Psychology Chapter 2: Research Methods

deductive reasoning

• general → specific

  • general principles are tested against the real word

• start with explanation/hypothesis and perform experiment to determine validity

• all conclusions drawn from deductive reasoning are correct

  • ex. all living things require energy to survive. ducks are living things, so they therefore require energy to survive (test - true)

  • ex. all ducks are born with sight. quackers is a duck. therefore, quackers can see. (test - false)

  • the hypotheses may be false, but not the conclusion you make about them

inductive reasoning

• specific → general

  • specifc facts are used to reach a broad generalization

• start with data and form a broad explanation

• not all conclusions drawn from inductive reasoning are correct

  • ex. apples, bananas, and oranges grow from trees. therefore, all fruit grows on trees (false)

how is deductive and inductive reasoning cyclic

• inductive reasoning makes a theory

• that theory generates a hypothesis

• the hypothesis is used for deductive reasoning to prove/disprove

• based on the conclusion from that, a general theory can be made for inductive reasoning

theory definition

• set of ideas that propose an explanation for observed phenomena

• hard to prove because of its complexity

• ex. evolution by natural selection

importance of research methods

written in such a way that someone can repeat the exact process and get the same results

hypothesis

• a testable prediction about how the world works

• if-then statement

  • ex. if the amount of sunlight a tomato plant receives increases, then its fruit yield will increase

• falsifiable - can be proven incorrect

difference between hypothesis and theory

hypotheses are easily testable while theories arent

importance of falsifiability

acts as a filter to ensure all proven information is correct

types of research

• observational

  • case studies

• survey research

  • archival research

• longitudinal study

case study

• research on one/few unique people

• lots of data but hard to generalize to general population

naturalistic observation

• seeing how people act when they’re not watched

  • observer has to blend in

• easier to generalize

• difficult to detect when/if behavior is observed due to observer bias (each having their own criteria) - why you set guidelines for what counts as a behavior

  • ex. the fidgeting in class

• good with animals

surveys

• big data → generalize

• con: only getting from people who are willing to take the survey + they can lie

• good for mode, median, mean

• surface-level conclusions

archival research

• using existing records for research - never interact with participants

• less time and money

• no guarantee of consistency, have to tailor research question to existing data

longitudinal research

• long-term research with same participants

• attrition - majority of people dropping out. have to start with big # of people knowing most of them will leave

• protection from generational bias

• good for psych experiments cause its the same mind changing over time

cross-sectional research

• researcher compares multiple segments of population at the same time

• ex. compare dietary habits of 20 yos to 30 yos instead of comparing when they’re 20 yo and then 10 years later

• does not protect from generational bias

types of variables

• independent

• dependent

independent variables

• the variable manipulated by the experimenters - the cause

• the experimenters are the ones changing it

dependent variables

• measured during the experiment - the effect

• seeing how the independent variable influences it

• experimenters arent directly manipulating it

which is independent and dependent variable: an experiment seeing how sunlight affects plant growth

independent: how mucsunlight being administered to the plant

dependent: plant height

confounding variable

• an outside factor that affects both the independent variable and the dependent variable

• making it falsely appear that a direct causal link exists between them

• ex.

  • independent variable: amount of time studying

  • dependent variable: test scores

  • confounding variable: amount of prior knowledge on the subject

biases in research

• experimenter bias

• participant bias

• single-blind

• double-blind

experimenter bias

experimenter wanting their hypothesis to be true, so they refuse to look at their data without this bias influencing them (confirmation bias)

participant bias

placebo effect (they really want the medicine to work)

single-blind

participants are unaware of which experimental group/control they are part of

double-blind

researchers and participants are unaware of which group they are part of

• people are numbered and assigned to each other and computers/researchers behind the scenes know who belongs to which group

• helps combat experimenter bias

samples in research - definition and how they should be dealt with

• number of participants = n

  • 500 participants ; n=500

• need enough to be accurate but not too much to where it does harm/waste of money

  • the amount depends on your personal goals with the study. case studies need a little bit but surveys/longitudinal need a lot

• groups should be randomly assigned

  • dont put same gender/race/socioeconomic people in the same category

  • but also make your sample size inclusive to avoid confounding variable

• remove bias through statistical analysis

• further investigate into outliers

  • if a drug only works on 2% of people, find out why on them specifically

correlation does not determine causation

A and B are correlated, but to determine if A causes B, experimenters must change A and look at the outcome of B.

correlational research

• only in performed simulations to prevent unethical experiments

• measure the strength of relationship (r) of two variables without manipulating them (good for ethics/predictions)

• attempt to predict the value of one based on the other

• con: can’t establish cause-and-effect relationship using correlation because you can’t manipulate the variables in an experiment

correlation coefficient (r)

• a number from -1 to +1

• indicates the strength/direction of the relationship between variables

• the closer r is to ±1, the more strongly related/predictable the variables are

  • the tighter the dots are together, the closer they are to ±1

• the closer r is to 0, the weaker the relationship/predictability is

statistical analysis

• determines how likely any difference between experimental groups is due to chance (p-value)

  • to find out if there are meaningful differences between two groups

• if your p-value is 0.05 or less, your data is NOT happening by chance and your data is statistically significant

how did they prevent confounding variable in baby lab

• observer not knowing which toy is good and which is evil when having baby choose

• switched shirt colors so the babies weren’t choosing certain shirt colors but actually who they thought were good

• used previous research that proved that babies’ eyes look longer at stuff they like to prove that 3 mos can have morality

A researcher wants to determine if the color of an office has any effect on worker productivity. In an experiment, one group performs a task in a yellow room while another performs the same task in a blue room.

independent: color

dependent: productivity

Researchers want to learn whether listening to fast-paced music helps runners perform better during a marathon. In an experiment, one group of runners listens to fast-paced music while another group listens to slow-paced music.

independent: music choice

dependent: runner performance

Researchers want to determine if a new type of treatment will lead to a reduction in anxiety for patients suffering from social phobia. In an experiment, some volunteers receive the new treatment, another group receives a different treatment, and a third group receives no treatment.

independent: type of therapy

dependent: effects on anxiety

Imagine that you conducted an experiment to address the hypothesis that greater detail in a letter of recommendation would make the letter more persuasive. To address this hypothesis, you create two versions of a letter of recommendation that differ with respect to amount of detail. In one condition, the letter of recommendation has many details. In the other condition, the letter of recommendation has few details. Moreover, you randomly assign participants to conditions. In this experiment, you find that, on the average, the letter of recommendation is perceived to be more persuasive in the condition in which there are many details than in the condition in which there are few details.

independent: amount of detail

dependent: amount of persuasiveness

confounding: number of words

• recs might like more words as opposed to amount of detail

charlie tries to lose weight over 2 months. the first week, he exercises vigorously. the second week, he gives up meat. the third week, he drinks large amounts of water. the fourth week, he eats just bananas. he continues to vary his approach each week. at the end of his 8-week experiment, he concludes that his best weight loss comes from exercise.

correlational study or experiment?

experiment

• independent variable (type of exercise)

• dependent variable (amount of weight lost)

ms. tucker wants to see which teaching method is best for her students. for the first half of the semester, she teaches using lecture. she sees what students have learned using a 50-point multiple choice test. for the second half, she teaches using demonstrations/active learning exercises. she evaluates using the same test again, and discovers that studies have better test scores through active learning strategies

• experiment

  • independent: teaching style

  • dependent: test scores

a new movie shows teen violence against senior citizens. crime rates have increased since the movie release. a reporter compares crime rates before and after the movie to see if the movie causes teen violence.

correlational

• no experiment actually conducted

• confounding variables: time of year, not knowing who did/didnt watch the movie + whether they committed crime, etc.

• violent movie and violence rates are maybe loosely correlated

physician flips a coin to see success of new expensive medicine compared to old cheap medicine. if heads, he administers expensive medicine. if tails, it’s cheap medicine. after 20 patients and 3 months, he asks them to rate the degree of improvement on a 10 point scale. he discovers that the new medicine has no real advantage over the old one.

experiment

• randomized groups

• independent: type of medicine

• dependent: degree of impact

illusory correlation

• when people believe that relationships exist between two things when no such relationship exists

• ex. moon phases determine behavior

operational definition

• precise description of variables

• necessary for readers to understand criteria of what was measured

• helps repeat the experiment

quasi-experimental

experiments where the independent variable isn’t necessarily controlled by the experimenters (ex. sex)

replication crisis

contemporary issue in higher academia. most well-known studies/scientists have made research that is impossible to replicate

reliability

• the ability to produce a given result

• doesnt have to be correct, it just has to consistently produce that result (within small margin of error)

types of reliability

• inter-rater reliability

• internal consistency

• test-retest reliability

inter-rater reliability

how much observers agree on the observations

internal consistency

how much different survey items that measure the same thing correlate with one another

test-retest reliability

how much the outcome of a measure remains consistent through repetition

validity

• how accurate an instrument accurately measures

• answering the thing it was supposed to answer

types of validity

• ecological validity

• construct validity

• face validity

ecological validity

how much results can be generalized/applied to the real world

construct validity

how much a variable actually measures what it’s supposed to

face validity

how seemingly valid a variable is on the surface

research guidelines with human/animal participants

• informed consent

• debriefing (if deception was used)

• minimize harm/pain

informed consent

• description of what you can expect during the study

• includes risks and implications

• TELL THEM THEY CAN LEAVE AT ANY TIME

• confidential information

• parents will sign for minors

deception

• sometimes used if necessary, to maintain the integrity of the experiment

• ex. if testing how people’s opinion of someone is affected by their clothes, you would not tell the participants that that is what they judge them on because it could make them respond inauthentically

debriefing

• says the full experiment including the deception and why it was used

how is the baby lab still meaningful if it was 80% of babies producing results

• it is still reliable, producing consistent results when people repeat it

• though maybe it’s not necessarily valid

IRB - Institutional Review Board

ethics committee that approves/rejects proposed studies + gives feedback

1. Melissa is running a study to see if girls and boys average different amounts of participation in classrooms. She hypothesizes that girls participate more than boys and plans to observe students in various classes + record how often they raise their hands. For the first part of her data collection, Melissa selects two boys and two girls to study. She finds that the girls raise their hands an average of 4.7 times per week while th eboys only raise their hands an average of 1.3 times in the same classes. Do you think this difference is likely to be statistically significant? Why or why not?

2. Next, Melissa expands her study to 50 boys and 50 girls. She finds that girls raise their hands an average of 3.1 times per week and boys raise their hands an average of 3.0 times per week in their social studies class. Do you think this difference is likely to be statistically significant? Why or why not?

3. In a third data collection, Melissa studies another 50 boys and 50 girls. Girls raise their hands an average of 3.2 times per week and boys 2.4 times in math class. She finds out there is tremendous variability because some students (both boys and girls) raise their hands 15 times a week, while others don’t raise their hands at all. How does the variability in responses affect the likelihood of the difference between girls’ and boys’ participation to be statistically significant?

1. no because the sample size is too small. those 2 girls could’ve been super extroverted/confident about their capabilities compared to the boys. we also don’t know about the students’ prior knowledge of the classes and any learning differences any of them may have.

2. same as last question. sample size is hard to generalize and too many confounding variables. also it changed from all classes to just social studies, making it even harder to apply.

3. the outliers could screw the average and make the difference in raising hands statistically insignificant. use IQR to fix. also the class changed again to math…

tyrone wants to study the impact of watching sexually suggestive/explicit television on people’s attitudes toward sex. he plans to test ninth graders because he believes they are still young enough to be highly impressionable. he will solicit volunteers to come after school. half will be assigned to watch one hour of sexually explicit clips from a cable TV show while the other half will view an hour of clips from the same show that aren’t sexual. after watching the TV shows, all participants will fill out a questionnaire about the attitudes toward sex.

1. what additional information might you want to know as an IRB member to decide whether or not it should be supported?

2. benefits and harms?

3. what would your recommendation be?

1. research that proves 9th graders are impressionable, promise of parental consent, option to opt out, how many students in each group and how long the clips are

2. benefits: know impact of porn on young minds - harms: could further explore it when they shouldn’t, get groomed in those spaces, develop porn addiction → unhealthy relationship with sex

3. not do the study at all cause the harms outweigh the pros. you also need a bigger age range not just 9th graders. do a correlational study instead?

priya wants to see whether listening to music while working out makes people exercise harder. she plans to ask college students to come to the gym and run on a treadmill for half an hour either while listening to music or in silence. the dependent measure will be the number of miles run in that time period.

1. what additional information might you want to know as an IRB member to decide whether or not it should be supported?

2. benefits and harms?

3. what would your recommendation be?

1. what kind of music specifically? hertz level? what does “exercise harder” mean? how many students?

2. benefits: people who want to lose weight can use the results to help; harms: maybe someone falls/faints but not much harm

3. divide in gendered groups as well with similar heights/weights + address questions above

charlotte wants to see the effect of labeling students (gifted vs struggling) on their achivement in second grade. she proposes that students in an elementary school be divided into reading groups in whcih ability levels (determined by previous test scores) are evenly mixed. one group will be told they are gifted, another struggling, and another where they say nothing. charlotte theorizes that by the end of the second-grade year, the gifted group will outperform those in the struggling group on the same reading test.

1. what additional information might you want to know as an IRB member to decide whether or not it should be supported?

2. benefits and harms?

3. what would your recommendation be?

1. i feel like i know everything and confounding variables have been cleared up (like if the abilities are actually mixed but we know the test scores)

2. benefits: know the power of language in children; harm: the struggling group will internalize it for life, as proven by The Monster Study (1939)

3. reject it just because the harms have been proven to be permanent, more harm done than good