Component 2- Psychology Eduquas

Confounding variable

A variable that isn't the independent variable under study but which varies systematically with the IV. Changes in the dependent variable may be due to the confounding variable rather than the IV and therefore the outcome is meaningless. To 'confound' means to cause confusion.

Extraneous variable

Do not vary systematically with the IV and therefore don't act as an alternative IV but may have an effect on the DV. They are nuisance variables (may affect the DV but not in a systematic way) that muddy the waters and make it more difficult to detect a significant effect.)

Realism

The aim of any psychological study is to provide information about how people behave in 'real life' - everyday settings . If the set-up of a study is too artificial or contrived then the participants will not act as they would normally.

E.g. Loftus and Palmer investigated EWT by showing participants a film of a car accident ad asking questions about speed of car - is this realistic?

Mundane realism

Mundane realism = how an experiment mirrors the real world. Watching a car accident on film lacks mundane realism = it's not like everyday experience and this means results of the study may not be very useful in terms of understanding behaviour in the real world.

Generalisation

Point of realism is to be able to generalise the results beyond the particular unique research setting - in particular to be able to understand behaviour in everyday life = the 'real world.'

-If the environment and materials are real (high realism) a study can still lack generalisability. E.g. if all participants in study are American university students, it may not be reasonable to generalise the findings to behaviour of all people

Validity

How true/legitimate something is as an explanation of behaviour. It involves issues of control, realism and generalisability.

Internal validity

Concerns what goes on inside a study. The degree to which an observed effect was due to the experimental manipulation rather than other factors like confounding/extraneous variables.

Concerned with things like:

• Whether the IV produced change the DV or did something else affect the DV like a confounding variable.

• Whether the researcher tested what they intended to test. E.g. to find if watching TV affects quality of homework - you cannot be certain you're testing 'watching the TV' by just having it on = the person my not be watching it.

• Whether the study possessed/lacked mundane realism.

To gain high internal validity researchers must design the research carefully - controlling confounding and extraneous variables and ensuring they are testing what they intended to test.

External validity

Affected by internal validity - you cannot generalise results of a study that was low in internal validity as results have no real meaning for the behaviour in question. The degree to which a research finding can be generalised to other settings/people/over time.

External validity also concerns:

• The place where research was conducted = ecological validity - it may not be appropriate to generalise from the research setting to other settings/everyday life.

• The people who are studied = population validity. If research involved just students/men/Americans etc. It may not be appropriate to generalise findings to all people.

• The historical period = historical validity. If a study was done in the 1950s it may not be appropriate to generalise the findings to people today because many other factors affect behaviour now.

Directional hypothesis (one-tailed)

States the expected direction of the results e.g. people who have plentiful sleep have better marks than people with a lower sleep average. OR Changing it to people who have plentiful sleep have lower marks in class tests than people with a lower sleep average. - still directional but stating the results are expected to go in the opposite direction.

Non-directional hypothesis

States there is a difference between the 2 conditions but doesn't state the direction of the difference. e.g. people who have plentiful sleep have different marks on class tests than people with lower sleep average. = all have a hunch sleep affects performance but there no real evidence to suggest if the effect would be positive or negative.

Null hypothesis

Statement of no effect. E.g. there is no difference in the marks on tests between people who sleep an average 8+ hours to those with a lower sleep average.

Alternative hypothesis

The alternative to the null hypothesis e.g. people who have plentiful sleep have better marks than people with lower sleep average.

In an experiment the alternative hypothesis is called an experimental hypothesis.

Repeated measures design

All participants receive all levels of the IV e.g. each participant does the task with the TV on, then a week later each does a similar test without the TV on. We compare the performance (DV) of the participant on the 2 tests.

Disadvantages of the repeated measures design

• The order of conditions may affect performance = an order effect. E.g. participants might do better on the second because of a practice effect or because they are less anxious. In some situations participants might do worse because of the boredom effect where they get bored of doing the same test again.

• When they do the second test they might guess the purpose of the experiment which can affect their behaviour e.g. participants could do purposely do worse on the second because they want it to appear as if they work less well in afternoon.

Dealing with disadvantages of repeated measures

• Researchers could use 2 different tests to reduce a practice effect but the tests must be equivalent. This can be done by constructing a test of 40 items and randomly allocating them to test A and test B.

• Main way to deal with order effects is using counterbalancing.

• To avoid participants guessing the aims of the study, a cover story can be presented about the purpose of the tests.

Independent groups design

Participants placed in separate (independent) groups. Each group does one level of IV e.g.:

• Group A = does task with TV on = 1 level of IV.

• Group B = does task with TV off = other level of IV.

Compare performance (DV) of 2 groups.

Disadvantages of independent groups design

• Researcher cannot control effects of participant variables like differences in ability/characteristic of each participant. E.g. participants in group A could have better memories then group B = confounding variable.

• Independent groups design needs more participants than repeated measures design to get the same amount of data.

Dealing with disadvantages of independent groups design

• Randomly allocate participants to conditions which should distribute participant variables equally.

• Random allocation can be done by picking participants names out of a hat etc.

Matched pairs design

Use 2 groups of participants but match participants on key characteristics believed to affect their performance on the DV. e.g. IQ or time spent watching TV.

One member of pair is put in Group A and the other in Group B - procedure is then same as independent groups design.

Important to realise characteristics for matching must be relevant to the study, you wouldn't need to match participants on gender if testing memory unless there was evidence women had better memories.

Disadvantages of matched pairs design

• Very time consuming and difficult to match participants on key variables. The researcher has to start with a large group of participants to ensure they can obtain matched pairs on key variables.

• It isn't possible to control all participant variables as you can only match variables known to be relevant but others could be important. E.g. in memory experiment you can match on memory abilities but find some participants had been involved in teaching programme to boost memory skills and should've been matched on this.

Dealing with disadvantages of a matched pairs design

• Restrict the number of variables to match on to make it easier.

• Conduct a pilot study to consider key variables that might be important when matching.

Counterbalancing

Ensures each condition in a repeated measures design is tested first or second in equal amounts. If participants do same memory test first in morning then in afternoon we might expect them to do better on the second test because they have had some practice or do worse because they are bored with the task = order effects - dealt with using counterbalancing. 2 ways to do counterbalancing = 2 conditions:

• Condition A = test done in morning

• Condition B = test done in afternoon.

Way 1 for counterbalancing = AB or BA

Divide participants into 2 groups.

• Group 1 = each does A then B

• Group 2 = each does B then A.

This is still a repeated measures design even with 2 groups - comparison will be made for each participant on performance on 2 conditions = morning and afternoon.

Way 2 for counterbalancing = ABBA

All participants take part in each condition twice

• Trail 1 = Condition A = morning.

• Trial 2 = Condition B = afternoon.

• Trail 3 = Condition B = afternoon.

• Trial 4 = Condition A = morning.

Then compare scores on trial 1 and 4 with trial 2 and 3. Still repeated measures because we are comparing scores of same person.

Approaches preferring lab based experiments

Cognitive, biological, behaviourist approaches.

Approaches preferring non-experimental outside of lab

Positive and psychodynamic.

Laboratory experiment

The most scientific way to conduct research. Laboratory = room equipped to allow scientific research and measurement. Most psychological research that gets conducted in a laboratory tends to be experimental. Many researchers also conduct observational research in a laboratory setting.

Advantages of Laboratory Experiments

• Allow researchers to measure research variables easily making it easier to control confounding or extraneous variables and easier for other researchers to replicate the research.

• Some psychologists need to utilise equipment that isn't portable in research e.g. PET scan machines like Raine et al. used are very large

Disadvantages of Laboratory Experiments

• May cause participants to show artificial behaviour because of laboratory surrounding - reminds them they are taking part in piece of research = participants are on their best behaviour.

• Lack of ecological validity

In field research

Where psychologists conduct and collect research data outside of a laboratory - in a more natural setting. Can happen in many locations like shopping centres, hospitals, schools etc. Environments may be new to participants = not participants natural environments but setting is more natural.

• Participants have gone to the research = Laboratory.

• Researchers have gone to participants = Field.

Advantages of field experiments

• Useful if you want to minimise the artificial nature of research - when people are in real everyday environments they are less likely to be aware of participation in research meaning they act more naturally.

• Allow psychologists to examine behaviour in a huge range of contexts that would be difficult to accommodate in laboratories.

Disadvantages of field research

• More difficult to measure research variables and more difficult to control confounding or extraneous variables. A second researcher replicates the research and may not find the same results because of differences in the settings.

• Difficult for researchers to utilise a full complement of equipment. E.g. brain scanning machines aren't portable and aren't easily used by field researchers.

Online research

being online allows researchers to be able to access participants via internet or social networking tools. Websites allow people to visit the site and become participants of psychological research. Often questionnaires or maybe experimental.

Advantages of online research

• Researcher has ability to access large group of participants. = allows to seek out a diverse sample that may be less culturally biased

• It is cost effective. Researchers frequently use free, cheap software which allows them to quickly put research onto the internet. Costs of this are generally less than posting surveys or having a researcher assistant to complete telephone interviews. Data analysis is generally quicker as participants have already transcribed their responses and analysis can be conducted.

Disadvantages of online research

• Methods used online tend to be limited, most being surveys or questionnaires, some experimental research being conducted online but tends to be infrequent.

• Ethical issues like consent and protection of participants from any risk of harm may be more difficult to deal with. Most internet users are simply asked to tick box if they agree to terms and conditions so hard to tell if they give valid consent. Difficult to appropriately debrief the participants if the need arises. The British Psychological Society offer guidance on how to conduct online research ethically.

Quasi experiment

research method that the experimenter hasn't manipulated the IV directly. Researcher records the effect of the IV on a dependent variable - IV occurs naturally

Natural experiments

Conducted when it isn't possible for ethical/practical reasons to deliberately manipulate an IV. IV varies 'naturally' the DV may be tested in a laboratory.

Quasi Experiment types

Natural experiments = Those with an IV.

Difference studies = Those with no IV.

Examples of natural experiments

• Effects of watching violence = Berkowitz 1970 = did monthly FBI crime statistics on frequency of violent crimes from January 1990 to December 1966. Reported steep increase in violent crimes recorded from November 1963. Proposed result of American population exposed to TV footage of Kennedy's assassination. IV = exposure to TV footage of Kennedy's assassination.

• Effects of TV = before 1995 people living on small island St Helena in middle of Atlantic had no TV, arrival of TV allowed researchers to see how exposure to Western programmes may influence behaviour - Charlton et al. 2000 found no difference in pro/anti social behaviour after introduction of Western TV - IV was no TV and exposure to TV.

In both examples IVs weren't controlled by researchers - took advantage of something that would have been ethically and practically difficult to control.

Difference studies

The apparent IV is also naturally occurring and the DV may be measured in a laboratory. Key feature = IV not made to vary by anyone - simply a difference between people that exists - isn't actually a variable at all. A kind of quasi-experiment = the IV is not something that varies at all - it is a condition that exists. Researcher records the effect of this 'quasi-IV' on a DV. Lack of manipulation of IV and lack of random allocation = causal conclusions can only be tentatively drawn.

Examples of different studies

• Gender differences = Sheridan and King 1972 = tested obedience asking male participants to give genuine electric shocks of increasing advantage to puppy - 54% males gave maximum non-fatal shock but obedience rate for women was 100%. IV = gender - difference that can't be manipulated = not true 'IV'.

• Olfactory abilities of non-sighted individuals = Rosenbluth et al. 2000 = compared olfactory (smell) abilities of sighted and non-sighted children. 1 group of 30 non-sighted children matched to 30 sighted children by age, sex and ethnicity. Found 2 groups of children didn't differ significantly in olfactory abilities but the non-sighted children scored higher (12.1 out of 25) compared to sighted children (10.4) when labelling smells.

In both the characteristics of people, gender or visual ability is investigated = IV of research.

Evaluation: Manipulation of IV

Lack of control over IV = we cannot say for certain that any change in the DV was caused by the IV. causal conclusion not guaranteed as poor experimental design can make them unjustified in lab/field experiment.

Evaluation: Random allocation

In experiment with independent groups design participants are randomly allocated to conditions, this isn't possible in natural/quasi-experiments = may be biases in the different groups of participants.

E.g. In olfactory abilities study the non-sighted people were matched with sighted people - this was done with the intention of removing any possible biases it could mean that people in the 2 conditions are still very different = lack of random allocation may be uncontrolled confounding variables.x

Evaluation: Unique characteristics of participants

Sample studied may have unique characteristics. E.g. in St Helena TV study the people were in a particular helpful pro-social community explaining why violence on TV didn't affect their behaviour but in other studies the advent of TV did have an effect.

Unique characteristics of the sample mean the findings can't be generalised to other groups of people = low population validity.

Demand characteristics

They are a cue that makes participants unconsciously aware of the aims of a study or helps participants to work out what the researcher expects to find.

Example study of demand characteristics

Orne and Scheibe 1964 = Participants had to sit in a room on their own for 4 hours. One group of participants were asked to sign a form at the beginning releasing the experiment from responsibility if anything happened to them during the experiment. Also given a panic button to push if they felt overly stressed. The other group given no information to arouse their expectations. Group 1 showed extreme signs of distress during isolation. This can only be explained in terms of expectations created by the situation.

Orne invented the term demand characteristics to describe the effect of expectations.

Experimental demand characteristics

In experiment participants are usually unsure of what to do, they actively look for clues for how they should behave in that situation - the clues are demand characteristics which collectively convey the experimental hypothesis to the participants.

Researcher bias

Information other than the IV from a researcher that encourages certain behaviours in the participant which might lead to a fulfilment of the investigator's expectations. Anything the researcher does that has an effect on participant's performance in a study other than what was intended. Investigator effects may act as a confounding or extraneous variable.

Investigators unconsciously encourage participants by e.g. spending more time with one group of participants or being more positive with them. The way an investigator asks a question may lead a participant to give the answer the investigator 'wants'.