Longitudinal Methods

Longitudinal developmental design

Making a measurement repeatedly across time.

Repeatedly measuring the same variable over some amount of time, in one sample or in multiple overlapping samples 

Keeping measurement waves spaced over the same amount of time makes statistical models much more simple 

The longitudinal developmental design does not…

do a good job of explaining why something happens or being able to control it. It describes a phenomena and sometimes being able to predict the trajectory 

Longitudinal intervention or experimental design is…

what is trying to accomplish that last bit of explaining and controlling.

You have to split your sample into a control group and an intervention/experimental group. 

Quasi experiment means…

you do not get to decide who is in what group, it is used when you cannot control the thing that is happening (there is no random allocation). 

Lagged regression equation:  

Outcome variable at time 2 = intercept + Outcome variable at time 1 + predictor variables + error 

Advantages of Longitudinal methods

One of the most valid and powerful ways to measure change  

Provides stronger evidence (not causal evidence unless its experimental) for the direction of statistical relationships 

Disadvantages of Longitudinal methods

Time/effort

Time/money

Hawthorne effect

Problems that are unique to longitudinal designs: 

Instrument decay

Regression to the mean

Sample Attrition

Instrument decay…

This is where your measurements or equipment or researcher become better or worse across time. 

Regression to the mean…

Extreme values are more likely to be less extreme when measured again 

Sample Attrition…

Loss of participants over time (a reason can be due to things like stress) 

Important confounds: 

Historical/ societal effect

Testing/ practice/ learning effect

Normative developmental effect

Historical/ societal effect

An external event may effect variables in some or all of your sample, during the study (Covid for example)

Testing/ practice/ learning effect

Participants' task performance may change, or variables may change, because they have completed the task before or answered questions 

Normative developmental effect

Changes in variables may be due to other factors in lifespan developmental (e.g aging)