Experimental designs

 

The workshop on experimental design on the 19th March gave me pause for thought. Did my study fall into the experimental design category or not?

Scenario

  • Observational vs experimental. Observing eg. Observing neural activity. (Note to self – I am doing an observational design in my study?) Experimental – researcher intervenes.
  • Experimental assumes causal relationship – cause and effect. For example…Lack of sleep affects concentration in a MARC class the following day.
  • Hypothesis: Lack of sleep (i) will cause lower concentration (d) in a MARC class the following day.
  • Population = marc class
  • Randomise 2 groups of 6 (pick numbers out of hat?)
  • Experimental group = lack of sleep no more than 3 hours (according to the literature)
  • Control group = more than 3 hours
  • Class sleeps in a laboratory and sleep monitored. Once a month over 5 months on sequential days of the week.
  • To test: Use a concentration questionnaire? Or task? (according to literature)

5 conditions needed to meet conditions of causality

  1. Cause has to precede the effect
  2. There should be a correlation (a relationship)
  3. The effect should not happen in the absence of the cause
  4. The cause and effect should take place close in time
  5. All other explanations of why the cause-effect relationship is happening are ruled out.

Hypotheses

  1. Cause = independent. Outcome = dependant variable eg. Lack of sleep = independent. Alertness = dependant

Manipulation of variables

  1. Manipulating cause /independent variable
  2. Measure outcome – dependent variable
  • Participants should not know what experiment is about.
  • Control group – didn’t receive treatment or manipulation
  • Other group is experimental group
  • Simple experiment one control group and 1 experimental group.

Variables

  1. Variables can be manipulated or measured or observed.
  2. Direct measurements g. speed, height etc.
  3. Indirect measurements. Measuring of what we can’t directly see ie. Latent measurements.

Categorical data = nominal or ordinal. Only binominal.

Continuous data = always quantifiable. Can be put on a continuum

Simple experiment – 2 conditions

Multilevel experiment – 3 or more conditions.

Field experiments

  1. Not randomly allocating subjects
  2. Less control over your setting

There is a difference between Experimental approach vs. experimental design.

Factorial experimental design: 2 or more cause variables