Research Methods

What are the 4 types of research?

Empirical, Nonemperical, Qualitative, and Quantitative research

Involves a collection of new information or data through observation of measurement of behavior or physical properties.

empirical research

Typically uses existing information rather than gathering new data

nonemperical research

Data based off things other than numbers; often relies on verbal information. Ex: Describing a persons behavior or statements

qualitative research

related to numerical information, frequency counts, percentages etc.

quantitative research

What are the 4 types of quantitative research?

Experimental, Quasi-experimental, Nonexperimental

Researchers identify one or more factors they will manipulate or control. Ex: comparing different treatment approaches for fluency disorders

experimental quantitative research

Experimental research that lacks random assignments. Ex: Children in certain classes received different reading curriculum. Children are not assigned to classrooms randomly.

quasi-experimental quantitative research

More descriptive in nature. Ex: case studies, surveys, fundamental frequency measurements, comparison studies, longitudinal studies

nonexperimental quantitative research

The characteristic the researcher wants to study. Ex: age, race, sex, TBI or no TBI

independent variable

The outcome measures that measure the effects of the experimental manipulation. Ex: Test score, level of aspiration, MLU, speech intelligibility

dependent variable

In ethics in research what are the two major areas of concern?

1. Protection of human subjects

2. Fraudulent data

What are the three principles of human participant protection?

1. Respect of persons - giving sufficient information to make voluntary decisions.

2. Beneficience - researchers' obligation to protect the wellbeing of the participants

3. Justice - need to make equitable decisions re: who is allowed to participate in research.

There are 2 types of hypotheses. What are they?

Null hypothesis and Research hypothesis

stated in the negative form in which the researcher assumes that there will not be a significant effect.

null hypothesis

State in the positive form in which the researcher assumes there will be a positive effect.

research hypothesis

What does PICO stand for?

Population/Patient

Intervention

Comparison

Outcome

T or F: Qualitative: natural environment or situation. Emphasize the meaning of participants statements and behaviors (e.g. home).

True

Need to control environment so that all participants have an equal experience in the study (lab space).

True, for quantitative

T or F: Qualitative often a small number of participants, spend a lot of time reviewing data from these participants.

True

T or F: Quantitative large number of participants, spend a short time in the lab setting, analyze in batches of large group data.

True

What is the aim of experimental research design?

The aim of the experiment research is to find a cause- effect relationship, that changing one variable will be the only cause of changing the other variables

What are the two characteristics of the experimental research?

1. Manipulating the independent variable

2. Control over the extraneous variables

What are some advances of experimental research?

The researcher is able to identify and determine the cause - effect relationship. This means that changing the independent variable may produce a change in the dependent variable.

What are the 5 types of experimental research designs?

- Posttest only design

- Pretest-posttest randomized group

- Solomon randomized four-group

- Switching replications

- Factorial

What is post-test only randomized design?

1) Randomization of participants to assigned groups

2) Implementation of treatments

3) Observation and measurement of behavior

What is the most common true experimental design?

Pretest-Posttest randomized control group design

a combination of the two factors. Pre-test evidence gives you an indication that groups were not different before the treatment started—treatment effect

mixed model design

What is switching replications design?

•Switching roles - 2 phases

•Phase I: Participants are randomly assigned to treatment and control groups.

•Conduct posttest.

•Phase II: Switch roles; the treatment group became the control group and the control group becomes the treatment group

•Conduct posttest.

This design provides a way to deal with the problem of denying treatment to a control group

What is a factorial design?

•The researcher is manipulating two or more independent variables in the same experiment.

•Each independent variable is called a "factor".

•Each factors has two or more levels.

•The research is able to determine how more than one independent variable can affect the dependent variable and how the two independent variables influence one another.

More than one measure of the same variable; only one group of participants; Participants serve as their own controls and experience all levels of the IV (e.g. different treatments)

repeated measures design

Results for each participant are reported individually. Does not mean the study only had one participant rather refers to how researchers report their findings. Participants experience both treatment and control conditions.

single subject design

The strongest design to establish cause and effect is

true experimental design

What are the 4 measurement scales?

Nominal

Ordinal

Interval

Ratio

categorical data, or labels; usually consist of "count data"

nominal data

data organized in some hierarchical or rank order

Ordinal data

data organized in order,with equal intervals between data

interval data

Numbers on _______ can be added, subtracted, multiplied and divided in a meaningful way.

interval scale

data organized in order,with equal intervals between dataand an "absolute zero"

ratio data

T or F: Frequencies and percentages are most often used with nominal level measures.

True

T or F: Measures of central tendency (such as the mean) only describe one aspect of a frequency distribution.

True

best measure of central tendency only if the data are normally distributed.

mean

Used to determine the likelihood that the findings from a sample represent the population as a whole.

inferential statistics

what are designed to determine the probability that the null hypothesis accurately represents the population.

Inferential statistics

rejecting a true null hypothesis; the researcher concludes that the groups are significantly different when the groups were not different. "false alarm"

type one error

failing to reject a false null hypothesis; says there is no difference between groups when in fact there is difference.

type two error

Parametric statistical procedure requires ______ and _____. Most well-known statistics are parametric.

interval scale or ratio scale

T or F: Non-Parametric Statistical Procedure: (normal distribution not required). Data measured on any scale.

True

Most common procedure for analyzing the difference between 2 sets of data when you have interval or ratio level measures (test scores, speech rate, accuracy, etc.)

t-test

What are the 3 factors influences whether you will find a difference in the t-test?

1) Magnitude of difference between means - the larger the difference between means the more likely the t-test will show significance.

2) Amount of variability in the data (less varianceà more likely to see a difference).

3) Sample size (larger sampleà more likely to find a difference - it represents the population better).

matched participants or two measures on the same participants

Paired t-test

refers to identifying individuals who have a disorder.

Sensitivity

refers to excluding individuals who are free of a disorder.

Specificity

an index (or measure) of the degree to which the differences you observed in your sample groups actually exist in the real population. "How large an effect do I expect exists in the population?"

effect size

refers to the probability that you will be able to detect a difference between subject groups, if such a difference actually existswithin the population.

power

subjects receiving the treatment are referred to as

"experimental group"

T or F: Subjects receiving no treatment are referred to as the "control group"

True

If subjects receive two or more conditions (i.e., Treatment vs. No Treatment or Treatment A vs. Treatment B) it is a ________.

"repeated measures" design

If each condition is tested using a different group of subjects (i.e., each subject is tested under only one condition), you are studying what?

"independent samples"

If subjects in the various groups are matched by some criteria (e.g., age, gender, cognitive ability, pre-treatment performance) it is a

"matched pairs" design

is a procedure that measures the strength of a relationship between two sets of data.

correlation

is a procedure that uses the strength of that relationship (i.e., the correlation) to make predictions of one data set from another.

regression

T or F: Correlation does not imply causation! Just because two sets of data are related does not mean one causes (or affects) the other.

One has to be very careful in drawing causitive conclusions from correlational data.

True

persons of interest for a particular study, defined at the beginning stages of the study, all members have one or more predetermined characteristics

•Ex. 4-5 yr old children who are bilingual.

population

people who actually participate in a study (from larger population) AKA participants

sample

T or F: APA guidelines states that the use of "subjects" should be avoided. Refer to participants by who they are e.g. "high-schoolers" "1st graders" "bilingual pre-schoolers" or by what they do e.g."listeners", "participants" *but also use person-first language (e.g. avoid "aphasics" "stutterers"

True

all the people the researchers want the results to apply to.

intended population

the group from which participants are recruited (because of study limitations i.e. geography, bureaucracy, etc.).

accessible population

T or F: A sample is representative if characteristics of a sample are a good match to characteristics of the population e.g. education level, mental health status, etc.

True

What are the three sources of bias in sampling?

1. Failing to identify all members of a group.

2. convenience sample (a sample that is easy to access)

3. volunteerism (we can't force people to participate)

every member has an equal chance of being selected, uses random numbers—via a table of random numbers or random number function on spreadsheet

simple random

What is systematic sampling?

•Systematic Sampling:

•Start with a list of potential participants

•Establish a sampling interval

•Select every so many participants according to your interval (e.g.list of 2500 participants

•Goal is to recruit 125, 2500/125=20

•So your sampling interval is 20.

•Every 20th person will be selected to be in the study

Breaking the total population into stratified groups and then auditing each group as if it were a different population (excluding the immaterial strata). More efficient because you would do less tests than if you tested it as a whole

stratified sampling

a researcher develops a sampling frame that includes a list of all the states, cities, institutions, or organizations in which elements of the identified populations can be linked. A random sample of these can then be used in the study.

cluster sampling

The goal is to recruit participants that are the best source of information for a specific issue (e.g. infants with palatal cleft, children with selective mutism. Goal is not to generalize findings but get an expert opinion. Used in qualitative research to study individuals with unique experiences.

purposive sampling

assigning participants to experimental and control groups by chance, thus minimizing preexisting differences between the different groups

random assignment

What 3 factors affect decisions about sample size?

•The more variable a population is the larger a sample you need

•How large the group is

•Too small of a sample→ unable to detect differences

What is the criteria for judging how applicable findings are?

o Similarity of participants in the study and your clients

o Intensity and type of treatment appropriate for your clinical setting

o Required equipment available in your setting

o Benefits of the treatment justify any additional cost