PSYCH 350 Exam 1 Essay Practice

Briefly describe each of the sources of new knowledge. What is the accepted role of each source of knowledge in modern scientific psychological research?

- Intuition is knowledge based on how we feel about things, natural instinctive feelings about things to produce new knowledge.

- Authority is knowledge gained from other people whose intuitions have been right often enough that we trust them to produce new knowledge.

- Rational inductive is based on the process of a proof to gather current knowledge, start with an axiom, build together things we already know to result in new knowledge

- Empiricism is to apply the scientific method to gather current knowledge, form a hypothesis about some new knowledge, gather data relevant to the hypothesis, test the hypothesis, the results are the new knowledge.

- All four are legitimate sources of research hypotheses, but only empiricism is an accepted source of new knowledge with in modern scientific psychology.

Contrast "proof" vs. "evidence." Which is preferred, what keeps us from obtaining it, and what do we do instead? What do we do to convince ourselves that our new knowledge is correct?

- Proof gives us absolute certainty our answer is correct, while evidence provides us only probabilistic answers.

- We would rather have proof than evidence.

- We can only prove something using rational deductive proofs, like in logic or mathematics, which require that we have an axiom or starting point about which we are absolutely certain.

- Since we have no axioms in behavioral science, we are forced to use evidence obtained from empirical research.

- To provide the most convincing evidence possible, we use the best empirical methodologies from our research area and conduct replication and programmatic research studies.

Describe the research loop (be sure to briefly describe each stage). Tell the (3) different ways that it is applied and what we learn from that each type of application.

- The research loop is the steps in the scientific method.

- Library research to learn what we already know; a research hypothesis to guide our next study; a research design of how we will collect data to test the hypothesis; data collection followed by data analysis to test the research hypothesis, and we draw conclusions about whether the research hypothesis was supported or not.

- We apply the research loop in three ways.

1) Testing of novel research hypotheses (the sexiest part).

2) We also complete replication research rerunning exactly the same study to be sure we got the proper results.

3) And we run programmatic research/convergent research with different variations of the study to determine which variation produce similar and different results.

Briefly describe the kinds of validity we want our research to have and the dependent nature among them.

- Having measurement validity means that our data accurately reflects the behaviors and characteristics we are are trying to study.

- External validity means that our choices of participants, setting, and task/stimulus reflect the who, where and doing what we are trying to study.

- Internal validity means that we have controlled confounds so that we are studying the effects we are trying to study.

- Statistical conclusion validity means that we get the right empirical results from our study.

- Statistical conclusion validity is dependent on �chance� and the other three types of validity, measurement, internal and external validity.

What is required to have a "truly random sample"? Is this often accomplished? When you are told that a sample is "random," what has usually been done?

- You need a complete sampling frame.

- You need a 100% return rate or complete substitution, and no attrition (participants dropping out as the study goes along).

- This is rarely accomplished, usually because of limits in sampling frames and return rates.

- When we hear the term random sample, it usually means they used a good purposive sampling frame, researcher selection, and had a good return rate (which varies from 15% to 75% depending on research area).

Compare and contrast IVs & confounds. Respond to the statement, "You only have to worry about confounds when you are testing a causal research hypothesis."

- IVs and confounds are both causal variables that have an influence to change participant's values on the DV.

- The IV is the causal variable were are trying to study when we want to know the IV-DV relationship.

- Confounds interfere with our study of the IV-DV relationship because, if there are confounds, we don't know if the DV was caused by the IV, caused by the confounds, or by some combination.

Describe the variables that exist "before the study begins" and "after the study is completed" and how they are related. What determines what variables exist after the study is completed?

- We have the DV (the response or outcome variable we are studying), the IV (the presumed causal effect variable influencing the DV), and everything else is a potential confound.

- After the study we still have our DV and our IV, but the potential confounds have become either constants, control variables, or confounding variables.

- The resources and expertise of the researcher determine whether potential confounds are controlled or become confounds that limit the internal validity and statistical conclusion validity of our results.

Distinguish between participant selection and participant assignment and tell the specific type of validity associated with each. Tell how "randomization" is applied to each and whether or not it is considered necessary.

- Participant selection, also called sampling, is who will participate in the study.

- Participant assignment is who will be in which IV conditions of the study.

- Selection related to the Population part of External Validity, while assignment is related to the Initial Equivalence part of Internal Validity

- Randomized assignment is absolutely required to provide the initial equivalence to test a causal research hypothesis.

- Randomized selection is desirable, but is not necessary to provide a representative sample to help with External validity.

Describe the two different characterizations of the relationship between internal validity and external validity. Which do you prefer and why?

- The precursor model is based on a primary emphasis on internal validity, causal interpretability and true experiments.

- The precursor model suggests that without internal validity and causal interpretability, external validity is unimportant (i.e., internal validity is a precursor to external validity).

- The trade-off model is based on research that primarily tests associative hypotheses using non-experiments.

- The trade-off model suggests that each decision a researcher makes about the research design or procedure influences the internal and external validity of the study and that they must consider the tradeoff between the two.

- I prefer the precursor model because I think that causal research hypotheses are most important and that the internal validity of a study is most important.

Can all causal research hypotheses be studied? Why or why not? (Be sure to give examples to support your answer!)

- No! Not all causal research hypotheses can be tested.

- You need a true experiment with no confounds.

- You can't always run a true experiment because you can't always randomly assign and manipulate the intended causal variable (the IV).

- Even if you can run a true experiment, you can't always provide for ongoing equivalence of all procedural variables

- The reasons you might not be able to randomly assign and manipulate and/or control ongoing equivalence are:

1) we lack the technical ability to manipulate all variables,

2) we often have ethical reasons not to manipulate variables that we are able to manipulate, and

3) sometimes we can ethically manipulate variables, but don't have the resources to do so in our research.

Respond to each of these statements. "Unless you run a true experiment there is no way you can causally interpret your results." "Running a true experiment guarantees your results will be causally interpretable."

- The first statement is true! Causal interpretability requires a true experiment with no confounds, so, without a true experiment you can't causal interpret your results

- The second statement is false! For two reasons

1) Sometimes random assignment doesn't produce initial equivalence of all subject variables

2) Sometimes confounds are introduced during IV manipulation, task completion or DV measurement, so that you don't have ongoing equivalence.

Suppose a colleague said to you, "Why even bother running non-experiments? We can't get any useful information from them!" What seems to be the type of information this colleague thinks is the only useful kind? How should you respond to this statement?

- Clearly this colleague believes that only causal knowledge is useful.

- You should respond by reminding the colleague that most of the information we apply in science, medicine, law and everyday life is associative.

- Associative information allows for accurate prediction, and prediction is often enough to support our scientific, medical, and societal decisions.

Describe the key steps in the research process, briefly describing the type(s) of validity "at stake" during the completion of each.

- Stating the research hypothesis and whether it is associative or causal, sets the stage for evaluating the quality of the study and its results.

- Participant selection/sampling (who will be in the study) is about the Population portion of External Validity

- Participant assignment (who will be in what conditions of the study, when) is about the Initial Equivalence portion of Internal Validity (control of subject variables) � this step is critical for True Experiments

- Manipulation of the IV involves Measurement validity, Setting and Task/stimulus parts of External validity and ongoing equivalence part of Internal Validity.

- Measurement of the DV also involves Measurement validity, Setting and Task/stimulus parts of External validity and ongoing equivalence part of Internal Validity.

- Data analysis is all about statistical conclusion validity � did we get the right answer?.

Identify the attributes of a research study that do and do not directly influence the causal interpretability of the results. Also, tell the attributes of a study that can make it harder to maintain ongoing equivalence, and so, casual interpretability.

- Two attributes of a study directly influence.

- The 2 are:

1) Random assignment of individual participants to IV conditions and

2) manipulation of the IV by the researcher.

- Four attributes of a study do not directly influence the causal interpretability or the research results

- The 4 are: 1) participant selection/sampling (population part of external validity), 2) Setting (part of external validity), 3) data collection method (measurement validity) and 4) statistical analysis used (statistical conclusion validity)

- Two attributes that do make ongoing equivalence control and causal interpretability harder are:

1) studies in field settings make it harder to maintain ongoing equivalence (of procedural variables) than studies in lab settings and

2) longer studies make it harder to maintain ongoing equivalence (or procedural variables) than shorter studies