validity?

Fluctuations in scores of any low-reliability assessment (eg. involving luck)

Order effect & the 3 types

Participation in one treatment may have an influence on the scores in the following treatments

1. Practice effect

2. Fatigue effect

3. Contrast effect

Practice effect

Progressive improvement in performance as participants gain experience through a series of treatment conditions

Fatigue effect

Progressive decline in performance as participants work through a series of treatment conditions

Contrast effect

When subjective perception of a treatment condition is influenced by its contrast with the previous treatment

Eg. participants who just got out from a room of 10°C may feel that a room of 40°C is not hot

3 Ways to deal with time-related threats and order effects

1. Controlling time

2. Switch to BSD

3. Counterbalancing: matching treatments w.r.t. time

Controlling time

Have a good balance in time interval: short enough to avoid time-related confounds, but long enough to dissipate order effects

Switching to BSD is better in situations when:

1. There is strong order effect eg. comparing two methods of teaching reading to children

2. Measurement can only work effectively for one time eg. using confederate to measure helping behaviour

Counterbalance: general procedure

Split the pool of participants into two, one group go through the original order (control > treatment), and the other group go through the reverse order (treatment > control)

Counterbalance: how it works

Distribute any possible confounds evenly across all conditions (BUT does NOT eliminate it!)

Use counterbalance when:

You insist on doing WSD, as you are limited with your sample size