Midterm 2 Flashcards
What are descriptive methods?
Do not involve the manipulation of any variables by the researcher. We can only speculate about the causation that may be involved.
Archival and Previously Recorded Sources of Data
Researchers may not gather their own data; may answer their research question using data recorded by other individuals for other purposes.
Examples of Archival and Previously Recorded Sources of Data
Public health and census data may be analyzed years later to answer questions about socioeconomic status, religion, or political party affiliation.
Problems with Archival and Previously Recorded Sources of Data (1)
- Unless you are dealing with the papers and documents of a few clearly identified individuals, you will not know exactly who left the data you are investigating. Not knowing the participants who make up your sample will make it difficult to understand and generalize your results. Your ability to make statements other than those that merely describe your data is limited.
Problems with Archival and Previously Recorded Sources of Data (2)
- Participants may have been selective in what they chose to write. Titchener translated Wundt’s book but severely misrepresented him which could affect our impression. His original writings are still available to be retranslated but we still face the problem that Wundt may have omitted things he did not wish to share.
You cannot avoid this problem of selective deposit.
What is selective deposit?
Problem with archival and previously recorded sources of data in which participants may have been selective in what they chose to write.
Problems with Archival and Previously Recorded Sources of Data (3)
- Survival of such records. In graffiti example, are the walls scrubbed clean every day or is the graffiti allowed to accumulate? The data in which you are interested will probably not have a very high survival rate.
Type of paper on which magazine was printed, high acid content of paper led to disintegration of these magazines, hence only a precious few have survived.
Comparisons with The Experimental Method - Archival and Previously Recorded Data
Aware of nonrepresentative samples, data that was purposely not recorded, and data that may be lost.
Because we examined data that was recorded at a different time under unknown circumstances, we are unable to control the gathering of this data.
Therefore, we cannot make any cause-and-effect statement, the best we can do is speculate about what might have occurred.
Naturalistic Observation
Seeking answers to research questions by observing behaviour in the real world, hallmark of qualitative research.
Can also use naturalistic observation to collect numerical data and answer more focused research questions.
Example; Researcher interested in child behaviour may go to a daycare center and record observations/their behaviour.
First goal of naturalistic observation
To describe behaviour as it occurs in the natural setting without the artificiality of the lab.
Second goal of naturalistic observation
Describe the variables that are present and the relations among them.
Naturalistic observation may provide clues concerning why birds migrate at particular times of the year and what factors determine the length of stay in a certain area.
What is important for the researcher to not do in naturalistic observation?
Not interfere with or intervene in the behaviour being studied. Observer should be as inconspcious as possible, one-way mirrors are popular.
Why must researcher be unobtrusive in naturalistic observation?
Avoid influencing or changing the behaviour of the participants of the study. Presence of an observer is not part of the natural setting, therefore they may behave differently in the presence of an observer.
Reactance or Reactivity Effect/Hawthorne Effect
Biasing of the participants responses because they know they are being observed.
Also called the Hawthorne effect because of Hawthorne experiment, which measured lighting on productivity. When lighting was dimmed, people were more productive than the plant. They knew they were research participants in an experiment, thus performed better.
Main drawback of naturalistic observation
Inability to make cause-and-effect statements, we do not manipulate any variables when we use this technique.
High on external validity, low on internal validity
Objectivity
Why use naturalistic observation if it does not allow us to make cause-and-effect statements? (1)
May be our only choice of research techniques to study a particular type of behaviour. Psychologists interested in seeing reactions after natural disasters cannot recreate these life-threatening situations, they must make their observations under naturally occurring conditions.
Why use naturalistic observation if it does not allow to make cause-and-effect statements? (2)
Adjunct to experimental method. Can use naturalistic observation before conducting experimental method to get an idea of the relevant variables in the situation. Once you have an idea about which variables are important, you can conduct systematic, controlled studies of this variable in a lab. After conducting lab experiment, you may return to natural setting to see if insights gained in lab are indeed applied to real life.
Time Sampling
Making observations at different time periods in order to obtain a more representative sampling of the behaviour of interest.
Selection of time periods may be determined randomly or in a more systematic manner.
Use of time sampling may apply to the same or different participants.
Situation Sampling
Observing the same type of behaviour in several different situations.
Two advantages of situation sampling
- Able to determine whether the behaviour in question changes as a function of the context in which you observed. Amount of personal space preference differs from one culture to another or one geographic region to another.
- Researchers likely to observe different participants in the different situations. Ability to generalize any behavioural consistencies has increased.
What major decision do you need to make before you conduct your research project?
Whether to present the results in a qualitative or quantitative manner.
Qualitative Approach
Description of the behaviour in question, a narrative record and conclusions prompted by this description.
Form of written or tape-recorded notes, during or immediately after observing behaviour. Avoid making speculative comments.
Two reasons for using more than one observer
- One observer may miss or overlook the behaviour.
- Disagreement concerning exactly what was seen and how it should be categorized or recorded.
Interobserver Reliability
The extent to which observers agree.
Low and high interobserver reliability
Observers disagree about the behaviour they observed = low
Observers agree about the behaviour they observed = high
How to measure interobserver reliability
Number of times observers agree / Number of opportunities to agree times 100 = percentage of agreement
Participant Observation
When the observer imbeds himself/herself within the group being studied.
e.g; How do homeless maintain a positive self-ID?
Researcher worked as a volunteer at a homeless shelter and had conversations with homeless people.
Clinical Perspective
Descriptive approach aimed at understanding and correcting a particular behavioural problem
Does after death call help people with mental illness?
How is clinical perspective different from participant observation?
- Client chooses clinician, whereas participant observer chooses others to study
- Clinicians cannot be unobtrusive or passive because they have been asked to participate in the situation
- Participant observers goal is understanding, whereas clinicians’ goal is helping
Descriptive Surveys
Seek to determine what % of the population have particular characteristics, beliefs, or behaviours
Characteristics
Who are you?
How many children do you have?
Do you suffer from diabetes? DEPRESSION? Asthma?
Do you have a driver’s license?
Are you employed?
Beliefs
What do you think?
Do you believe in gay marriage?
Do you like pepsi or coke better?
Who do you think will win the Stanley Cup?
How would you rate the service at this restaurant?
Behaviours
How will/do you act?
Who dyou plan to vote for?
Do you engage in unprotected sex?
Do you smoke?
Would you buy this product?
What TV programs do you watch?
Goal of descriptive Surveys
Representation/assessment of population
Using a sample to draw conclusions about the population
Analytic Surveys
Seek to determine the relevant variables and how they are related
Is aggression related to health behaviours in adolescents?
Analytic Survey Steps
Step 1
- What are the relevant variables?
Variables: Aggression and health behaviours
Construct 1: Aggression, aggression questionnaire based off 5-point Likert scale
Construct 2: Health Behaviours
How many times smoked cigs, alcohol, diet.
Analytic Survey Steps
Step 2
How are they related?
Results.
Research Strategies
- Single-strata approach
- Cross-sectional research
- Longitudinal research
What is single-strata approach?
Select from one subgroup of a population
Cross-sectional research
Compares multiple subgroups at the same time
Longitudinal research
Looks at one group over an extended period of time
This group is called a cohort
Choosing your Survey
Lots of surveys out there, useful because generally have been tested.
- How will you administer your survey. Pros.
- Can be completed without researcher
- Can be sent to a large number of people
- With online surveys, little/no data entry
- Random-digit dialing, random sampling
- Can enter data immediately on computer
- Can clear up ambiguous answers
- Can control order that questions are answered
- Increased response rate
- Can clear up ambiguous answers
- Can control the order questions are answered
How will you administer your survey? Cons.
- Can’t be sure who answered the survey
- Can’t be sure participants paying attention
- Low response rates
- Caller ID and voicemail
- Can’t use visual aids or nonverbal cues
- Difficult to establish rapport
- Takes more time of experimenter and participant
- Is more expensive
- Possibility of interviewer bias
What kind of questions will you use?
Yes-No questions
Forced alternative - Select between two alternative responses
Multiple choice - best response from alternatives
Likert - response based on designated scale
Open-ended - construct one’s own answer
- Write the items
Questions should
- be clear, concise, short
- familiar vocabulary
- appropriate reading level for sample
- be specific
- Pilot test
Preliminary test to try out procedures and make any needed changes or adjustments
ALL STEPS TO DEVELOP A SURVEY
- How will you administer your survey?
- What type of questions will you use?
- Write out the items
- Pilot test
- Add/remove/clarify questions
- Create survey instructions
How do tests and inventories differ from surveys?
Surveys examine an opinion on an issue or topic
Tests assess a specific attribute, characteristic, or ability of the person being tested
Correlational Research
Both a statistical technique and a research method
Research designed to determine whether an association exists between two variables.
Does not mean causation
Correlations, positive
As one variable increases, scores on the other variable also increase.
Test scores are positively correlated if a student who makes a low score on Test 1 also scores low on Test 2, whereas a student who scores high on test 1 also scores high on test 2.
Height and weight are positively correlated.
As A increases, B increases.
Negative Correlation
Increase in one variable is accompanied by a decrease in the second variable.
As A increases, B decreases.
Drinking water and thirst are negatively correlated, when you drink water, you reduce thirst.
Zero Correlation
Two variables under consideration are not related. High scores on one variable may be paired with low, intermediate, or high on the second variable or vice versa.
Knowing the score on variable 1 does not help us with variable 2.
Internal Validity
Deals with experimental control.
Extent to which we can be sure the IV is the cause of the DV. IV actually created any change that can be observed in the DV
External Validity
Deals with generalizability
Extent to which we can be sure we can generalize our results to different populations
Threats to Internal Validity 1
- Selection
If participants are not equal prior to the experiment, don’t know whether results are a function of the IV or a pre-existing difference
Threats to Internal Validity 2
History.
When events occur during an experiment. After IV manipulation but before DV measure is taken.
Affects DV
Threats to Internal Validity 3
Maturation
When changes occur in the participants over time during their participation in the experiment that affects the DV
Threats to Internal Validity 4
Testing
When measuring the DV causes a change in the DV.
Threats to Internal Validity 5
Instrumentation
Measurement criterion is changed during experiment or faulty.
Threats to Internal Validity 6
Statistical Regression
Remeasure participants who have extreme scores, very high or very low, scores likely to regress or move towards mean.
Threats to Internal Validity 7
Experimental Mortality
Participants from different groups drop out at different rates
Obvious Fix for Threat TO internal validity
Random assignment
Why use pretest-posttest control group design?
Controls for selection effects.
Pretest makes sure RA equalized groups
Without introducing any additional confounds
Both groups experience pre-test
May be harder to detect effect of manipulation due to testing effects
Why use a posttest only control group design?
Increased protection from testing effects. If worried about pretest effects on posttest scores.
Lack of pretest prevents training/practice/learning effects on posttest scores.
We can’t know how scores change.
Decreased protection from selection effects.
Assume RA will equalize groups but there is no pretest to make sure.
The Solomon Four-Group Design
Control for selection
Allow us to see the specific effects of the pretest on the posttest scores
Gain in external validity
Tabulating Data
We need to find a way to summarize our numbers!
We can do this by
1. Create a frequency distribution
2. Create a graph
Allows us to see our data pictorially, preferable to tables.
What is a frequency distribution?
A table that shows classes (intervals of data) with a count of the number of entries in each interval.
The frequency of a class is the number of data entries in that interval.
Constructing a Frequency Distribution
- Number of classes = 7
- Find the class width.
How to find the class width
Determine range of the data max - min and divide by number of classes. Round up, 11.29 is rounded up to 12 to ensure all data is included.
Find the class limits
Use minimum data value as the lower limit of the first class.
Find the remaining lower limits, add the class width to the lower limit of the preceding class.
7 will be at the top, and then you add 12 to 7, 12 to 19 etc to 79 (which is our max-min)
Find the upper limits
Upper limit of the first class is 18, one less than 19. And then 30 which is one less than 31, and then 42 which is one less than 43. 78+12 for the last one.
Make a tally mark for each data entry in the row of the appropriate class.
7-18. Find all data points within this range and tally it up and then count tallies to find the total frequency for each class.
Frequency Histogram
Consecutive bars must touch.
A bar graph that represents the frequency distribution
Horizontal (x) axis is quantitative and measures the data values
Vertical (Y) axis measures the frequencies of the classes
Constructing a Frequency Histogram
- Calculate Class Boundaries
Numbers that separate classes without gaps
Distance from the upper limit of the first class to the lower limit of the second
class is 19-8=1
Half of the distance is 0.5
7-0.5
18+0.5
Using Midpoints
Lower class limit + Upper class limit /2
Which of the following is an issue with the use of frequency histograms/polygons?
Individual data values are lost
Stem-and-leaf plot
Each number is separated into a stem and a leaf.
Similar to a histogram
Still contains original data values.
Graphing Correlational Data
Each entry in one data set corresponds to one entry in a second data set
Graph using a scatter plot.
Ordered pairs are graphed as points in a coordinate plane. Used to show the relationship between two quantitative variables.
Unimodel Symmetrical
One clear peak. Values increase and rise to one highest point before falling
Bimodal Symmetrical
Two peaks are mirror images.
Negative Skew
Higher on the right side. Longer fatter tail on the left side
Positive Skew
Higher on the left side. Long fatter tail on the right side
Kurtosis
Extent of deviation from normal curve in width of curve and thickness of tails
Mesokurtic Curve, Leptokurtic curve, platykurtic curve
Normal distibution, lepto is very high up and narrow tall and skinny, platy is like a flat lime dome flat and wide
