Experimental_research
- 1. 1/04/2023 Tom Cockburn Wintech 2003 1 Lecture 8 Experimental research method
- 2. 1/04/2023 Tom Cockburn Wintech 2003 2 Experimental method Experimental research is a comparison-between- groups. The experimenters are seeking to discover or corroborate cause and effect relationships by changing something to see if the change causes a difference to another thing. They alter the independent variable (IV) to see if it causes a difference to a dependent variable (DV).
- 3. 1/04/2023 Tom Cockburn Wintech 2003 3 A Classic, experimental method involves assigning subjects at random to either an ‘experimental’ group or a ‘control’ group. The ‘control’ group is compared with the ‘experimental’ group to assess the effects of the experiment on the experimental group.
- 4. 1/04/2023 Tom Cockburn Wintech 2003 4 Switching on the light The independent variable(IV) can be likened to a light switch and the dependent variable (DV) can be likened to a light bulb. An experiment compares the results of exposure of one (independent) variable upon another (the dependent variable). Switch on (the IV ‘switch’) and the (DV) bulb (metaphorically) illuminates the room.
- 5. 1/04/2023 Tom Cockburn Wintech 2003 5 Advantages of experimental research methods The experiment is probably the best scientific method for attempting to prove or corroborate causal relationships. It allows us to locate the variables which control or cause a problem to occur. For some researchers such as those in the physical sciences the experiment has good internal validity because of the degree to which all ‘extraneous’ variables are controlled.
- 6. 1/04/2023 Tom Cockburn Wintech 2003 6 The experimental process Consider-as a group-how we could set up an experiment to discover the impact of noise upon the recall of radio news. How could we test the hypothesis that noise affects recall of radio news? What would we need?What is the IV ‘switch’? What is the DV ‘light bulb’ being switched on/off?
- 7. 1/04/2023 Tom Cockburn Wintech 2003 7 The noise hypothesis –1 experimental procedure 1. Select two groups. (caveats/qualifiers?) 2. Expose them to the same news broadcast. 3. One is under ‘noisy condition’ and other is under ‘silent condition’.(definitions?) 4. After exposure to the news each group is tested separately,using same recall test.(test admin. Issues= ?) 5. IV=noise and DV=recall of news content.
- 8. 1/04/2023 Tom Cockburn Wintech 2003 8 Aims of experimental research Establish cause and effect relationships. Then, like a switch we can turn the phenomena ‘on’ and ‘off’. This is useful as the ability to switch things on or off means, a) we can be reasonably confident we’ve found variables that cause or impact on the problem significantly, and b) we can use that knowledge to control or adjust such phenomena. Such knowledge has advanced natural science and medicine as mentioned.
- 9. 1/04/2023 Tom Cockburn Wintech 2003 9 Issues in any experimental research Internal validity--This is summed up in the following question -Does the experiment test what we say it tests? How sure can we be that the changes in the DV we measured were caused by the changes in the IV? Are all extraneous variables or ‘confounding’ variables (ie one which changes as IV changes) controlled? For example, how sure can we be about the ‘noise’ and recall of news experiment’s internal validity?
- 10. 1/04/2023 Tom Cockburn Wintech 2003 10 Possible alternative explanations or extraneous variables? Boredom-don’t listen to news normally and not interested in the experiment. Nothing else to do-cognitive dissonance ie make it seem better use of time/less of a waste of valuable time? Trying to curry favour by participating in teacher’s test Hawthorne effect. Non-aural learner eg visual or tactile or sociable
- 11. 1/04/2023 Tom Cockburn Wintech 2003 11 External validity means how strong are the generalizations we can make from this research (if any)?
- 12. 1/04/2023 Tom Cockburn Wintech 2003 12 Threats to validity of experiments Campbell in Bynner & Stribley (1978) identified 9 threats to internal validity and 6 to external validity. Internal validity threats 1. History : mediating event or factor between pre- and post-test= alternative possible explanation of effects.e.g. Easterby and Ashton’s study of University Management Development Programme’s Impact on students’ careers. (Managers’ Colleagues used absences to their own advantage).
- 13. 1/04/2023 Tom Cockburn Wintech 2003 13 More threats to internal validity of experiments 2. Maturation:changes in subjects or society such as growth,fatigue, social trends. 3. Instability:instability of measures, sampling fluctuations or of repeated or ‘equivalent’ measures. 4. Testing:The effect of taking a first test upon a second or subsequent test or the effects of one social indicator upon future readings of that indicator e.g. ‘priming’ subjects ,stock prices.
- 14. 1/04/2023 Tom Cockburn Wintech 2003 14 Even more threats 5. Instrumentation:changes in calibration of equipment, or observers, or scores used. 6. Regression artifacts: so-called ‘pseudo-shifts’ which occur when people or treatment units have been selected upon the basis of their extreme scores.ie. Regression to the ‘norm’ or middle scores in following groups or years eg selection of years of droughts or floods or high fatality years-following years may show improvement even if ‘untreated’.
- 15. 1/04/2023 Tom Cockburn Wintech 2003 15 Yet other threats 7. Selection: bias resulting from differential recruitment of comparison groups-leads to different mean scores for measures of effects. 8. Experimental mortality: Loss of people from comparison groups ( they die, they move, they get new jobs etc).
- 16. 1/04/2023 Tom Cockburn Wintech 2003 16 Finally--Interactive threat to validity 9. Selection-maturation interaction: selection biases resulting in different rates of maturation or autonomous change. Eg on e group matures quicker than another causing selection bias.
- 17. 1/04/2023 Tom Cockburn Wintech 2003 17 Threats to external validity and interpretation of results 1. Interaction effects of testing- ‘priming’. 2. Interaction of selection and ‘treatment’- unduly ‘responsive’ outcomes. 3. Reactive effects- ‘articifiality’ & ‘Hawthorne’. 4. Multiple treatment interference- cumulative effects untypical of separate treatment eg First Gulf war syndrome
- 18. 1/04/2023 Tom Cockburn Wintech 2003 18 5. Irrelevant responsiveness measures- complex measures,some irrelevant ones may appear to produce effects. 6. Irrelevant replicability of treatments- complex treatments & replications may ‘forget’/omit some components actually responsible for the effects.
- 19. 1/04/2023 Tom Cockburn Wintech 2003 19 Social science experiments Asch(1955)-Opinion & social pressure Ekman et al (1987)-Universals & cultural differences in judgments of facial expressions of emotion. Milgram (1964)-Obedience to authority Gordon(1992)Treatment of depressed women by nurses. Zimbardo(1974)-psychology of imprisonment