- Lack of internal validity - the presence of other's may have influenced the rate of conformity.
Crutchfield (1955): Study did not involve face-to-face contact. Participants answered questions by pressing buttons in private booths. Experiment was rigged to participants were last to answer - they could see other's answers but not each other. The results were similar to Aschs' with 30% conforming to wrong answers - clearly influenced by what they believed to be answers of other participants.
Supports: Demonstrates pressure to conform can operate without face-to-face communication.
- Culture bias - All american participants from 3 US colleges. Importance placed on conformity may differ amongst cultures.
Smith & Bond (1997): Replicated using participants from Zimbabwe and Belgium. Found percentage of errors varied considerably from 14% in Belgium to 51% in Zimbabwe. Maybe due to experimental design which resulted in minor changes.
Contradicts: Demonstrates different cultures (Zimbabwe+Belgium) generate results considerably different to American participants' results.
Asch (1955) - Continued
- Lacks validity
Perrin & Spencer (1980): Called the study "a child of it's time", as it was era specific to the 1950s. The USA in the 1950s was a highly conformist society due to the Cold War.
Repeated Asch's study in Britain 1970s and found only 1 student conformed in 396 trials. Maybe cultural customs of the time 1950s distorted the results of Aschs'
Contradicts: The assumption was people would adhere to group norms, whatever the situation and whenever it happens, but only 1 did in 1970s Britain.
- Lack of ecological validity
Hoffling et al (1966) addresses this as found that nurses in a real hospital setting (field study) obeyed the orders from a 'fictional' Dr Smith over the phone, which was against hospital policy. 21 out of 22 nurses administered a 'drug' they believed was real to patients in spite of warnings on the bottle.
Supports: Shows that ordinary people in ordinary settings are just as likwly to obey figures of authority even though, beforehand most nurses claim they would disobey, as Milgram showed what we say and what we do in real life may be different.
- Lack of reliability - conducted in 1960s where obedience levels may have been affected by WW1 - a recent event.
Burger (2009) addressed this by conducting a partial replication and found 70% of participants went to 150V. Due to ethical restrictions it's not possible to assess further levels of obedience.
Supports: Further reinforces that the majority of people would still endorse a lethal amount of volts in a modernised replication of the study, under the orders of someone they percieved to be of authority.
Milgram (1963) - Continued
- Lack of experimental validity - were participants displaying demand characteristics and playing along?
Sheridan and King (1972): assessed this and found similarly high levels of obedience when using real shocks. Small puppy used in place of 'victim' and shocked, even though seen yelping 75% of time, participants still gave maximum shock.
Supports: participants had no doubt it was real yet still obeyed.
- Established correlation between TLCU and illness, but didn't consider cause and effect of relationship
Cohen et al (1993) Asked participants to complete the SRRS. Then were split into 2 groups: the control group - given non-infectious nasal drops and the experimental group - given drops contraining common cold virus. Both groups quarantined for 7 days. They found participants with high LCU scores = more likely to be infected.
Supports: It establishes explicitely that stress does cause illness.
- Didn't consider that personality can effect stress response
Kobasa (1979) Did research with 2 groups of students. One group considered themselves to have 'hardy' personality trait, whereas the other group didn't. They were given the same task to complete and found the group with the 'hardy' disposition found the task less stressful that those who considered themselves less hardy.
Contradicts: Demonstrates that personality traits such as 'hardiness' can influence stress
Rahe (1970) - Continued
- Only considered major life changing events, and not minor daily 'hassles'.
Lazarus (1990) suggested major life changes are relatively rare, but minor daily stressors (hassles) are more significant.
Delongis et al (1998) studied stress in 75 married couples. Gave participants life events questionnaire and hassles and uplifts scale. He found a correlation of +.59 between hassles scores and next day health problems (e.g. flu/sore throat).
Contradicts: Found no relationship between life events and illness, but did regarding minor stresses e.g. hassles, proving them to be more significant.
Gardner and Gardner (1969)
- Washoe was not learning equivalent of human linguistics
Terrace (1979) taught ASL to chimp named Nim Chimpsky (after linguist Naom Chomsky - proposed humans have language acquisition device to determine linguistic ability). Nim learnt 125 different signs altogether and in combinations. Terrace recorded over 20,000 combinations over 2 year period, but was disappointed to find difference between Nim's communication and child language as Nim couldn't combine words to create meaning.
Contradicts: Nim's language different to human child's, suggesting Washoe's language acquisition down to imitating previous combinations without using them meaningfully.
- Washoe was chosen as particularly intelligent for species - not a true reflection of primates abilities.
Savage-Rumbaugh (1986) began research with two bonobo chimp's: Kanzi and Malika, taught to use language through lexigrams. Kanzi first learnt through interactions with mother who was taught. Neither specifically taught how to use them - they acquired the knowledge through interest in communicating with human companions. After 17 months, Kanzi able to produce more than 2500 non-imatative communications.
Supports: Suggests it is possible for chimps of other breeds to have the same propensity for language as Kanzi and Mulika learned 2500 communications.
Gardner and Gardner (1969) - Continued
- Most language research shows how language can be simulated - was it truly being learned? For 30 years - computers have been able to simulate conversation.
ELIZA: programme devised in 1960s by Weizenbaum that could respond to a user's input as if a non-directive psychotherapist. Expert judges were not always able to tell difference between ELIZA's responses an that of a real therapist.
Contradicts: Suggests there is a difference between being able to produce language and actually being able to understand it.
- Lacks validity - participants may have asnwered questions with socially desirable responses.
Waynforth & Dunbar (1995) analysed 900 lonely hearts ads in American newspapers to assess differences in male and female choices of mate. They found males sought youth and attractiveness and females used phsyical attraction to attract a mate whereas males used economic status / earning power.
Supports: Shows clearly in real life, humans still tend to indicate the same characteristics in mate preferences to our ancient ancestors when they are not away they are being measured.
- Used a questionnaire to assess mate preferences - less internally valid way of measuring as what is deemed as good looks is not clearly defined.
Singh (1993) found in an experiment that there is universal preference for males, for women to have a low waist to hip ratio of 0.7. This would indicate she is a) not pregnant b) has child-bearing hips - both signs of potential fertility.
Supports: Shows a male mate preference for young healthy females, indicating that the universal trait equate to an overal preference in a mate.
Buss (1989) - Continued
- Assumed physical attributes males were interested in reflected differences in fertility, however, 'baby like' features indicate a need for a caring characteristic - not factored in Buss' study. Thus, criticised as attraction was based on personal characteristics of a mate.
Little (2007) suggested facial symmetry is important in mate selection as it shows an absence of gene mutation. Little found in both UK and a primitive society, they saw this as a more attractive trait.
Supports and develops: demonstrates that reproductive capacity in a partner is valued,due to indicators of physical attractiveness such as symmetry of the face.
Gibson & Walk (1960)
- Only mobile infants could be used on visual cliff, so were at least 6 months old, thus, may have learned depth perception from environment rather than it being innate.
Bower et al (1970) found that human infants as young as 6 days old were more upset by small disc that moved to 8cm from then than a large disc that moved to 20cm from them. This was despite the fact that both objects would have produced the same sized image on the retina.
Supports and develops: Demonstrates that depth perception is innate, as it appears to be present in newborns which did not have time to learn it.
- While G&G's research strongly suggests depth perception is innate, it tells us nothing about which visual cues infants use to percieve depth
Pei, Nettett & Norcia (2007) found that infants can tell depth perception when patterns are bold (such as checkerboard) but not subtle differences in texture as adults can.
Contradicts (to some extent): while basic depth perception may be innate, it suggests the fine-tuned depth perception of adults have, needs experience to mature.
Gibson & Walk (1960) - Continued
- Lacks internal validity - was the visual cliff apparatus truly measuring depth perception.
Witherington et al (2005), noticed in footage of Gibson and Walk's study that some infants braced themselves before touching the glass. He performed his own study and found that the infants with the strongest aversion were those with the most experience of crawling/walking.
Contradicts to some extent: Suggests it may be an innate ability, but one which is subject to maturation when the infant becomes mobile.
- Lacks reliability - carried out in 1973, would same diagnosis be made in modern times?
Slater (2004) replicated, although not as systematic research as Slater had been previously diagnosed with clinical depression. After presenting herself at 9 psychiatric emergency rooms compaining of hearing one auditory hallucination ('thud'), she found in almost all cases she was diagnosed with psychotic depression and given anti-psychotics/depressants.
Supports: Suggests there is still a lack of reliability today in diagnosing mental disorders.
- Lacks internal validity - was it truly measuring diagnostic systems?
Spitzer (1976) found people admitted to his own hospital with schizophrenia were rarely discharged with the diagnosis of 'schizophrenia in remission', as well as in the records of 13 psychiatric units. He argued Rosenhan's findings were unsurprising because diagnosis of psychiatric illness often relies on verbal reports.
Contradicts: In reality, people also would not simple use trickery to get admission to psychiatric hospitals, thus, doesn't truly measure the diagnosis of mental illness.
Rosenhan (1973) - Continued
- Lacks ecological validity - Rosenhan called 'experimenter bias'.
Kety (1974) pointed out that if someone was to drink a quart of blood and present themselves vomitting blood at a hospital, they would predictably be diagnoses with a peptic ulcer, but this does not mean that doctors cannot treat or diagnose ulcers. Thus, it is unsurprising that somebody can decieve a psychiatrist into giving a diagnosis of schizophrenia, but doesn't mean it doesn't exist, cannot be treated or is not appropriate.
Contradicts: Suggests that by the pseudopatients admitting to hearing voices it was only appropriate for them to be treated as if they had a form of psychiatric illness.